This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Research ArticleRectifier Fault Diagnosis and Fault Tolerance ofa Doubly Fed Brushless Starter Generator
Liwei Shi and Zhou Bo
School of Traffic amp Vehicle Engineering Shandong University of Technology Zibo Shandong 250049 China
Correspondence should be addressed to Liwei Shi liwei10nuaaeducn
Received 25 August 2014 Accepted 7 October 2014
Academic Editor Sergiu Dan Stan
Copyright copy 2015 L Shi and Z Bo This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited
This paper presents a rectifier fault diagnosis method with wavelet packet analysis to improve the fault tolerant four-phase doublyfed brushless starter generator (DFBLSG) system reliability The system components and fault tolerant principle of the highreliable DFBLSG are given And the common fault of the rectifier is analyzed The process of wavelet packet transforms faultdetectionidentification algorithm is introduced in detail The fault tolerant performance and output voltage experiments weredone to gather the energy characteristics with a voltage sensor The signal is analyzed with 5-layer wavelet packets and theenergy eigenvalue of each frequency band is obtained Meanwhile the energy-eigenvalue tolerance was introduced to improvethe diagnostic accuracy With the wavelet packet fault diagnosis the fault tolerant four-phase DFBLSG can detect the usual open-circuit fault and operate in the fault tolerant mode if there is a fault The results indicate that the fault analysis techniques in thispaper are accurate and effective
1 Introduction
A starter generator can start the engine and supply electri-cal power for the aeronautics and automobiles A variablefrequency starter generator is developed for the Boeing 787which is introduced in [1] The switched reluctance machine(SRM) [2] and the PM machine [3] were also considered asstarter generators However the controlled power electroniccircuit for switched reluctance generator is very complexand expensive And the PM material is very strict with theworking temperature [4 5] It is difficult for both of the SRMand PM machine to regulate the output voltage when theywork as generators
Doubly fed brushless starter generator (DFBLSG) is anew type of brushless DC machine that comes from doublysalient PMmachine by using field winding instead of perma-nent magnet steel [6] Its armature windings and excitationwindings are both mounted on the salient poles of the statorand it has similar salient rotor structure to SRM rotor As agenerator DFBLSG does not need rotor position informationand controlled power electronic circuit in the SRM generatorDFBLSG has such advantages as simple structure low costhigh reliability and good fault tolerance Thus it has good
application prospects in many fields including wind powergeneration aeronautics astronautics automobiles and ships[7] In particular it can be employed as aeronautics startergenerator because it has excellent performance on both powergeneration and starting [8 9]
The fault tolerant schematic of amachine system is shownin Figure 1 When a multiphase machine with more thanthree phases is implemented in the system it can continueoperating even with one or two open-circuit phases at themonitor of deferent sensors So it is necessary to build asystem for fault tolerant application with the fault detectingsensors and controller [10 11]
Generating mode is of particular importance for thestarter and generator As a weak part of the doubly fedbrushless generator system the rectifier is prone to failurePractice shows that the faults of rectifier could be dividedinto short-circuit fault and open-circuit fault Short-circuitfault would result in stopping operation by a fuse [8] Onoccurrence of open-circuit fault the unbalanced operatingcondition of rectification circuit would cause output voltagevariation phase current distortion and even system crashesTherefore the research on rectifier faults detecting for doublyfed brushless starter generator is significantly important
Hindawi Publishing CorporationJournal of SensorsVolume 2015 Article ID 961894 9 pageshttpdxdoiorg1011552015961894
2 Journal of Sensors
OutputLoad
Phase A
Controller
Positionsensor
Current sensor
Input Phase BPhase C
Phase N
Machine
Voltagesensor
Figure 1 Fault tolerant schematic of a machine system
Recently there are many different methods raised forrectifier fault diagnosis applications The primary idea isdiagnosing fault information included in output voltage Todetect the fault of a rectifier with 12 phases an approach isgiven on the basis of voltage analysis with voltage eigenvaluein [12] An extended Kalman filter is used to estimate theunknown model parameters for nonlinear dynamic systems[13] Fourier transform and wavelet analysis has been used infault diagnosis in [14ndash16] In [14] a novel method for gearboxfault detection based on biorthogonal B-spline wavelet wasproposed And in [15] an automatic identification methodfor the leakage signal of petroleum pipeline is studied Thewavelet analysis is used to process the input signal because ithas the superiority of varied size of the analysis window to thefrequency [16]
This paper proposed a fault tolerant four-phase DFBLSGand presents a rectifier fault diagnosis method with waveletpacket analysis The system components and fault tolerantprinciple of the 129-pole high reliable DFBLSG are givenAnd the common fault of the rectifier is analyzedThe processof wavelet packet transforms fault detectionidentificationalgorithm is introduced in detail The output voltage signaldetected by a voltage sensor is used as a wavelet packetinput signal And the experiments were done to gather theenergy characteristicsThewavelet coefficients were obtainedfrom wavelet packet analysis of fault voltage signals andthen the wavelet reconstruction of each frequency band wasprocessed Meanwhile the energy-eigenvalue tolerance wasintroduced to improve the diagnostic accuracy With thewavelet packet fault diagnosis the fault tolerant four-phaseDFBLSG can detect the usual open-circuit fault and operatein the fault tolerantmode if there is a faultThe results indicatethat the fault analysis techniques in this paper are accurateand effective
2 System Components and Principle
21 System Components The DFBLSG comprises machinebody and its controller as shown in Figure 2 A four-phasefull-bridge inverter is used for DFBLSG drive Phase A andphase C connect in series and they build up an independentchannelThe other channel is composed of phase B and phaseD The position sensors are employed to detect the rotorposition and rotor speed on the starting mode And a voltage
+Q3
D3
U1
AC
Q4D4
Q1D1
Q2D2
Q5D5
Q6D6
D
D7
D8
B
Q7
Q8Positionsensors
Voltagesensor
Currentsensor
DSPcontroller
minus
Figure 2 The system of the DFBLSG
A1 B1
A3
C1
C2
B2B3
D1D3
C3
51 2
A2
D2
3 4
Figure 3 The structure and magnetic flux of the DFBLSG
sensor is used to detect the rectifier fault The controller isutilized to control the output voltage the rotating of themachine and the fault detection
When the DFBLSG works as a generator all the switchedtubes are turned off and four-phase output voltage is rectifiedby the eight diodes The output voltage is stabilized byadjusting the field winding current under the monitor of avoltage sensor
The starter generator can output 14V DC source for themotor when it runs as a generator The 12V DC battery cansupply DC source when the machine is required to start theengine
22 Structure and Principle of DFBLSG Figure 3 shows thestructure of the novel four-phase photograph It is differentwith the traditional 128-pole three-phase machine [17]because it has 12 stator poles and 9 rotor poles (Table 2)
Obviously each phase winding consists of three concen-trated coils For example phase A consists of A1 A2 andA3 which are connected in series as shown in Figure 3It should be noted that the phase winding direction ofthis configuration is different from the topology in of thetraditional four-phase DFBLSG which wounds the field coils
Journal of Sensors 3
Table 1 Classification of diode open faults
Item Fault type ExampleT1 One diode or two diodes of the same loop open circuit D1 D3 D5 D7 D2 D4 D6 D8 D1D4 D3D2 D5D8 D7D6T2 Other two diodes or more of the same channel open circuit D1D3 D5D7 D2D4 D6D8 D1D2 D3D4 D5D6 D7D8 D1D2D3
T3 Two diodes of the different channel open circuit D1D5 D3D7 D2D6 D4D8 D1D6 D1D8 D3D6 D3D8 D5D2D5D4 D7D2 D7D4
T4 Three diodes of the different channel open circuit D1D2D5 D1D2D6
Table 2 Key parameters of the generator
Item DataNumber of stator poles 12Number of rotor poles 9Stator outer diameter (mm) 136Rotor outer diameter (mm) 835Air gap (mm) 025Axle length (mm) 40Stator tooth height (mm) 138Stator yoke height (mm) 12Stator pole arc coefficient 0667Rotor pole arc coefficient 05Rated power (W) 300Rated voltage (V) 14
around four stator poles But both of them have the samedirection as the field coils The phase windings of the twosections are not connected directly Therefore there are twochannels which are labeled as A C and B D
23 Rectifier and Its Fault The traditional three-phase gener-ator has three types of rectifiers [18] which are positive full-bridge rectifier half-wave rectifier and negative half-waverectifier Similarly the four-phase DFBLSG can also use theabove rectifiers The four-phase full-bridge rectifier shownin Figure 4 which has two independent output channels isstudied because it has the ability of fault isolation
Rectifier is the weaknesses of the power generationsystem Potential faults of the DFBLSG can be divided intoshort circuit and open circuit Because the short-circuit faultsare very harmful to the system it should be protected by afuse Hence this paper will mainly study diode open-circuitfaults including single diode open-circuit and double diodesopen-circuit faults
On occurrence of open faults the output voltagewould bedistorted in accordance with certain laws For example whenD1 andD2 are open the output voltage will be distorted in thefirst half period So the diode open faults can be classified into4 types according to distortion laws of waveform as shown inTable 1
3 Fault Detection Process
In Figure 2 the voltage is detected by a voltage sensor andthe analog voltage signal is sent to DSP after conditioningThen DSP controller analyzed the sampled signal and select
T1 T5 T7
T2 T6 T8
F9
ABC
D1 D3 D5
D2D2 D6
F1
F2F5
F4F3
F6
F7
F8
MF15
T3
T4
D3
D4
Figure 4 Potential faults of the DFBLSG
fault energy eigenvalue with a wavelet packet firstly Thenthe eigenvalue is compared with a sample stored in DSPcontroller If the fault condition is reached the DSP willmake appropriate decisions and output fault indication signalthrough the IO port
The fault detection process includes the following steps(1) Analyze the normalized voltage sampling signal with
the wavelet packet let the decomposed layer be 119895 andextract wavelet coefficient of every band [19]The normalized equation of array [119875
119894] can be
described as
1198751015840
119894=119875119894minus 119875min119875max minus 119875min
(1)
(2) Reconstruct the decomposition coefficients of thewavelet packet and extract the signal from every band
(3) Calculate each band signal energy eigenvaluesaccording to
Figure 5 Voltage waveforms under normal and one phase open-circuit fault
Journal of Sensors 5(
)
0
20
40
60
80
100
THDK
N T1 T2 T3 T4
Figure 6 Voltage pulsating ratio and THDunder normal and open-circuit faultN normal T1 faultD1 open circuit T2 faultD1D2opencircuit T3 fault D1D5 open circuit T4 fault D1D2D5 open circuit
And the new characteristic vector is shown as
119879 = [1199040 1199041 119904
2119895minus1] (5)
(5) Establish the sample fault file We should collect alarge number of experimental data samples and thenthe fault characteristic vector can be determined withthe statistical average value of the sample
The statistical average value can be calculated with
119862119896=119873
sum119897=1
119904119896 (119897)
119899 119896 = 0 1 2
119895minus1 (6)
where 119899 is the number of experimental timesThe error discriminant vector Δ119875 is used to describe
the tolerance range of statistical average value 119862119896 and the
element of Δ119875 can be caculated by
Δ119862119896= 119870120590 = 119870radic
1
119873
119873
sum119897=1
[119904119896 (119897) minus 119862119896]
2 (7)
where 120590 is the sample standard deviation and 119870 is thetolerance factor generally taken to be 1
Equation (6) establishes mappings between energy fea-ture vector and fault state With the obtained fault modeparameter a table that describes the relationship betweenfault state and change parameter is stored in DSP Then thefaults can be identified and displayed
4 Experiments and Wavelet PacketAnalysis Conclusions
