PARTIAL DISCHARGE SITE LOCATION SYSTEM FOR ...

Note: Within nine months from the publication of the mention of the grant of the European patent, any person may givenotice to the European Patent Office of opposition to the European patent granted. Notice of opposition shall be filed ina written reasoned statement. It shall not be deemed to have been filed until the opposition fee has been paid. (Art.99(1) European Patent Convention).

Printed by Jouve, 75001 PARIS (FR)

Europäisches Patentamt

European Patent Office

Office européen des brevets

(19)

EP

1 15

3 30

8B

1*EP001153308B1*(11) EP 1 153 308 B1

(12) EUROPEAN PATENT SPECIFICATION

(45) Date of publication and mentionof the grant of the patent:23.11.2005 Bulletin 2005/47

(21) Application number: 99964328.1

(22) Date of filing: 29.12.1999

(51) Int Cl.7: G01R 31/11, G01R 31/10

(86) International application number:PCT/US1999/031032

(87) International publication number:WO 2000/040982 (13.07.2000 Gazette 2000/28)

(54) PARTIAL DISCHARGE SITE LOCATION SYSTEM FOR DETERMINING THE POSITION OFFAULTS IN A HIGH VOLTAGE CABLE

TEILENTLADUNGS-ORTUNGSSYSTEM ZUR FEHLERORTUNG IN EINEMHOCHSPANNUNGSKABEL

SYSTEME DE LOCALISATION DE SITE DE DECHARGE PARTIELLE POUR LOCALISER LESPANNES DANS UN CABLE A HAUTE TENSION

(84) Designated Contracting States:CH DE FR GB LI NL

(30) Priority: 05.01.1999 US 225305

(43) Date of publication of application:14.11.2001 Bulletin 2001/46

(73) Proprietor: HUBBELL INCORPORATEDOrange, Connecticut 06477-4024 (US)

(72) Inventor: FAWCETT, TimothyRuncorn Cheshire WA7 1UB (GB)

(74) Representative: Bubb, Antony John Allen et alWilson GunnChancery House,Chancery LaneLondon WC2A 1QU (GB)

(56) References cited:US-A- 4 491 782 US-A- 5 272 439US-A- 5 416 418 US-A- 5 481 195US-A- 5 530 365 US-A- 5 600 248

EP 1 153 308 B1

2

5

10

15

20

25

30

35

40

45

50

55

Description

Field of the Invention:

[0001] The invention relates to a system for determin-ing the location of faults using multiple passes of pulseactivity on a conductor being tested and averaging ofdetected pulse activity. The invention also relates to asystem which performs partial discharge measurementand enhanced data acquisition using multiple passes ofpulses and averaging without requiring the pulse trig-gering function of a digital oscilloscope.

Background of the Invention

[0002] Partial discharge may occur along cables inelectric power transmission and distribution systemswhen cable insulation breaks down. For example, a cav-ity in cable insulation can cause partial discharges undernormal operating conditions and test conditions whenthe cable is energized, that is, when high voltage signalsare introduced into the cable. When a partial discharge(PD) occurs, high frequency current and voltage pulsesemanate from the site of the discharge, which is here-inafter referred to as a fault site. These current and volt-age pulses can be useful as an indication of the pres-ence of a fault (e.g., an insulation defect) for partial dis-charge site location and assessment. The type of cablefault that causes PD is non-reversible and damage tothe cable progressively degrades the insulation until acatastrophic failure occurs. A cable that is exhibiting PDand is unattended deteriorates due to a combination offactors such as moisture ingression, mechanical fatigueand thermal cycling, among others. It is most cost effec-tive to monitor the cable at the time of manufacture andonce installed to detect partial discharge activity and de-termine its location as soon as such conditions arise toallow for preventive maintenance to be performed be-fore catastrophic failure occurs.[0003] US4491782 and US5272439 disclose meth-ods for locating faults in cables.[0004] US55303365 discloses a method for locatingfaults in cables using a complex scheme of pulse reflec-tion.[0005] Partial discharge site location (PDSL) is a tech-nique for determining the position of a fault within a highvoltage (HV) cable by using detection means at a nearend of the cable to detect pulses which are generatedby PD within the fault. The reflection of those pulsesfrom the far end of the cable is also detected at the nearend. The time difference between the PD pulse and itsreflected pulse is proportional with respect to the dis-tance of the fault site from the far end of the cable.[0006] An existing partial discharge measurement(PDM) system employs a relatively simple dischargesite location system which allows a user to perform asingle time-related capture of pulse activity on a samplecable being tested. A user can employ the graphic user

interface and processing capability of the PDM systemto zoom in on pulses and their reflections and to deter-mine the position of a fault within the cable. This singletime-related capture technique has some deficiencieswhen compared with other equipment such as a digitiz-ing oscilloscope. Since the above-described techniqueis a single shot process, no averaging of the data occurs.Thus, the PDM system sensitivity is limited by the non-correlated noise on the system. A digitizing oscilloscopeis advantageous in that it can average pulse activity bydigitizing the pulse activity in short bursts based on itstime-base setting in response to a triggering event cre-ated by the presence of a pulse. These short bursts ofdigitized pulse activity represent