41 Prototype Experiments A four-phase full-bridge recti-fier was built in the experimental bench The prototypemachine operating in normal state and fault state was testedFigure 5(a) shows the voltage waveform when the machinehas no fault Figures 5(b) and 5(c) display the voltagewaveform with one diode open circuit The other two diodesor more of the same channel open-circuit fault two diodesof the different channel open-circuit fault and three diodesof the different channel open-circuit fault were shown inFigures 5(d)ndash5(f) respectively
The equation for calculation the voltage ripple is shownin
119870V =119880max minus 119880min119880119900
(8)
And the total harmonic distortion (THD) of voltage isexpressed as
THD =radicsuminfin
2119881119894
2
1198811
times 100 (9)
With the fast Fourier transform analysis of the voltagewaveform obtained from the oscilloscope the normal andvarious fault voltage ripple ratio and THD can be drawnin Figure 6 As can be seen from the figure the four-phasefull-bridge rectifier can tolerate the fault list in Table 1 Butthe voltage ripple and THD are very high when there is anopen-circuit fault So the in time detection of fault type isvery necessary for us to relieve the negative influence of thisvoltage ripple
Figure 7 presents the photograph of the generator and thecontroller Figure 8(a) shows the nonload characteristic witha four-phase full bridgeAnd the external characteristic underdifferent fault is shown in Figure 8(b) when the excitationcurrent is 4A From the figure we can see that the generatorcan tolerate one-diode open-circuit faults well because theother diode in the same leg can rectify the positive or negativecurrent Comparing the characteristics curves under variousfaults the output voltage of single phase open circuit fault islower than one diode open-circuit fault The characteristicexperiments results show the same conclusion with thevoltage waveforms in Figure 5 Because the number of phasewindings is relatively small the external characteristic of themachine is ldquohardrdquo compared to common generators Whenthere is a failure the DFBLSG can achieve fault toleranceby increasing the field current or outputting a relatively lowpower with the same field current
42 Wavelet Packet Analysis The rectified voltage samplingfrequency is 50 kHz the sample length of oscilloscope is2500 and the sampling repeated 10 times According to theabove diagnostic methods we use db1 wavelet as the baseswave to carry out the wavelet packet analysis The waveletpacket decomposition tree is shown in Figure 9 and thedecomposition layer 119895 = 5 As an example Figure 10 presents
6 Journal of Sensors
(a) Rotor (b) Stator (c) DSP controller
Figure 7 Rotor stator and controller photograph
0
5
10
15
20
25
0 2 4 6 8 10Field current (A)
Out
put v
olta
ge (V
)
Normal circuitD1 open circuit
D1D2 open circuitD1D5 open circuit
(a) Nonload characteristic
02468
1012141618
0 5 10 15 20Output current (A)
Out
put v
olta
ge (V
)
Normal circuitD1 open circuit
D1D2 open circuitD1D5 open circuit
(b) External characteristic
Figure 8 Characteristic experiments results with full bridge mode
(00)
(10) (11)
(20) (21)
(30) (31)
(40) (41)
(50) (51)
Figure 9 The five-layer wavelet packet decomposition tree
the wavelet packet decomposition results of deferent bandBecause the waveform describes a relative energy size it hasno unit
Table 3 Energy-eigenvalue tolerance statistic on each fault condi-tion
Δ1198621Δ1198622Δ1198623Δ1198624Δ1198625
Normal 0008 00027 00027 0 00027D1 0075 00023 0008 0004 0004D1D1 open circuit 011 0041 0010 00032 00032D1D5 open circuit 004 0001 00003 00003 00003D1D2D5 open circuit 016 005 00012 00006 00006
The energy eigenvalue statistic on each fault conditionis shown in Figure 11 From the figure we can see that theelement vector data after 6th harmonic is almost zero Sothese frequency band energy eigenvalues have no practicalsignificance and they can be negligible
With (7) we can get the energy-eigenvalue tolerancestatistic on each fault condition (Table 3)
If the system is normal the energy characteristics of allthe frequency bands are very little If D1 is open circuitcharacteristic of the first frequency band is high If D1 and
Journal of Sensors 7
10000 2000
0
1
minus1
(a) Analyzed signal in case of D1 open circuit
804020 600
4
16
12
8
(b) Packet (50)
806040200
0
1
2
minus1
(c) Packet (51)
80 120400
0
08
minus08
(d) Packet (41)
200 400 600
0
01
0
02
minus01
(e) Packet (21)
12008004000
0
01
minus01
(f) Packet (11)
Figure 10 The wavelet packet decomposition results
D2 are open circuit the energy characteristic of the secondfrequency band is higher than the one of the first bandIf there is an open-circuit fault on D1D5 the first and thesecond frequency band are almost the same All the energycharacteristics are very high if there is a D1D2D5 open-circuitfault
The experiments prove that when the generator rectifiercircuit fails the wavelet packet analysis results of voltage onthe sampling signal can be used for the fault diagnosis Withthe established relationship between energy characteristicsand fault state the open-circuit fault can be recognized by thewavelet packet analysis
8 Journal of Sensors
0
1
2
3
4
5
6
7
8
9
1 2 3 4 5 6 7 8
D1D2D5 open circuitNormal D1D5 open circuit
D1D2 open circuitD1 open circuit
Figure 11 Energy eigenvalue statistic on each fault condition
5 Conclusions
The doubly fed brushless starter generator has broad appli-cation prospects in the aerospace and automotive industrywhich need high reliability for the whole system To meetthe requirement of the fault tolerant capability a four-phaseDFBLSG which has the characteristic of phase redundancyphase isolation and fault diagnosis is presented in this paperAs the weak part of the system the rectifier faults are dividedinto one diode or two diodes of the same loop open circuittwo diodes of the same channel open-circuit two diodes ofthe different channel open circuit and three diodes of thedifferent channel open circuit
To detect the open-circuit fault of the rectifier a rectifierfault diagnosis method with wavelet packet analysis is pre-sented Wavelet packet analysis can decompose the voltagesignal into several layers in the whole frequency rangeWith the energy eigenvalues extracted from full-bandwaveletpacket eigenvectors can be established to describe differentfaults The fault tolerant performance and output voltageexperiments were done to gather the energy characteristicswith a voltage sensor With the established relationshipbetween energy characteristics and fault state the open-circuit fault can be diagnosed by the wavelet packet analysis
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This paper is financially supported by the Natural ScienceFoundation of Shandong (no ZR2012EM011) The authorswould like to thank APSC in Nanjing University of Aeronau-tics and Astronautics
References
[1] Z Chen H Wang and Y Yan ldquoA doubly salient start-ergenerator with two-section twisted-rotor structure forpotential future aerospace applicationrdquo IEEE Transactions onIndustrial Electronics vol 59 no 9 pp 3588ndash3595 2012
[2] W Ding and D Liang ldquoA fast analytical model for an integratedswitched reluctance startergeneratorrdquo IEEE Transactions onEnergy Conversion vol 25 no 4 pp 948ndash956 2010
[3] H Mirahki M Moallem and S A Rahimi ldquoDesign opti-mization of IPMSM for 42 v integrated starter alternatorusing lumped parameter model and genetic algorithmsrdquo IEEETransactions on Magnetics vol 50 no 3 pp 114ndash119 2014
[4] C Liu K T Chau and X Zhang ldquoAn efficient wind-photovoltaic hybrid generation system using doubly excitedpermanent-magnet brushless machinerdquo IEEE Transactions onIndustrial Electronics vol 57 no 3 pp 831ndash839 2010
[5] D Lee Y-H Jeong and S-Y Jung ldquoDesign of wound rotorsynchronous machine for ISG and performance comparisonwith interior permanent magnet synchronous machinerdquo Trans-actions of the Korean Institute of Electrical Engineers vol 62 no1 pp 37ndash42 2013
[6] Y Du K T ChauM Cheng Y FanW Zhao and F Li ldquoA linearstator permanentmagnet vernierHTSmachine for wave energyconversionrdquo IEEE Transactions on Applied Superconductivityvol 22 no 3 Article ID 5202505 2012
[7] Z Zhang Y Yan and Y Tao ldquoA new topology of low speed dou-bly salient brushless DC generator for wind power generationrdquoIEEE Transactions on Magnetics vol 48 no 3 pp 1227ndash12332012
[8] L Sun Z Zhang and L Qian ldquoCalculation and analysis of ironloss in Doubly Salient Brushless DC generatorrdquo in Proceedingsof the 39th Annual Conference of the IEEE Industrial ElectronicsSociety pp 2650ndash2655 November 2013
[9] L Zhang B Zhou Y-S Zhang F-S Cheng and J-D WeildquoAnalysis and suppression of torque ripple for doubly salientelectro-magnetic motorrdquo Proceedings of the Chinese Society ofElectrical Engineering vol 30 no 3 pp 83ndash89 2010
[10] W Cao B C Mecrow G J Atkinson J W Bennett and DJ Atkinson ldquoOverview of electric motor technologies used formore electric aircraft (MEA)rdquo IEEE Transactions on IndustrialElectronics vol 59 no 9 pp 3523ndash3531 2012
[11] W Xu J Zhu Y Zhang and J Hu ldquoCogging torque reduc-tion for radially laminated flux-switching permanent magnetmachine with 1214 polesrdquo in Proceedings of the 37th AnnualConference of the IEEE Industrial Electronics Society (IECON rsquo11)pp 3590ndash3595 Melbourne Australia November 2011
[12] C-F Zhang M Yan J He C Luo and Y-Y Xiao ldquoCurrentsensor fault diagnosis for doubly-fed wind generator controlsystemrdquo Information Technology Journal vol 12 no 18 pp4368ndash4373 2013
[13] S Mariani and A Ghisi ldquoUnscented Kalman filtering fornonlinear structural dynamicsrdquo Nonlinear Dynamics vol 49no 1-2 pp 131ndash150 2007
[14] G Zhang and Y Ge ldquoA novel method for gearbox faultdetection based on biorthogonal B-spline waveletrdquo Sensors andTransducers vol 133 no 10 pp 83ndash94 2011
[15] B Zhou JWang S Li andWWang ldquoAnewkey predistributionscheme for multiphase sensor networks using a new deploy-ment modelrdquo Journal of Sensors vol 2014 Article ID 57391310 pages 2014
[16] M Aktas ldquoA novel method for inverter faults detection anddiagnosis in PMSM drives of HEVs based on discrete wavelettransformrdquo Advances in Electrical and Computer Engineeringvol 12 no 4 pp 33ndash38 2012
[17] F Yang and Z Chen ldquoResearch on torque ripple of doublysalient electro-magnetic generator under no-load conditionrdquo
Journal of Sensors 9
Transactions of China Electrotechnical Society vol 28 no 1 pp290ndash295 2013
[18] B A Welchko T A Lipo T M Jahns and S E Schulz ldquoFaulttolerant three-phase AC motor drive topologies a comparisonof features cost and limitationsrdquo IEEE Transactions on PowerElectronics vol 19 no 4 pp 1108ndash1116 2004
[19] X Chen G Lin and Y Zhang ldquoDenoising method based onsparse representation for WFT signalrdquo Journal of Sensors vol2014 Article ID 145870 10 pages 2014
Figure 1 Fault tolerant schematic of a machine system
Recently there are many different methods raised forrectifier fault diagnosis applications The primary idea isdiagnosing fault information included in output voltage Todetect the fault of a rectifier with 12 phases an approach isgiven on the basis of voltage analysis with voltage eigenvaluein [12] An extended Kalman filter is used to estimate theunknown model parameters for nonlinear dynamic systems[13] Fourier transform and wavelet analysis has been used infault diagnosis in [14ndash16] In [14] a novel method for gearboxfault detection based on biorthogonal B-spline wavelet wasproposed And in [15] an automatic identification methodfor the leakage signal of petroleum pipeline is studied Thewavelet analysis is used to process the input signal because ithas the superiority of varied size of the analysis window to thefrequency [16]
This paper proposed a fault tolerant four-phase DFBLSGand presents a rectifier fault diagnosis method with waveletpacket analysis The system components and fault tolerantprinciple of the 129-pole high reliable DFBLSG are givenAnd the common fault of the rectifier is analyzedThe processof wavelet packet transforms fault detectionidentificationalgorithm is introduced in detail The output voltage signaldetected by a voltage sensor is used as a wavelet packetinput signal And the experiments were done to gather theenergy characteristicsThewavelet coefficients were obtainedfrom wavelet packet analysis of fault voltage signals andthen the wavelet reconstruction of each frequency band wasprocessed Meanwhile the energy-eigenvalue tolerance wasintroduced to improve the diagnostic accuracy With thewavelet packet fault diagnosis the fault tolerant four-phaseDFBLSG can detect the usual open-circuit fault and operatein the fault tolerantmode if there is a faultThe results indicatethat the fault analysis techniques in this paper are accurateand effective
2 System Components and Principle
21 System Components The DFBLSG comprises machinebody and its controller as shown in Figure 2 A four-phasefull-bridge inverter is used for DFBLSG drive Phase A andphase C connect in series and they build up an independentchannelThe other channel is composed of phase B and phaseD The position sensors are employed to detect the rotorposition and rotor speed on the starting mode And a voltage
+Q3
D3
U1
AC
Q4D4
Q1D1
Q2D2
Q5D5
Q6D6
D
D7
D8
B
Q7
Q8Positionsensors
Voltagesensor
Currentsensor
DSPcontroller
minus
Figure 2 The system of the DFBLSG
A1 B1
A3
C1
C2
B2B3
D1D3
C3
51 2
A2
D2
3 4
Figure 3 The structure and magnetic flux of the DFBLSG
sensor is used to detect the rectifier fault The controller isutilized to control the output voltage the rotating of themachine and the fault detection
When the DFBLSG works as a generator all the switchedtubes are turned off and four-phase output voltage is rectifiedby the eight diodes The output voltage is stabilized byadjusting the field winding current under the monitor of avoltage sensor
The starter generator can output 14V DC source for themotor when it runs as a generator The 12V DC battery cansupply DC source when the machine is required to start theengine
22 Structure and Principle of DFBLSG Figure 3 shows thestructure of the novel four-phase photograph It is differentwith the traditional 128-pole three-phase machine [17]because it has 12 stator poles and 9 rotor poles (Table 2)
Obviously each phase winding consists of three concen-trated coils For example phase A consists of A1 A2 andA3 which are connected in series as shown in Figure 3It should be noted that the phase winding direction ofthis configuration is different from the topology in of thetraditional four-phase DFBLSG which wounds the field coils
Journal of Sensors 3
Table 1 Classification of diode open faults
Item Fault type ExampleT1 One diode or two diodes of the same loop open circuit D1 D3 D5 D7 D2 D4 D6 D8 D1D4 D3D2 D5D8 D7D6T2 Other two diodes or more of the same channel open circuit D1D3 D5D7 D2D4 D6D8 D1D2 D3D4 D5D6 D7D8 D1D2D3
T3 Two diodes of the different channel open circuit D1D5 D3D7 D2D6 D4D8 D1D6 D1D8 D3D6 D3D8 D5D2D5D4 D7D2 D7D4
T4 Three diodes of the different channel open circuit D1D2D5 D1D2D6
Table 2 Key parameters of the generator
Item DataNumber of stator poles 12Number of rotor poles 9Stator outer diameter (mm) 136Rotor outer diameter (mm) 835Air gap (mm) 025Axle length (mm) 40Stator tooth height (mm) 138Stator yoke height (mm) 12Stator pole arc coefficient 0667Rotor pole arc coefficient 05Rated power (W) 300Rated voltage (V) 14
around four stator poles But both of them have the samedirection as the field coils The phase windings of the twosections are not connected directly Therefore there are twochannels which are labeled as A C and B D
23 Rectifier and Its Fault The traditional three-phase gener-ator has three types of rectifiers [18] which are positive full-bridge rectifier half-wave rectifier and negative half-waverectifier Similarly the four-phase DFBLSG can also use theabove rectifiers The four-phase full-bridge rectifier shownin Figure 4 which has two independent output channels isstudied because it has the ability of fault isolation
Rectifier is the weaknesses of the power generationsystem Potential faults of the DFBLSG can be divided intoshort circuit and open circuit Because the short-circuit faultsare very harmful to the system it should be protected by afuse Hence this paper will mainly study diode open-circuitfaults including single diode open-circuit and double diodesopen-circuit faults
On occurrence of open faults the output voltagewould bedistorted in accordance with certain laws For example whenD1 andD2 are open the output voltage will be distorted in thefirst half period So the diode open faults can be classified into4 types according to distortion laws of waveform as shown inTable 1
3 Fault Detection Process
In Figure 2 the voltage is detected by a voltage sensor andthe analog voltage signal is sent to DSP after conditioningThen DSP controller analyzed the sampled signal and select
T1 T5 T7
T2 T6 T8
F9
ABC
D1 D3 D5
D2D2 D6
F1
F2F5
F4F3
F6
F7
F8
MF15
T3
T4
D3
D4
Figure 4 Potential faults of the DFBLSG
fault energy eigenvalue with a wavelet packet firstly Thenthe eigenvalue is compared with a sample stored in DSPcontroller If the fault condition is reached the DSP willmake appropriate decisions and output fault indication signalthrough the IO port
The fault detection process includes the following steps(1) Analyze the normalized voltage sampling signal with
the wavelet packet let the decomposed layer be 119895 andextract wavelet coefficient of every band [19]The normalized equation of array [119875
119894] can be
described as
1198751015840
119894=119875119894minus 119875min119875max minus 119875min
(1)
(2) Reconstruct the decomposition coefficients of thewavelet packet and extract the signal from every band
(3) Calculate each band signal energy eigenvaluesaccording to
Figure 5 Voltage waveforms under normal and one phase open-circuit fault
Journal of Sensors 5(
)
0
20
40
60
80
100
THDK
N T1 T2 T3 T4
Figure 6 Voltage pulsating ratio and THDunder normal and open-circuit faultN normal T1 faultD1 open circuit T2 faultD1D2opencircuit T3 fault D1D5 open circuit T4 fault D1D2D5 open circuit
And the new characteristic vector is shown as
119879 = [1199040 1199041 119904
2119895minus1] (5)
(5) Establish the sample fault file We should collect alarge number of experimental data samples and thenthe fault characteristic vector can be determined withthe statistical average value of the sample
The statistical average value can be calculated with
119862119896=119873
sum119897=1
119904119896 (119897)
119899 119896 = 0 1 2
119895minus1 (6)
where 119899 is the number of experimental timesThe error discriminant vector Δ119875 is used to describe
the tolerance range of statistical average value 119862119896 and the
element of Δ119875 can be caculated by
Δ119862119896= 119870120590 = 119870radic
1
119873
119873
sum119897=1
[119904119896 (119897) minus 119862119896]
2 (7)
where 120590 is the sample standard deviation and 119870 is thetolerance factor generally taken to be 1
Equation (6) establishes mappings between energy fea-ture vector and fault state With the obtained fault modeparameter a table that describes the relationship betweenfault state and change parameter is stored in DSP Then thefaults can be identified and displayed
4 Experiments and Wavelet PacketAnalysis Conclusions
41 Prototype Experiments A four-phase full-bridge recti-fier was built in the experimental bench The prototypemachine operating in normal state and fault state was testedFigure 5(a) shows the voltage waveform when the machinehas no fault Figures 5(b) and 5(c) display the voltagewaveform with one diode open circuit The other two diodesor more of the same channel open-circuit fault two diodesof the different channel open-circuit fault and three diodesof the different channel open-circuit fault were shown inFigures 5(d)ndash5(f) respectively
The equation for calculation the voltage ripple is shownin
119870V =119880max minus 119880min119880119900
(8)
And the total harmonic distortion (THD) of voltage isexpressed as
THD =radicsuminfin
2119881119894
2
1198811
times 100 (9)
With the fast Fourier transform analysis of the voltagewaveform obtained from the oscilloscope the normal andvarious fault voltage ripple ratio and THD can be drawnin Figure 6 As can be seen from the figure the four-phasefull-bridge rectifier can tolerate the fault list in Table 1 Butthe voltage ripple and THD are very high when there is anopen-circuit fault So the in time detection of fault type isvery necessary for us to relieve the negative influence of thisvoltage ripple
Figure 7 presents the photograph of the generator and thecontroller Figure 8(a) shows the nonload characteristic witha four-phase full bridgeAnd the external characteristic underdifferent fault is shown in Figure 8(b) when the excitationcurrent is 4A From the figure we can see that the generatorcan tolerate one-diode open-circuit faults well because theother diode in the same leg can rectify the positive or negativecurrent Comparing the characteristics curves under variousfaults the output voltage of single phase open circuit fault islower than one diode open-circuit fault The characteristicexperiments results show the same conclusion with thevoltage waveforms in Figure 5 Because the number of phasewindings is relatively small the external characteristic of themachine is ldquohardrdquo compared to common generators Whenthere is a failure the DFBLSG can achieve fault toleranceby increasing the field current or outputting a relatively lowpower with the same field current
42 Wavelet Packet Analysis The rectified voltage samplingfrequency is 50 kHz the sample length of oscilloscope is2500 and the sampling repeated 10 times According to theabove diagnostic methods we use db1 wavelet as the baseswave to carry out the wavelet packet analysis The waveletpacket decomposition tree is shown in Figure 9 and thedecomposition layer 119895 = 5 As an example Figure 10 presents
6 Journal of Sensors
(a) Rotor (b) Stator (c) DSP controller
Figure 7 Rotor stator and controller photograph
0
5
10
15
20
25
0 2 4 6 8 10Field current (A)
Out
put v
olta
ge (V
)
Normal circuitD1 open circuit
D1D2 open circuitD1D5 open circuit
(a) Nonload characteristic
02468
1012141618
0 5 10 15 20Output current (A)
Out
put v
olta
ge (V
)
Normal circuitD1 open circuit
D1D2 open circuitD1D5 open circuit
(b) External characteristic
Figure 8 Characteristic experiments results with full bridge mode
(00)
(10) (11)
(20) (21)
(30) (31)
(40) (41)
(50) (51)
Figure 9 The five-layer wavelet packet decomposition tree
the wavelet packet decomposition results of deferent bandBecause the waveform describes a relative energy size it hasno unit
Table 3 Energy-eigenvalue tolerance statistic on each fault condi-tion
Δ1198621Δ1198622Δ1198623Δ1198624Δ1198625
Normal 0008 00027 00027 0 00027D1 0075 00023 0008 0004 0004D1D1 open circuit 011 0041 0010 00032 00032D1D5 open circuit 004 0001 00003 00003 00003D1D2D5 open circuit 016 005 00012 00006 00006
The energy eigenvalue statistic on each fault conditionis shown in Figure 11 From the figure we can see that theelement vector data after 6th harmonic is almost zero Sothese frequency band energy eigenvalues have no practicalsignificance and they can be negligible
With (7) we can get the energy-eigenvalue tolerancestatistic on each fault condition (Table 3)
If the system is normal the energy characteristics of allthe frequency bands are very little If D1 is open circuitcharacteristic of the first frequency band is high If D1 and
Journal of Sensors 7
10000 2000
0
1
minus1
(a) Analyzed signal in case of D1 open circuit
804020 600
4
16
12
8
(b) Packet (50)
806040200
0
1
2
minus1
(c) Packet (51)
80 120400
0
08
minus08
(d) Packet (41)
200 400 600
0
01
0
02
minus01
(e) Packet (21)
12008004000
0
01
minus01
(f) Packet (11)
Figure 10 The wavelet packet decomposition results
D2 are open circuit the energy characteristic of the secondfrequency band is higher than the one of the first bandIf there is an open-circuit fault on D1D5 the first and thesecond frequency band are almost the same All the energycharacteristics are very high if there is a D1D2D5 open-circuitfault
The experiments prove that when the generator rectifiercircuit fails the wavelet packet analysis results of voltage onthe sampling signal can be used for the fault diagnosis Withthe established relationship between energy characteristicsand fault state the open-circuit fault can be recognized by thewavelet packet analysis
8 Journal of Sensors
0
1
2
3
4
5
6
7
8
9
1 2 3 4 5 6 7 8
D1D2D5 open circuitNormal D1D5 open circuit
D1D2 open circuitD1 open circuit
Figure 11 Energy eigenvalue statistic on each fault condition
5 Conclusions
The doubly fed brushless starter generator has broad appli-cation prospects in the aerospace and automotive industrywhich need high reliability for the whole system To meetthe requirement of the fault tolerant capability a four-phaseDFBLSG which has the characteristic of phase redundancyphase isolation and fault diagnosis is presented in this paperAs the weak part of the system the rectifier faults are dividedinto one diode or two diodes of the same loop open circuittwo diodes of the same channel open-circuit two diodes ofthe different channel open circuit and three diodes of thedifferent channel open circuit
To detect the open-circuit fault of the rectifier a rectifierfault diagnosis method with wavelet packet analysis is pre-sented Wavelet packet analysis can decompose the voltagesignal into several layers in the whole frequency rangeWith the energy eigenvalues extracted from full-bandwaveletpacket eigenvectors can be established to describe differentfaults The fault tolerant performance and output voltageexperiments were done to gather the energy characteristicswith a voltage sensor With the established relationshipbetween energy characteristics and fault state the open-circuit fault can be diagnosed by the wavelet packet analysis
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This paper is financially supported by the Natural ScienceFoundation of Shandong (no ZR2012EM011) The authorswould like to thank APSC in Nanjing University of Aeronau-tics and Astronautics
References
[1] Z Chen H Wang and Y Yan ldquoA doubly salient start-ergenerator with two-section twisted-rotor structure forpotential future aerospace applicationrdquo IEEE Transactions onIndustrial Electronics vol 59 no 9 pp 3588ndash3595 2012
[2] W Ding and D Liang ldquoA fast analytical model for an integratedswitched reluctance startergeneratorrdquo IEEE Transactions onEnergy Conversion vol 25 no 4 pp 948ndash956 2010
[3] H Mirahki M Moallem and S A Rahimi ldquoDesign opti-mization of IPMSM for 42 v integrated starter alternatorusing lumped parameter model and genetic algorithmsrdquo IEEETransactions on Magnetics vol 50 no 3 pp 114ndash119 2014
[4] C Liu K T Chau and X Zhang ldquoAn efficient wind-photovoltaic hybrid generation system using doubly excitedpermanent-magnet brushless machinerdquo IEEE Transactions onIndustrial Electronics vol 57 no 3 pp 831ndash839 2010
[5] D Lee Y-H Jeong and S-Y Jung ldquoDesign of wound rotorsynchronous machine for ISG and performance comparisonwith interior permanent magnet synchronous machinerdquo Trans-actions of the Korean Institute of Electrical Engineers vol 62 no1 pp 37ndash42 2013
[6] Y Du K T ChauM Cheng Y FanW Zhao and F Li ldquoA linearstator permanentmagnet vernierHTSmachine for wave energyconversionrdquo IEEE Transactions on Applied Superconductivityvol 22 no 3 Article ID 5202505 2012
[7] Z Zhang Y Yan and Y Tao ldquoA new topology of low speed dou-bly salient brushless DC generator for wind power generationrdquoIEEE Transactions on Magnetics vol 48 no 3 pp 1227ndash12332012
[8] L Sun Z Zhang and L Qian ldquoCalculation and analysis of ironloss in Doubly Salient Brushless DC generatorrdquo in Proceedingsof the 39th Annual Conference of the IEEE Industrial ElectronicsSociety pp 2650ndash2655 November 2013
[9] L Zhang B Zhou Y-S Zhang F-S Cheng and J-D WeildquoAnalysis and suppression of torque ripple for doubly salientelectro-magnetic motorrdquo Proceedings of the Chinese Society ofElectrical Engineering vol 30 no 3 pp 83ndash89 2010
[10] W Cao B C Mecrow G J Atkinson J W Bennett and DJ Atkinson ldquoOverview of electric motor technologies used formore electric aircraft (MEA)rdquo IEEE Transactions on IndustrialElectronics vol 59 no 9 pp 3523ndash3531 2012
[11] W Xu J Zhu Y Zhang and J Hu ldquoCogging torque reduc-tion for radially laminated flux-switching permanent magnetmachine with 1214 polesrdquo in Proceedings of the 37th AnnualConference of the IEEE Industrial Electronics Society (IECON rsquo11)pp 3590ndash3595 Melbourne Australia November 2011
[12] C-F Zhang M Yan J He C Luo and Y-Y Xiao ldquoCurrentsensor fault diagnosis for doubly-fed wind generator controlsystemrdquo Information Technology Journal vol 12 no 18 pp4368ndash4373 2013
[13] S Mariani and A Ghisi ldquoUnscented Kalman filtering fornonlinear structural dynamicsrdquo Nonlinear Dynamics vol 49no 1-2 pp 131ndash150 2007
[14] G Zhang and Y Ge ldquoA novel method for gearbox faultdetection based on biorthogonal B-spline waveletrdquo Sensors andTransducers vol 133 no 10 pp 83ndash94 2011
[15] B Zhou JWang S Li andWWang ldquoAnewkey predistributionscheme for multiphase sensor networks using a new deploy-ment modelrdquo Journal of Sensors vol 2014 Article ID 57391310 pages 2014
[16] M Aktas ldquoA novel method for inverter faults detection anddiagnosis in PMSM drives of HEVs based on discrete wavelettransformrdquo Advances in Electrical and Computer Engineeringvol 12 no 4 pp 33ndash38 2012
[17] F Yang and Z Chen ldquoResearch on torque ripple of doublysalient electro-magnetic generator under no-load conditionrdquo
Journal of Sensors 9
Transactions of China Electrotechnical Society vol 28 no 1 pp290ndash295 2013
[18] B A Welchko T A Lipo T M Jahns and S E Schulz ldquoFaulttolerant three-phase AC motor drive topologies a comparisonof features cost and limitationsrdquo IEEE Transactions on PowerElectronics vol 19 no 4 pp 1108ndash1116 2004
[19] X Chen G Lin and Y Zhang ldquoDenoising method based onsparse representation for WFT signalrdquo Journal of Sensors vol2014 Article ID 145870 10 pages 2014
Item Fault type ExampleT1 One diode or two diodes of the same loop open circuit D1 D3 D5 D7 D2 D4 D6 D8 D1D4 D3D2 D5D8 D7D6T2 Other two diodes or more of the same channel open circuit D1D3 D5D7 D2D4 D6D8 D1D2 D3D4 D5D6 D7D8 D1D2D3
T3 Two diodes of the different channel open circuit D1D5 D3D7 D2D6 D4D8 D1D6 D1D8 D3D6 D3D8 D5D2D5D4 D7D2 D7D4
T4 Three diodes of the different channel open circuit D1D2D5 D1D2D6
Table 2 Key parameters of the generator
Item DataNumber of stator poles 12Number of rotor poles 9Stator outer diameter (mm) 136Rotor outer diameter (mm) 835Air gap (mm) 025Axle length (mm) 40Stator tooth height (mm) 138Stator yoke height (mm) 12Stator pole arc coefficient 0667Rotor pole arc coefficient 05Rated power (W) 300Rated voltage (V) 14
around four stator poles But both of them have the samedirection as the field coils The phase windings of the twosections are not connected directly Therefore there are twochannels which are labeled as A C and B D
23 Rectifier and Its Fault The traditional three-phase gener-ator has three types of rectifiers [18] which are positive full-bridge rectifier half-wave rectifier and negative half-waverectifier Similarly the four-phase DFBLSG can also use theabove rectifiers The four-phase full-bridge rectifier shownin Figure 4 which has two independent output channels isstudied because it has the ability of fault isolation
Rectifier is the weaknesses of the power generationsystem Potential faults of the DFBLSG can be divided intoshort circuit and open circuit Because the short-circuit faultsare very harmful to the system it should be protected by afuse Hence this paper will mainly study diode open-circuitfaults including single diode open-circuit and double diodesopen-circuit faults
On occurrence of open faults the output voltagewould bedistorted in accordance with certain laws For example whenD1 andD2 are open the output voltage will be distorted in thefirst half period So the diode open faults can be classified into4 types according to distortion laws of waveform as shown inTable 1
3 Fault Detection Process
In Figure 2 the voltage is detected by a voltage sensor andthe analog voltage signal is sent to DSP after conditioningThen DSP controller analyzed the sampled signal and select
T1 T5 T7
T2 T6 T8
F9
ABC
D1 D3 D5
D2D2 D6
F1
F2F5
F4F3
F6
F7
F8
MF15
T3
T4
D3
D4
Figure 4 Potential faults of the DFBLSG
fault energy eigenvalue with a wavelet packet firstly Thenthe eigenvalue is compared with a sample stored in DSPcontroller If the fault condition is reached the DSP willmake appropriate decisions and output fault indication signalthrough the IO port
The fault detection process includes the following steps(1) Analyze the normalized voltage sampling signal with
the wavelet packet let the decomposed layer be 119895 andextract wavelet coefficient of every band [19]The normalized equation of array [119875
119894] can be
described as
1198751015840
119894=119875119894minus 119875min119875max minus 119875min
(1)
(2) Reconstruct the decomposition coefficients of thewavelet packet and extract the signal from every band
(3) Calculate each band signal energy eigenvaluesaccording to
Figure 5 Voltage waveforms under normal and one phase open-circuit fault
Journal of Sensors 5(
)
0
20
40
60
80
100
THDK
N T1 T2 T3 T4
Figure 6 Voltage pulsating ratio and THDunder normal and open-circuit faultN normal T1 faultD1 open circuit T2 faultD1D2opencircuit T3 fault D1D5 open circuit T4 fault D1D2D5 open circuit
And the new characteristic vector is shown as
119879 = [1199040 1199041 119904
2119895minus1] (5)
(5) Establish the sample fault file We should collect alarge number of experimental data samples and thenthe fault characteristic vector can be determined withthe statistical average value of the sample
The statistical average value can be calculated with
119862119896=119873
sum119897=1
119904119896 (119897)
119899 119896 = 0 1 2
119895minus1 (6)
where 119899 is the number of experimental timesThe error discriminant vector Δ119875 is used to describe
the tolerance range of statistical average value 119862119896 and the
element of Δ119875 can be caculated by
Δ119862119896= 119870120590 = 119870radic
1
119873
119873
sum119897=1
[119904119896 (119897) minus 119862119896]
2 (7)
where 120590 is the sample standard deviation and 119870 is thetolerance factor generally taken to be 1
Equation (6) establishes mappings between energy fea-ture vector and fault state With the obtained fault modeparameter a table that describes the relationship betweenfault state and change parameter is stored in DSP Then thefaults can be identified and displayed
4 Experiments and Wavelet PacketAnalysis Conclusions
41 Prototype Experiments A four-phase full-bridge recti-fier was built in the experimental bench The prototypemachine operating in normal state and fault state was testedFigure 5(a) shows the voltage waveform when the machinehas no fault Figures 5(b) and 5(c) display the voltagewaveform with one diode open circuit The other two diodesor more of the same channel open-circuit fault two diodesof the different channel open-circuit fault and three diodesof the different channel open-circuit fault were shown inFigures 5(d)ndash5(f) respectively
The equation for calculation the voltage ripple is shownin
119870V =119880max minus 119880min119880119900
(8)
And the total harmonic distortion (THD) of voltage isexpressed as
THD =radicsuminfin
2119881119894
2
1198811
times 100 (9)
With the fast Fourier transform analysis of the voltagewaveform obtained from the oscilloscope the normal andvarious fault voltage ripple ratio and THD can be drawnin Figure 6 As can be seen from the figure the four-phasefull-bridge rectifier can tolerate the fault list in Table 1 Butthe voltage ripple and THD are very high when there is anopen-circuit fault So the in time detection of fault type isvery necessary for us to relieve the negative influence of thisvoltage ripple
Figure 7 presents the photograph of the generator and thecontroller Figure 8(a) shows the nonload characteristic witha four-phase full bridgeAnd the external characteristic underdifferent fault is shown in Figure 8(b) when the excitationcurrent is 4A From the figure we can see that the generatorcan tolerate one-diode open-circuit faults well because theother diode in the same leg can rectify the positive or negativecurrent Comparing the characteristics curves under variousfaults the output voltage of single phase open circuit fault islower than one diode open-circuit fault The characteristicexperiments results show the same conclusion with thevoltage waveforms in Figure 5 Because the number of phasewindings is relatively small the external characteristic of themachine is ldquohardrdquo compared to common generators Whenthere is a failure the DFBLSG can achieve fault toleranceby increasing the field current or outputting a relatively lowpower with the same field current
42 Wavelet Packet Analysis The rectified voltage samplingfrequency is 50 kHz the sample length of oscilloscope is2500 and the sampling repeated 10 times According to theabove diagnostic methods we use db1 wavelet as the baseswave to carry out the wavelet packet analysis The waveletpacket decomposition tree is shown in Figure 9 and thedecomposition layer 119895 = 5 As an example Figure 10 presents
6 Journal of Sensors
(a) Rotor (b) Stator (c) DSP controller
Figure 7 Rotor stator and controller photograph
0
5
10
15
20
25
0 2 4 6 8 10Field current (A)
Out
put v
olta
ge (V
)
Normal circuitD1 open circuit
D1D2 open circuitD1D5 open circuit
(a) Nonload characteristic
02468
1012141618
0 5 10 15 20Output current (A)
Out
put v
olta
ge (V
)
Normal circuitD1 open circuit
D1D2 open circuitD1D5 open circuit
(b) External characteristic
Figure 8 Characteristic experiments results with full bridge mode
(00)
(10) (11)
(20) (21)
(30) (31)
(40) (41)
(50) (51)
Figure 9 The five-layer wavelet packet decomposition tree
the wavelet packet decomposition results of deferent bandBecause the waveform describes a relative energy size it hasno unit
Table 3 Energy-eigenvalue tolerance statistic on each fault condi-tion
Δ1198621Δ1198622Δ1198623Δ1198624Δ1198625
Normal 0008 00027 00027 0 00027D1 0075 00023 0008 0004 0004D1D1 open circuit 011 0041 0010 00032 00032D1D5 open circuit 004 0001 00003 00003 00003D1D2D5 open circuit 016 005 00012 00006 00006
The energy eigenvalue statistic on each fault conditionis shown in Figure 11 From the figure we can see that theelement vector data after 6th harmonic is almost zero Sothese frequency band energy eigenvalues have no practicalsignificance and they can be negligible
With (7) we can get the energy-eigenvalue tolerancestatistic on each fault condition (Table 3)
If the system is normal the energy characteristics of allthe frequency bands are very little If D1 is open circuitcharacteristic of the first frequency band is high If D1 and
Journal of Sensors 7
10000 2000
0
1
minus1
(a) Analyzed signal in case of D1 open circuit
804020 600
4
16
12
8
(b) Packet (50)
806040200
0
1
2
minus1
(c) Packet (51)
80 120400
0
08
minus08
(d) Packet (41)
200 400 600
0
01
0
02
minus01
(e) Packet (21)
12008004000
0
01
minus01
(f) Packet (11)
Figure 10 The wavelet packet decomposition results
D2 are open circuit the energy characteristic of the secondfrequency band is higher than the one of the first bandIf there is an open-circuit fault on D1D5 the first and thesecond frequency band are almost the same All the energycharacteristics are very high if there is a D1D2D5 open-circuitfault
The experiments prove that when the generator rectifiercircuit fails the wavelet packet analysis results of voltage onthe sampling signal can be used for the fault diagnosis Withthe established relationship between energy characteristicsand fault state the open-circuit fault can be recognized by thewavelet packet analysis
8 Journal of Sensors
0
1
2
3
4
5
6
7
8
9
1 2 3 4 5 6 7 8
D1D2D5 open circuitNormal D1D5 open circuit
D1D2 open circuitD1 open circuit
Figure 11 Energy eigenvalue statistic on each fault condition
5 Conclusions
The doubly fed brushless starter generator has broad appli-cation prospects in the aerospace and automotive industrywhich need high reliability for the whole system To meetthe requirement of the fault tolerant capability a four-phaseDFBLSG which has the characteristic of phase redundancyphase isolation and fault diagnosis is presented in this paperAs the weak part of the system the rectifier faults are dividedinto one diode or two diodes of the same loop open circuittwo diodes of the same channel open-circuit two diodes ofthe different channel open circuit and three diodes of thedifferent channel open circuit
To detect the open-circuit fault of the rectifier a rectifierfault diagnosis method with wavelet packet analysis is pre-sented Wavelet packet analysis can decompose the voltagesignal into several layers in the whole frequency rangeWith the energy eigenvalues extracted from full-bandwaveletpacket eigenvectors can be established to describe differentfaults The fault tolerant performance and output voltageexperiments were done to gather the energy characteristicswith a voltage sensor With the established relationshipbetween energy characteristics and fault state the open-circuit fault can be diagnosed by the wavelet packet analysis
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This paper is financially supported by the Natural ScienceFoundation of Shandong (no ZR2012EM011) The authorswould like to thank APSC in Nanjing University of Aeronau-tics and Astronautics
References
[1] Z Chen H Wang and Y Yan ldquoA doubly salient start-ergenerator with two-section twisted-rotor structure forpotential future aerospace applicationrdquo IEEE Transactions onIndustrial Electronics vol 59 no 9 pp 3588ndash3595 2012
[2] W Ding and D Liang ldquoA fast analytical model for an integratedswitched reluctance startergeneratorrdquo IEEE Transactions onEnergy Conversion vol 25 no 4 pp 948ndash956 2010
[3] H Mirahki M Moallem and S A Rahimi ldquoDesign opti-mization of IPMSM for 42 v integrated starter alternatorusing lumped parameter model and genetic algorithmsrdquo IEEETransactions on Magnetics vol 50 no 3 pp 114ndash119 2014
[4] C Liu K T Chau and X Zhang ldquoAn efficient wind-photovoltaic hybrid generation system using doubly excitedpermanent-magnet brushless machinerdquo IEEE Transactions onIndustrial Electronics vol 57 no 3 pp 831ndash839 2010
[5] D Lee Y-H Jeong and S-Y Jung ldquoDesign of wound rotorsynchronous machine for ISG and performance comparisonwith interior permanent magnet synchronous machinerdquo Trans-actions of the Korean Institute of Electrical Engineers vol 62 no1 pp 37ndash42 2013
[6] Y Du K T ChauM Cheng Y FanW Zhao and F Li ldquoA linearstator permanentmagnet vernierHTSmachine for wave energyconversionrdquo IEEE Transactions on Applied Superconductivityvol 22 no 3 Article ID 5202505 2012
[7] Z Zhang Y Yan and Y Tao ldquoA new topology of low speed dou-bly salient brushless DC generator for wind power generationrdquoIEEE Transactions on Magnetics vol 48 no 3 pp 1227ndash12332012
[8] L Sun Z Zhang and L Qian ldquoCalculation and analysis of ironloss in Doubly Salient Brushless DC generatorrdquo in Proceedingsof the 39th Annual Conference of the IEEE Industrial ElectronicsSociety pp 2650ndash2655 November 2013
[9] L Zhang B Zhou Y-S Zhang F-S Cheng and J-D WeildquoAnalysis and suppression of torque ripple for doubly salientelectro-magnetic motorrdquo Proceedings of the Chinese Society ofElectrical Engineering vol 30 no 3 pp 83ndash89 2010
[10] W Cao B C Mecrow G J Atkinson J W Bennett and DJ Atkinson ldquoOverview of electric motor technologies used formore electric aircraft (MEA)rdquo IEEE Transactions on IndustrialElectronics vol 59 no 9 pp 3523ndash3531 2012
[11] W Xu J Zhu Y Zhang and J Hu ldquoCogging torque reduc-tion for radially laminated flux-switching permanent magnetmachine with 1214 polesrdquo in Proceedings of the 37th AnnualConference of the IEEE Industrial Electronics Society (IECON rsquo11)pp 3590ndash3595 Melbourne Australia November 2011
[12] C-F Zhang M Yan J He C Luo and Y-Y Xiao ldquoCurrentsensor fault diagnosis for doubly-fed wind generator controlsystemrdquo Information Technology Journal vol 12 no 18 pp4368ndash4373 2013
[13] S Mariani and A Ghisi ldquoUnscented Kalman filtering fornonlinear structural dynamicsrdquo Nonlinear Dynamics vol 49no 1-2 pp 131ndash150 2007
[14] G Zhang and Y Ge ldquoA novel method for gearbox faultdetection based on biorthogonal B-spline waveletrdquo Sensors andTransducers vol 133 no 10 pp 83ndash94 2011
[15] B Zhou JWang S Li andWWang ldquoAnewkey predistributionscheme for multiphase sensor networks using a new deploy-ment modelrdquo Journal of Sensors vol 2014 Article ID 57391310 pages 2014
[16] M Aktas ldquoA novel method for inverter faults detection anddiagnosis in PMSM drives of HEVs based on discrete wavelettransformrdquo Advances in Electrical and Computer Engineeringvol 12 no 4 pp 33ndash38 2012
[17] F Yang and Z Chen ldquoResearch on torque ripple of doublysalient electro-magnetic generator under no-load conditionrdquo
Journal of Sensors 9
Transactions of China Electrotechnical Society vol 28 no 1 pp290ndash295 2013
[18] B A Welchko T A Lipo T M Jahns and S E Schulz ldquoFaulttolerant three-phase AC motor drive topologies a comparisonof features cost and limitationsrdquo IEEE Transactions on PowerElectronics vol 19 no 4 pp 1108ndash1116 2004
[19] X Chen G Lin and Y Zhang ldquoDenoising method based onsparse representation for WFT signalrdquo Journal of Sensors vol2014 Article ID 145870 10 pages 2014
Figure 5 Voltage waveforms under normal and one phase open-circuit fault
Journal of Sensors 5(
)
0
20
40
60
80
100
THDK
N T1 T2 T3 T4
Figure 6 Voltage pulsating ratio and THDunder normal and open-circuit faultN normal T1 faultD1 open circuit T2 faultD1D2opencircuit T3 fault D1D5 open circuit T4 fault D1D2D5 open circuit
And the new characteristic vector is shown as
119879 = [1199040 1199041 119904
2119895minus1] (5)
(5) Establish the sample fault file We should collect alarge number of experimental data samples and thenthe fault characteristic vector can be determined withthe statistical average value of the sample
The statistical average value can be calculated with
119862119896=119873
sum119897=1
119904119896 (119897)
119899 119896 = 0 1 2
119895minus1 (6)
where 119899 is the number of experimental timesThe error discriminant vector Δ119875 is used to describe
the tolerance range of statistical average value 119862119896 and the
element of Δ119875 can be caculated by
Δ119862119896= 119870120590 = 119870radic
1
119873
119873
sum119897=1
[119904119896 (119897) minus 119862119896]
2 (7)
where 120590 is the sample standard deviation and 119870 is thetolerance factor generally taken to be 1
Equation (6) establishes mappings between energy fea-ture vector and fault state With the obtained fault modeparameter a table that describes the relationship betweenfault state and change parameter is stored in DSP Then thefaults can be identified and displayed
4 Experiments and Wavelet PacketAnalysis Conclusions
41 Prototype Experiments A four-phase full-bridge recti-fier was built in the experimental bench The prototypemachine operating in normal state and fault state was testedFigure 5(a) shows the voltage waveform when the machinehas no fault Figures 5(b) and 5(c) display the voltagewaveform with one diode open circuit The other two diodesor more of the same channel open-circuit fault two diodesof the different channel open-circuit fault and three diodesof the different channel open-circuit fault were shown inFigures 5(d)ndash5(f) respectively
The equation for calculation the voltage ripple is shownin
119870V =119880max minus 119880min119880119900
(8)
And the total harmonic distortion (THD) of voltage isexpressed as
THD =radicsuminfin
2119881119894
2
1198811
times 100 (9)
With the fast Fourier transform analysis of the voltagewaveform obtained from the oscilloscope the normal andvarious fault voltage ripple ratio and THD can be drawnin Figure 6 As can be seen from the figure the four-phasefull-bridge rectifier can tolerate the fault list in Table 1 Butthe voltage ripple and THD are very high when there is anopen-circuit fault So the in time detection of fault type isvery necessary for us to relieve the negative influence of thisvoltage ripple
Figure 7 presents the photograph of the generator and thecontroller Figure 8(a) shows the nonload characteristic witha four-phase full bridgeAnd the external characteristic underdifferent fault is shown in Figure 8(b) when the excitationcurrent is 4A From the figure we can see that the generatorcan tolerate one-diode open-circuit faults well because theother diode in the same leg can rectify the positive or negativecurrent Comparing the characteristics curves under variousfaults the output voltage of single phase open circuit fault islower than one diode open-circuit fault The characteristicexperiments results show the same conclusion with thevoltage waveforms in Figure 5 Because the number of phasewindings is relatively small the external characteristic of themachine is ldquohardrdquo compared to common generators Whenthere is a failure the DFBLSG can achieve fault toleranceby increasing the field current or outputting a relatively lowpower with the same field current
42 Wavelet Packet Analysis The rectified voltage samplingfrequency is 50 kHz the sample length of oscilloscope is2500 and the sampling repeated 10 times According to theabove diagnostic methods we use db1 wavelet as the baseswave to carry out the wavelet packet analysis The waveletpacket decomposition tree is shown in Figure 9 and thedecomposition layer 119895 = 5 As an example Figure 10 presents
6 Journal of Sensors
(a) Rotor (b) Stator (c) DSP controller
Figure 7 Rotor stator and controller photograph
0
5
10
15
20
25
0 2 4 6 8 10Field current (A)
Out
put v
olta
ge (V
)
Normal circuitD1 open circuit
D1D2 open circuitD1D5 open circuit
(a) Nonload characteristic
02468
1012141618
0 5 10 15 20Output current (A)
Out
put v
olta
ge (V
)
Normal circuitD1 open circuit
D1D2 open circuitD1D5 open circuit
(b) External characteristic
Figure 8 Characteristic experiments results with full bridge mode
(00)
(10) (11)
(20) (21)
(30) (31)
(40) (41)
(50) (51)
Figure 9 The five-layer wavelet packet decomposition tree
the wavelet packet decomposition results of deferent bandBecause the waveform describes a relative energy size it hasno unit
Table 3 Energy-eigenvalue tolerance statistic on each fault condi-tion
Δ1198621Δ1198622Δ1198623Δ1198624Δ1198625
Normal 0008 00027 00027 0 00027D1 0075 00023 0008 0004 0004D1D1 open circuit 011 0041 0010 00032 00032D1D5 open circuit 004 0001 00003 00003 00003D1D2D5 open circuit 016 005 00012 00006 00006
The energy eigenvalue statistic on each fault conditionis shown in Figure 11 From the figure we can see that theelement vector data after 6th harmonic is almost zero Sothese frequency band energy eigenvalues have no practicalsignificance and they can be negligible
With (7) we can get the energy-eigenvalue tolerancestatistic on each fault condition (Table 3)
If the system is normal the energy characteristics of allthe frequency bands are very little If D1 is open circuitcharacteristic of the first frequency band is high If D1 and
Journal of Sensors 7
10000 2000
0
1
minus1
(a) Analyzed signal in case of D1 open circuit
804020 600
4
16
12
8
(b) Packet (50)
806040200
0
1
2
minus1
(c) Packet (51)
80 120400
0
08
minus08
(d) Packet (41)
200 400 600
0
01
0
02
minus01
(e) Packet (21)
12008004000
0
01
minus01
(f) Packet (11)
Figure 10 The wavelet packet decomposition results
D2 are open circuit the energy characteristic of the secondfrequency band is higher than the one of the first bandIf there is an open-circuit fault on D1D5 the first and thesecond frequency band are almost the same All the energycharacteristics are very high if there is a D1D2D5 open-circuitfault
The experiments prove that when the generator rectifiercircuit fails the wavelet packet analysis results of voltage onthe sampling signal can be used for the fault diagnosis Withthe established relationship between energy characteristicsand fault state the open-circuit fault can be recognized by thewavelet packet analysis
8 Journal of Sensors
0
1
2
3
4
5
6
7
8
9
1 2 3 4 5 6 7 8
D1D2D5 open circuitNormal D1D5 open circuit
D1D2 open circuitD1 open circuit
Figure 11 Energy eigenvalue statistic on each fault condition
5 Conclusions
The doubly fed brushless starter generator has broad appli-cation prospects in the aerospace and automotive industrywhich need high reliability for the whole system To meetthe requirement of the fault tolerant capability a four-phaseDFBLSG which has the characteristic of phase redundancyphase isolation and fault diagnosis is presented in this paperAs the weak part of the system the rectifier faults are dividedinto one diode or two diodes of the same loop open circuittwo diodes of the same channel open-circuit two diodes ofthe different channel open circuit and three diodes of thedifferent channel open circuit
To detect the open-circuit fault of the rectifier a rectifierfault diagnosis method with wavelet packet analysis is pre-sented Wavelet packet analysis can decompose the voltagesignal into several layers in the whole frequency rangeWith the energy eigenvalues extracted from full-bandwaveletpacket eigenvectors can be established to describe differentfaults The fault tolerant performance and output voltageexperiments were done to gather the energy characteristicswith a voltage sensor With the established relationshipbetween energy characteristics and fault state the open-circuit fault can be diagnosed by the wavelet packet analysis
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This paper is financially supported by the Natural ScienceFoundation of Shandong (no ZR2012EM011) The authorswould like to thank APSC in Nanjing University of Aeronau-tics and Astronautics
References
[1] Z Chen H Wang and Y Yan ldquoA doubly salient start-ergenerator with two-section twisted-rotor structure forpotential future aerospace applicationrdquo IEEE Transactions onIndustrial Electronics vol 59 no 9 pp 3588ndash3595 2012
[2] W Ding and D Liang ldquoA fast analytical model for an integratedswitched reluctance startergeneratorrdquo IEEE Transactions onEnergy Conversion vol 25 no 4 pp 948ndash956 2010
[3] H Mirahki M Moallem and S A Rahimi ldquoDesign opti-mization of IPMSM for 42 v integrated starter alternatorusing lumped parameter model and genetic algorithmsrdquo IEEETransactions on Magnetics vol 50 no 3 pp 114ndash119 2014
[4] C Liu K T Chau and X Zhang ldquoAn efficient wind-photovoltaic hybrid generation system using doubly excitedpermanent-magnet brushless machinerdquo IEEE Transactions onIndustrial Electronics vol 57 no 3 pp 831ndash839 2010
[5] D Lee Y-H Jeong and S-Y Jung ldquoDesign of wound rotorsynchronous machine for ISG and performance comparisonwith interior permanent magnet synchronous machinerdquo Trans-actions of the Korean Institute of Electrical Engineers vol 62 no1 pp 37ndash42 2013
[6] Y Du K T ChauM Cheng Y FanW Zhao and F Li ldquoA linearstator permanentmagnet vernierHTSmachine for wave energyconversionrdquo IEEE Transactions on Applied Superconductivityvol 22 no 3 Article ID 5202505 2012
[7] Z Zhang Y Yan and Y Tao ldquoA new topology of low speed dou-bly salient brushless DC generator for wind power generationrdquoIEEE Transactions on Magnetics vol 48 no 3 pp 1227ndash12332012
[8] L Sun Z Zhang and L Qian ldquoCalculation and analysis of ironloss in Doubly Salient Brushless DC generatorrdquo in Proceedingsof the 39th Annual Conference of the IEEE Industrial ElectronicsSociety pp 2650ndash2655 November 2013
[9] L Zhang B Zhou Y-S Zhang F-S Cheng and J-D WeildquoAnalysis and suppression of torque ripple for doubly salientelectro-magnetic motorrdquo Proceedings of the Chinese Society ofElectrical Engineering vol 30 no 3 pp 83ndash89 2010
[10] W Cao B C Mecrow G J Atkinson J W Bennett and DJ Atkinson ldquoOverview of electric motor technologies used formore electric aircraft (MEA)rdquo IEEE Transactions on IndustrialElectronics vol 59 no 9 pp 3523ndash3531 2012
[11] W Xu J Zhu Y Zhang and J Hu ldquoCogging torque reduc-tion for radially laminated flux-switching permanent magnetmachine with 1214 polesrdquo in Proceedings of the 37th AnnualConference of the IEEE Industrial Electronics Society (IECON rsquo11)pp 3590ndash3595 Melbourne Australia November 2011
[12] C-F Zhang M Yan J He C Luo and Y-Y Xiao ldquoCurrentsensor fault diagnosis for doubly-fed wind generator controlsystemrdquo Information Technology Journal vol 12 no 18 pp4368ndash4373 2013
[13] S Mariani and A Ghisi ldquoUnscented Kalman filtering fornonlinear structural dynamicsrdquo Nonlinear Dynamics vol 49no 1-2 pp 131ndash150 2007
[14] G Zhang and Y Ge ldquoA novel method for gearbox faultdetection based on biorthogonal B-spline waveletrdquo Sensors andTransducers vol 133 no 10 pp 83ndash94 2011
[15] B Zhou JWang S Li andWWang ldquoAnewkey predistributionscheme for multiphase sensor networks using a new deploy-ment modelrdquo Journal of Sensors vol 2014 Article ID 57391310 pages 2014
[16] M Aktas ldquoA novel method for inverter faults detection anddiagnosis in PMSM drives of HEVs based on discrete wavelettransformrdquo Advances in Electrical and Computer Engineeringvol 12 no 4 pp 33ndash38 2012
[17] F Yang and Z Chen ldquoResearch on torque ripple of doublysalient electro-magnetic generator under no-load conditionrdquo
Journal of Sensors 9
Transactions of China Electrotechnical Society vol 28 no 1 pp290ndash295 2013
[18] B A Welchko T A Lipo T M Jahns and S E Schulz ldquoFaulttolerant three-phase AC motor drive topologies a comparisonof features cost and limitationsrdquo IEEE Transactions on PowerElectronics vol 19 no 4 pp 1108ndash1116 2004
[19] X Chen G Lin and Y Zhang ldquoDenoising method based onsparse representation for WFT signalrdquo Journal of Sensors vol2014 Article ID 145870 10 pages 2014
Figure 6 Voltage pulsating ratio and THDunder normal and open-circuit faultN normal T1 faultD1 open circuit T2 faultD1D2opencircuit T3 fault D1D5 open circuit T4 fault D1D2D5 open circuit
And the new characteristic vector is shown as
119879 = [1199040 1199041 119904
2119895minus1] (5)
(5) Establish the sample fault file We should collect alarge number of experimental data samples and thenthe fault characteristic vector can be determined withthe statistical average value of the sample
The statistical average value can be calculated with
119862119896=119873
sum119897=1
119904119896 (119897)
119899 119896 = 0 1 2
119895minus1 (6)
where 119899 is the number of experimental timesThe error discriminant vector Δ119875 is used to describe
the tolerance range of statistical average value 119862119896 and the
element of Δ119875 can be caculated by
Δ119862119896= 119870120590 = 119870radic
1
119873
119873
sum119897=1
[119904119896 (119897) minus 119862119896]
2 (7)
where 120590 is the sample standard deviation and 119870 is thetolerance factor generally taken to be 1
Equation (6) establishes mappings between energy fea-ture vector and fault state With the obtained fault modeparameter a table that describes the relationship betweenfault state and change parameter is stored in DSP Then thefaults can be identified and displayed
4 Experiments and Wavelet PacketAnalysis Conclusions
41 Prototype Experiments A four-phase full-bridge recti-fier was built in the experimental bench The prototypemachine operating in normal state and fault state was testedFigure 5(a) shows the voltage waveform when the machinehas no fault Figures 5(b) and 5(c) display the voltagewaveform with one diode open circuit The other two diodesor more of the same channel open-circuit fault two diodesof the different channel open-circuit fault and three diodesof the different channel open-circuit fault were shown inFigures 5(d)ndash5(f) respectively
The equation for calculation the voltage ripple is shownin
119870V =119880max minus 119880min119880119900
(8)
And the total harmonic distortion (THD) of voltage isexpressed as
THD =radicsuminfin
2119881119894
2
1198811
times 100 (9)
With the fast Fourier transform analysis of the voltagewaveform obtained from the oscilloscope the normal andvarious fault voltage ripple ratio and THD can be drawnin Figure 6 As can be seen from the figure the four-phasefull-bridge rectifier can tolerate the fault list in Table 1 Butthe voltage ripple and THD are very high when there is anopen-circuit fault So the in time detection of fault type isvery necessary for us to relieve the negative influence of thisvoltage ripple
Figure 7 presents the photograph of the generator and thecontroller Figure 8(a) shows the nonload characteristic witha four-phase full bridgeAnd the external characteristic underdifferent fault is shown in Figure 8(b) when the excitationcurrent is 4A From the figure we can see that the generatorcan tolerate one-diode open-circuit faults well because theother diode in the same leg can rectify the positive or negativecurrent Comparing the characteristics curves under variousfaults the output voltage of single phase open circuit fault islower than one diode open-circuit fault The characteristicexperiments results show the same conclusion with thevoltage waveforms in Figure 5 Because the number of phasewindings is relatively small the external characteristic of themachine is ldquohardrdquo compared to common generators Whenthere is a failure the DFBLSG can achieve fault toleranceby increasing the field current or outputting a relatively lowpower with the same field current
42 Wavelet Packet Analysis The rectified voltage samplingfrequency is 50 kHz the sample length of oscilloscope is2500 and the sampling repeated 10 times According to theabove diagnostic methods we use db1 wavelet as the baseswave to carry out the wavelet packet analysis The waveletpacket decomposition tree is shown in Figure 9 and thedecomposition layer 119895 = 5 As an example Figure 10 presents
6 Journal of Sensors
(a) Rotor (b) Stator (c) DSP controller
Figure 7 Rotor stator and controller photograph
0
5
10
15
20
25
0 2 4 6 8 10Field current (A)
Out
put v
olta
ge (V
)
Normal circuitD1 open circuit
D1D2 open circuitD1D5 open circuit
(a) Nonload characteristic
02468
1012141618
0 5 10 15 20Output current (A)
Out
put v
olta
ge (V
)
Normal circuitD1 open circuit
D1D2 open circuitD1D5 open circuit
(b) External characteristic
Figure 8 Characteristic experiments results with full bridge mode
(00)
(10) (11)
(20) (21)
(30) (31)
(40) (41)
(50) (51)
Figure 9 The five-layer wavelet packet decomposition tree
the wavelet packet decomposition results of deferent bandBecause the waveform describes a relative energy size it hasno unit
Table 3 Energy-eigenvalue tolerance statistic on each fault condi-tion
Δ1198621Δ1198622Δ1198623Δ1198624Δ1198625
Normal 0008 00027 00027 0 00027D1 0075 00023 0008 0004 0004D1D1 open circuit 011 0041 0010 00032 00032D1D5 open circuit 004 0001 00003 00003 00003D1D2D5 open circuit 016 005 00012 00006 00006
The energy eigenvalue statistic on each fault conditionis shown in Figure 11 From the figure we can see that theelement vector data after 6th harmonic is almost zero Sothese frequency band energy eigenvalues have no practicalsignificance and they can be negligible
With (7) we can get the energy-eigenvalue tolerancestatistic on each fault condition (Table 3)
If the system is normal the energy characteristics of allthe frequency bands are very little If D1 is open circuitcharacteristic of the first frequency band is high If D1 and
Journal of Sensors 7
10000 2000
0
1
minus1
(a) Analyzed signal in case of D1 open circuit
804020 600
4
16
12
8
(b) Packet (50)
806040200
0
1
2
minus1
(c) Packet (51)
80 120400
0
08
minus08
(d) Packet (41)
200 400 600
0
01
0
02
minus01
(e) Packet (21)
12008004000
0
01
minus01
(f) Packet (11)
Figure 10 The wavelet packet decomposition results
D2 are open circuit the energy characteristic of the secondfrequency band is higher than the one of the first bandIf there is an open-circuit fault on D1D5 the first and thesecond frequency band are almost the same All the energycharacteristics are very high if there is a D1D2D5 open-circuitfault
The experiments prove that when the generator rectifiercircuit fails the wavelet packet analysis results of voltage onthe sampling signal can be used for the fault diagnosis Withthe established relationship between energy characteristicsand fault state the open-circuit fault can be recognized by thewavelet packet analysis
8 Journal of Sensors
0
1
2
3
4
5
6
7
8
9
1 2 3 4 5 6 7 8
D1D2D5 open circuitNormal D1D5 open circuit
D1D2 open circuitD1 open circuit
Figure 11 Energy eigenvalue statistic on each fault condition
5 Conclusions
The doubly fed brushless starter generator has broad appli-cation prospects in the aerospace and automotive industrywhich need high reliability for the whole system To meetthe requirement of the fault tolerant capability a four-phaseDFBLSG which has the characteristic of phase redundancyphase isolation and fault diagnosis is presented in this paperAs the weak part of the system the rectifier faults are dividedinto one diode or two diodes of the same loop open circuittwo diodes of the same channel open-circuit two diodes ofthe different channel open circuit and three diodes of thedifferent channel open circuit
To detect the open-circuit fault of the rectifier a rectifierfault diagnosis method with wavelet packet analysis is pre-sented Wavelet packet analysis can decompose the voltagesignal into several layers in the whole frequency rangeWith the energy eigenvalues extracted from full-bandwaveletpacket eigenvectors can be established to describe differentfaults The fault tolerant performance and output voltageexperiments were done to gather the energy characteristicswith a voltage sensor With the established relationshipbetween energy characteristics and fault state the open-circuit fault can be diagnosed by the wavelet packet analysis
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This paper is financially supported by the Natural ScienceFoundation of Shandong (no ZR2012EM011) The authorswould like to thank APSC in Nanjing University of Aeronau-tics and Astronautics
References
[1] Z Chen H Wang and Y Yan ldquoA doubly salient start-ergenerator with two-section twisted-rotor structure forpotential future aerospace applicationrdquo IEEE Transactions onIndustrial Electronics vol 59 no 9 pp 3588ndash3595 2012
[2] W Ding and D Liang ldquoA fast analytical model for an integratedswitched reluctance startergeneratorrdquo IEEE Transactions onEnergy Conversion vol 25 no 4 pp 948ndash956 2010
[3] H Mirahki M Moallem and S A Rahimi ldquoDesign opti-mization of IPMSM for 42 v integrated starter alternatorusing lumped parameter model and genetic algorithmsrdquo IEEETransactions on Magnetics vol 50 no 3 pp 114ndash119 2014
[4] C Liu K T Chau and X Zhang ldquoAn efficient wind-photovoltaic hybrid generation system using doubly excitedpermanent-magnet brushless machinerdquo IEEE Transactions onIndustrial Electronics vol 57 no 3 pp 831ndash839 2010
[5] D Lee Y-H Jeong and S-Y Jung ldquoDesign of wound rotorsynchronous machine for ISG and performance comparisonwith interior permanent magnet synchronous machinerdquo Trans-actions of the Korean Institute of Electrical Engineers vol 62 no1 pp 37ndash42 2013
[6] Y Du K T ChauM Cheng Y FanW Zhao and F Li ldquoA linearstator permanentmagnet vernierHTSmachine for wave energyconversionrdquo IEEE Transactions on Applied Superconductivityvol 22 no 3 Article ID 5202505 2012
[7] Z Zhang Y Yan and Y Tao ldquoA new topology of low speed dou-bly salient brushless DC generator for wind power generationrdquoIEEE Transactions on Magnetics vol 48 no 3 pp 1227ndash12332012
[8] L Sun Z Zhang and L Qian ldquoCalculation and analysis of ironloss in Doubly Salient Brushless DC generatorrdquo in Proceedingsof the 39th Annual Conference of the IEEE Industrial ElectronicsSociety pp 2650ndash2655 November 2013
[9] L Zhang B Zhou Y-S Zhang F-S Cheng and J-D WeildquoAnalysis and suppression of torque ripple for doubly salientelectro-magnetic motorrdquo Proceedings of the Chinese Society ofElectrical Engineering vol 30 no 3 pp 83ndash89 2010
[10] W Cao B C Mecrow G J Atkinson J W Bennett and DJ Atkinson ldquoOverview of electric motor technologies used formore electric aircraft (MEA)rdquo IEEE Transactions on IndustrialElectronics vol 59 no 9 pp 3523ndash3531 2012
[11] W Xu J Zhu Y Zhang and J Hu ldquoCogging torque reduc-tion for radially laminated flux-switching permanent magnetmachine with 1214 polesrdquo in Proceedings of the 37th AnnualConference of the IEEE Industrial Electronics Society (IECON rsquo11)pp 3590ndash3595 Melbourne Australia November 2011
[12] C-F Zhang M Yan J He C Luo and Y-Y Xiao ldquoCurrentsensor fault diagnosis for doubly-fed wind generator controlsystemrdquo Information Technology Journal vol 12 no 18 pp4368ndash4373 2013
[13] S Mariani and A Ghisi ldquoUnscented Kalman filtering fornonlinear structural dynamicsrdquo Nonlinear Dynamics vol 49no 1-2 pp 131ndash150 2007
[14] G Zhang and Y Ge ldquoA novel method for gearbox faultdetection based on biorthogonal B-spline waveletrdquo Sensors andTransducers vol 133 no 10 pp 83ndash94 2011
[15] B Zhou JWang S Li andWWang ldquoAnewkey predistributionscheme for multiphase sensor networks using a new deploy-ment modelrdquo Journal of Sensors vol 2014 Article ID 57391310 pages 2014
[16] M Aktas ldquoA novel method for inverter faults detection anddiagnosis in PMSM drives of HEVs based on discrete wavelettransformrdquo Advances in Electrical and Computer Engineeringvol 12 no 4 pp 33ndash38 2012
[17] F Yang and Z Chen ldquoResearch on torque ripple of doublysalient electro-magnetic generator under no-load conditionrdquo
Journal of Sensors 9
Transactions of China Electrotechnical Society vol 28 no 1 pp290ndash295 2013
[18] B A Welchko T A Lipo T M Jahns and S E Schulz ldquoFaulttolerant three-phase AC motor drive topologies a comparisonof features cost and limitationsrdquo IEEE Transactions on PowerElectronics vol 19 no 4 pp 1108ndash1116 2004
[19] X Chen G Lin and Y Zhang ldquoDenoising method based onsparse representation for WFT signalrdquo Journal of Sensors vol2014 Article ID 145870 10 pages 2014
Figure 8 Characteristic experiments results with full bridge mode
(00)
(10) (11)
(20) (21)
(30) (31)
(40) (41)
(50) (51)
Figure 9 The five-layer wavelet packet decomposition tree
the wavelet packet decomposition results of deferent bandBecause the waveform describes a relative energy size it hasno unit
Table 3 Energy-eigenvalue tolerance statistic on each fault condi-tion
Δ1198621Δ1198622Δ1198623Δ1198624Δ1198625
Normal 0008 00027 00027 0 00027D1 0075 00023 0008 0004 0004D1D1 open circuit 011 0041 0010 00032 00032D1D5 open circuit 004 0001 00003 00003 00003D1D2D5 open circuit 016 005 00012 00006 00006
The energy eigenvalue statistic on each fault conditionis shown in Figure 11 From the figure we can see that theelement vector data after 6th harmonic is almost zero Sothese frequency band energy eigenvalues have no practicalsignificance and they can be negligible
With (7) we can get the energy-eigenvalue tolerancestatistic on each fault condition (Table 3)
If the system is normal the energy characteristics of allthe frequency bands are very little If D1 is open circuitcharacteristic of the first frequency band is high If D1 and
Journal of Sensors 7
10000 2000
0
1
minus1
(a) Analyzed signal in case of D1 open circuit
804020 600
4
16
12
8
(b) Packet (50)
806040200
0
1
2
minus1
(c) Packet (51)
80 120400
0
08
minus08
(d) Packet (41)
200 400 600
0
01
0
02
minus01
(e) Packet (21)
12008004000
0
01
minus01
(f) Packet (11)
Figure 10 The wavelet packet decomposition results
D2 are open circuit the energy characteristic of the secondfrequency band is higher than the one of the first bandIf there is an open-circuit fault on D1D5 the first and thesecond frequency band are almost the same All the energycharacteristics are very high if there is a D1D2D5 open-circuitfault
The experiments prove that when the generator rectifiercircuit fails the wavelet packet analysis results of voltage onthe sampling signal can be used for the fault diagnosis Withthe established relationship between energy characteristicsand fault state the open-circuit fault can be recognized by thewavelet packet analysis
8 Journal of Sensors
0
1
2
3
4
5
6
7
8
9
1 2 3 4 5 6 7 8
D1D2D5 open circuitNormal D1D5 open circuit
D1D2 open circuitD1 open circuit
Figure 11 Energy eigenvalue statistic on each fault condition
5 Conclusions
The doubly fed brushless starter generator has broad appli-cation prospects in the aerospace and automotive industrywhich need high reliability for the whole system To meetthe requirement of the fault tolerant capability a four-phaseDFBLSG which has the characteristic of phase redundancyphase isolation and fault diagnosis is presented in this paperAs the weak part of the system the rectifier faults are dividedinto one diode or two diodes of the same loop open circuittwo diodes of the same channel open-circuit two diodes ofthe different channel open circuit and three diodes of thedifferent channel open circuit
To detect the open-circuit fault of the rectifier a rectifierfault diagnosis method with wavelet packet analysis is pre-sented Wavelet packet analysis can decompose the voltagesignal into several layers in the whole frequency rangeWith the energy eigenvalues extracted from full-bandwaveletpacket eigenvectors can be established to describe differentfaults The fault tolerant performance and output voltageexperiments were done to gather the energy characteristicswith a voltage sensor With the established relationshipbetween energy characteristics and fault state the open-circuit fault can be diagnosed by the wavelet packet analysis
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This paper is financially supported by the Natural ScienceFoundation of Shandong (no ZR2012EM011) The authorswould like to thank APSC in Nanjing University of Aeronau-tics and Astronautics
References
[1] Z Chen H Wang and Y Yan ldquoA doubly salient start-ergenerator with two-section twisted-rotor structure forpotential future aerospace applicationrdquo IEEE Transactions onIndustrial Electronics vol 59 no 9 pp 3588ndash3595 2012
[2] W Ding and D Liang ldquoA fast analytical model for an integratedswitched reluctance startergeneratorrdquo IEEE Transactions onEnergy Conversion vol 25 no 4 pp 948ndash956 2010
[3] H Mirahki M Moallem and S A Rahimi ldquoDesign opti-mization of IPMSM for 42 v integrated starter alternatorusing lumped parameter model and genetic algorithmsrdquo IEEETransactions on Magnetics vol 50 no 3 pp 114ndash119 2014
[4] C Liu K T Chau and X Zhang ldquoAn efficient wind-photovoltaic hybrid generation system using doubly excitedpermanent-magnet brushless machinerdquo IEEE Transactions onIndustrial Electronics vol 57 no 3 pp 831ndash839 2010
[5] D Lee Y-H Jeong and S-Y Jung ldquoDesign of wound rotorsynchronous machine for ISG and performance comparisonwith interior permanent magnet synchronous machinerdquo Trans-actions of the Korean Institute of Electrical Engineers vol 62 no1 pp 37ndash42 2013
[6] Y Du K T ChauM Cheng Y FanW Zhao and F Li ldquoA linearstator permanentmagnet vernierHTSmachine for wave energyconversionrdquo IEEE Transactions on Applied Superconductivityvol 22 no 3 Article ID 5202505 2012
[7] Z Zhang Y Yan and Y Tao ldquoA new topology of low speed dou-bly salient brushless DC generator for wind power generationrdquoIEEE Transactions on Magnetics vol 48 no 3 pp 1227ndash12332012
[8] L Sun Z Zhang and L Qian ldquoCalculation and analysis of ironloss in Doubly Salient Brushless DC generatorrdquo in Proceedingsof the 39th Annual Conference of the IEEE Industrial ElectronicsSociety pp 2650ndash2655 November 2013
[9] L Zhang B Zhou Y-S Zhang F-S Cheng and J-D WeildquoAnalysis and suppression of torque ripple for doubly salientelectro-magnetic motorrdquo Proceedings of the Chinese Society ofElectrical Engineering vol 30 no 3 pp 83ndash89 2010
[10] W Cao B C Mecrow G J Atkinson J W Bennett and DJ Atkinson ldquoOverview of electric motor technologies used formore electric aircraft (MEA)rdquo IEEE Transactions on IndustrialElectronics vol 59 no 9 pp 3523ndash3531 2012
[11] W Xu J Zhu Y Zhang and J Hu ldquoCogging torque reduc-tion for radially laminated flux-switching permanent magnetmachine with 1214 polesrdquo in Proceedings of the 37th AnnualConference of the IEEE Industrial Electronics Society (IECON rsquo11)pp 3590ndash3595 Melbourne Australia November 2011
[12] C-F Zhang M Yan J He C Luo and Y-Y Xiao ldquoCurrentsensor fault diagnosis for doubly-fed wind generator controlsystemrdquo Information Technology Journal vol 12 no 18 pp4368ndash4373 2013
[13] S Mariani and A Ghisi ldquoUnscented Kalman filtering fornonlinear structural dynamicsrdquo Nonlinear Dynamics vol 49no 1-2 pp 131ndash150 2007
[14] G Zhang and Y Ge ldquoA novel method for gearbox faultdetection based on biorthogonal B-spline waveletrdquo Sensors andTransducers vol 133 no 10 pp 83ndash94 2011
[15] B Zhou JWang S Li andWWang ldquoAnewkey predistributionscheme for multiphase sensor networks using a new deploy-ment modelrdquo Journal of Sensors vol 2014 Article ID 57391310 pages 2014
[16] M Aktas ldquoA novel method for inverter faults detection anddiagnosis in PMSM drives of HEVs based on discrete wavelettransformrdquo Advances in Electrical and Computer Engineeringvol 12 no 4 pp 33ndash38 2012
[17] F Yang and Z Chen ldquoResearch on torque ripple of doublysalient electro-magnetic generator under no-load conditionrdquo
Journal of Sensors 9
Transactions of China Electrotechnical Society vol 28 no 1 pp290ndash295 2013
[18] B A Welchko T A Lipo T M Jahns and S E Schulz ldquoFaulttolerant three-phase AC motor drive topologies a comparisonof features cost and limitationsrdquo IEEE Transactions on PowerElectronics vol 19 no 4 pp 1108ndash1116 2004
[19] X Chen G Lin and Y Zhang ldquoDenoising method based onsparse representation for WFT signalrdquo Journal of Sensors vol2014 Article ID 145870 10 pages 2014
Figure 10 The wavelet packet decomposition results
D2 are open circuit the energy characteristic of the secondfrequency band is higher than the one of the first bandIf there is an open-circuit fault on D1D5 the first and thesecond frequency band are almost the same All the energycharacteristics are very high if there is a D1D2D5 open-circuitfault
The experiments prove that when the generator rectifiercircuit fails the wavelet packet analysis results of voltage onthe sampling signal can be used for the fault diagnosis Withthe established relationship between energy characteristicsand fault state the open-circuit fault can be recognized by thewavelet packet analysis
8 Journal of Sensors
0
1
2
3
4
5
6
7
8
9
1 2 3 4 5 6 7 8
D1D2D5 open circuitNormal D1D5 open circuit
D1D2 open circuitD1 open circuit
Figure 11 Energy eigenvalue statistic on each fault condition
5 Conclusions
The doubly fed brushless starter generator has broad appli-cation prospects in the aerospace and automotive industrywhich need high reliability for the whole system To meetthe requirement of the fault tolerant capability a four-phaseDFBLSG which has the characteristic of phase redundancyphase isolation and fault diagnosis is presented in this paperAs the weak part of the system the rectifier faults are dividedinto one diode or two diodes of the same loop open circuittwo diodes of the same channel open-circuit two diodes ofthe different channel open circuit and three diodes of thedifferent channel open circuit
To detect the open-circuit fault of the rectifier a rectifierfault diagnosis method with wavelet packet analysis is pre-sented Wavelet packet analysis can decompose the voltagesignal into several layers in the whole frequency rangeWith the energy eigenvalues extracted from full-bandwaveletpacket eigenvectors can be established to describe differentfaults The fault tolerant performance and output voltageexperiments were done to gather the energy characteristicswith a voltage sensor With the established relationshipbetween energy characteristics and fault state the open-circuit fault can be diagnosed by the wavelet packet analysis
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This paper is financially supported by the Natural ScienceFoundation of Shandong (no ZR2012EM011) The authorswould like to thank APSC in Nanjing University of Aeronau-tics and Astronautics
References
[1] Z Chen H Wang and Y Yan ldquoA doubly salient start-ergenerator with two-section twisted-rotor structure forpotential future aerospace applicationrdquo IEEE Transactions onIndustrial Electronics vol 59 no 9 pp 3588ndash3595 2012
[2] W Ding and D Liang ldquoA fast analytical model for an integratedswitched reluctance startergeneratorrdquo IEEE Transactions onEnergy Conversion vol 25 no 4 pp 948ndash956 2010
[3] H Mirahki M Moallem and S A Rahimi ldquoDesign opti-mization of IPMSM for 42 v integrated starter alternatorusing lumped parameter model and genetic algorithmsrdquo IEEETransactions on Magnetics vol 50 no 3 pp 114ndash119 2014
[4] C Liu K T Chau and X Zhang ldquoAn efficient wind-photovoltaic hybrid generation system using doubly excitedpermanent-magnet brushless machinerdquo IEEE Transactions onIndustrial Electronics vol 57 no 3 pp 831ndash839 2010
[5] D Lee Y-H Jeong and S-Y Jung ldquoDesign of wound rotorsynchronous machine for ISG and performance comparisonwith interior permanent magnet synchronous machinerdquo Trans-actions of the Korean Institute of Electrical Engineers vol 62 no1 pp 37ndash42 2013
[6] Y Du K T ChauM Cheng Y FanW Zhao and F Li ldquoA linearstator permanentmagnet vernierHTSmachine for wave energyconversionrdquo IEEE Transactions on Applied Superconductivityvol 22 no 3 Article ID 5202505 2012
[7] Z Zhang Y Yan and Y Tao ldquoA new topology of low speed dou-bly salient brushless DC generator for wind power generationrdquoIEEE Transactions on Magnetics vol 48 no 3 pp 1227ndash12332012
[8] L Sun Z Zhang and L Qian ldquoCalculation and analysis of ironloss in Doubly Salient Brushless DC generatorrdquo in Proceedingsof the 39th Annual Conference of the IEEE Industrial ElectronicsSociety pp 2650ndash2655 November 2013
[9] L Zhang B Zhou Y-S Zhang F-S Cheng and J-D WeildquoAnalysis and suppression of torque ripple for doubly salientelectro-magnetic motorrdquo Proceedings of the Chinese Society ofElectrical Engineering vol 30 no 3 pp 83ndash89 2010
[10] W Cao B C Mecrow G J Atkinson J W Bennett and DJ Atkinson ldquoOverview of electric motor technologies used formore electric aircraft (MEA)rdquo IEEE Transactions on IndustrialElectronics vol 59 no 9 pp 3523ndash3531 2012
[11] W Xu J Zhu Y Zhang and J Hu ldquoCogging torque reduc-tion for radially laminated flux-switching permanent magnetmachine with 1214 polesrdquo in Proceedings of the 37th AnnualConference of the IEEE Industrial Electronics Society (IECON rsquo11)pp 3590ndash3595 Melbourne Australia November 2011
[12] C-F Zhang M Yan J He C Luo and Y-Y Xiao ldquoCurrentsensor fault diagnosis for doubly-fed wind generator controlsystemrdquo Information Technology Journal vol 12 no 18 pp4368ndash4373 2013
[13] S Mariani and A Ghisi ldquoUnscented Kalman filtering fornonlinear structural dynamicsrdquo Nonlinear Dynamics vol 49no 1-2 pp 131ndash150 2007
[14] G Zhang and Y Ge ldquoA novel method for gearbox faultdetection based on biorthogonal B-spline waveletrdquo Sensors andTransducers vol 133 no 10 pp 83ndash94 2011
[15] B Zhou JWang S Li andWWang ldquoAnewkey predistributionscheme for multiphase sensor networks using a new deploy-ment modelrdquo Journal of Sensors vol 2014 Article ID 57391310 pages 2014
[16] M Aktas ldquoA novel method for inverter faults detection anddiagnosis in PMSM drives of HEVs based on discrete wavelettransformrdquo Advances in Electrical and Computer Engineeringvol 12 no 4 pp 33ndash38 2012
[17] F Yang and Z Chen ldquoResearch on torque ripple of doublysalient electro-magnetic generator under no-load conditionrdquo
Journal of Sensors 9
Transactions of China Electrotechnical Society vol 28 no 1 pp290ndash295 2013
[18] B A Welchko T A Lipo T M Jahns and S E Schulz ldquoFaulttolerant three-phase AC motor drive topologies a comparisonof features cost and limitationsrdquo IEEE Transactions on PowerElectronics vol 19 no 4 pp 1108ndash1116 2004
[19] X Chen G Lin and Y Zhang ldquoDenoising method based onsparse representation for WFT signalrdquo Journal of Sensors vol2014 Article ID 145870 10 pages 2014
Figure 11 Energy eigenvalue statistic on each fault condition
5 Conclusions
The doubly fed brushless starter generator has broad appli-cation prospects in the aerospace and automotive industrywhich need high reliability for the whole system To meetthe requirement of the fault tolerant capability a four-phaseDFBLSG which has the characteristic of phase redundancyphase isolation and fault diagnosis is presented in this paperAs the weak part of the system the rectifier faults are dividedinto one diode or two diodes of the same loop open circuittwo diodes of the same channel open-circuit two diodes ofthe different channel open circuit and three diodes of thedifferent channel open circuit
To detect the open-circuit fault of the rectifier a rectifierfault diagnosis method with wavelet packet analysis is pre-sented Wavelet packet analysis can decompose the voltagesignal into several layers in the whole frequency rangeWith the energy eigenvalues extracted from full-bandwaveletpacket eigenvectors can be established to describe differentfaults The fault tolerant performance and output voltageexperiments were done to gather the energy characteristicswith a voltage sensor With the established relationshipbetween energy characteristics and fault state the open-circuit fault can be diagnosed by the wavelet packet analysis
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This paper is financially supported by the Natural ScienceFoundation of Shandong (no ZR2012EM011) The authorswould like to thank APSC in Nanjing University of Aeronau-tics and Astronautics
References
[1] Z Chen H Wang and Y Yan ldquoA doubly salient start-ergenerator with two-section twisted-rotor structure forpotential future aerospace applicationrdquo IEEE Transactions onIndustrial Electronics vol 59 no 9 pp 3588ndash3595 2012
[2] W Ding and D Liang ldquoA fast analytical model for an integratedswitched reluctance startergeneratorrdquo IEEE Transactions onEnergy Conversion vol 25 no 4 pp 948ndash956 2010
[3] H Mirahki M Moallem and S A Rahimi ldquoDesign opti-mization of IPMSM for 42 v integrated starter alternatorusing lumped parameter model and genetic algorithmsrdquo IEEETransactions on Magnetics vol 50 no 3 pp 114ndash119 2014
[4] C Liu K T Chau and X Zhang ldquoAn efficient wind-photovoltaic hybrid generation system using doubly excitedpermanent-magnet brushless machinerdquo IEEE Transactions onIndustrial Electronics vol 57 no 3 pp 831ndash839 2010
[5] D Lee Y-H Jeong and S-Y Jung ldquoDesign of wound rotorsynchronous machine for ISG and performance comparisonwith interior permanent magnet synchronous machinerdquo Trans-actions of the Korean Institute of Electrical Engineers vol 62 no1 pp 37ndash42 2013
[6] Y Du K T ChauM Cheng Y FanW Zhao and F Li ldquoA linearstator permanentmagnet vernierHTSmachine for wave energyconversionrdquo IEEE Transactions on Applied Superconductivityvol 22 no 3 Article ID 5202505 2012
[7] Z Zhang Y Yan and Y Tao ldquoA new topology of low speed dou-bly salient brushless DC generator for wind power generationrdquoIEEE Transactions on Magnetics vol 48 no 3 pp 1227ndash12332012
[8] L Sun Z Zhang and L Qian ldquoCalculation and analysis of ironloss in Doubly Salient Brushless DC generatorrdquo in Proceedingsof the 39th Annual Conference of the IEEE Industrial ElectronicsSociety pp 2650ndash2655 November 2013
[9] L Zhang B Zhou Y-S Zhang F-S Cheng and J-D WeildquoAnalysis and suppression of torque ripple for doubly salientelectro-magnetic motorrdquo Proceedings of the Chinese Society ofElectrical Engineering vol 30 no 3 pp 83ndash89 2010
[10] W Cao B C Mecrow G J Atkinson J W Bennett and DJ Atkinson ldquoOverview of electric motor technologies used formore electric aircraft (MEA)rdquo IEEE Transactions on IndustrialElectronics vol 59 no 9 pp 3523ndash3531 2012
[11] W Xu J Zhu Y Zhang and J Hu ldquoCogging torque reduc-tion for radially laminated flux-switching permanent magnetmachine with 1214 polesrdquo in Proceedings of the 37th AnnualConference of the IEEE Industrial Electronics Society (IECON rsquo11)pp 3590ndash3595 Melbourne Australia November 2011
[12] C-F Zhang M Yan J He C Luo and Y-Y Xiao ldquoCurrentsensor fault diagnosis for doubly-fed wind generator controlsystemrdquo Information Technology Journal vol 12 no 18 pp4368ndash4373 2013
[13] S Mariani and A Ghisi ldquoUnscented Kalman filtering fornonlinear structural dynamicsrdquo Nonlinear Dynamics vol 49no 1-2 pp 131ndash150 2007
[14] G Zhang and Y Ge ldquoA novel method for gearbox faultdetection based on biorthogonal B-spline waveletrdquo Sensors andTransducers vol 133 no 10 pp 83ndash94 2011
[15] B Zhou JWang S Li andWWang ldquoAnewkey predistributionscheme for multiphase sensor networks using a new deploy-ment modelrdquo Journal of Sensors vol 2014 Article ID 57391310 pages 2014
[16] M Aktas ldquoA novel method for inverter faults detection anddiagnosis in PMSM drives of HEVs based on discrete wavelettransformrdquo Advances in Electrical and Computer Engineeringvol 12 no 4 pp 33ndash38 2012
[17] F Yang and Z Chen ldquoResearch on torque ripple of doublysalient electro-magnetic generator under no-load conditionrdquo
Journal of Sensors 9
Transactions of China Electrotechnical Society vol 28 no 1 pp290ndash295 2013
[18] B A Welchko T A Lipo T M Jahns and S E Schulz ldquoFaulttolerant three-phase AC motor drive topologies a comparisonof features cost and limitationsrdquo IEEE Transactions on PowerElectronics vol 19 no 4 pp 1108ndash1116 2004
[19] X Chen G Lin and Y Zhang ldquoDenoising method based onsparse representation for WFT signalrdquo Journal of Sensors vol2014 Article ID 145870 10 pages 2014
Transactions of China Electrotechnical Society vol 28 no 1 pp290ndash295 2013
[18] B A Welchko T A Lipo T M Jahns and S E Schulz ldquoFaulttolerant three-phase AC motor drive topologies a comparisonof features cost and limitationsrdquo IEEE Transactions on PowerElectronics vol 19 no 4 pp 1108ndash1116 2004
[19] X Chen G Lin and Y Zhang ldquoDenoising method based onsparse representation for WFT signalrdquo Journal of Sensors vol2014 Article ID 145870 10 pages 2014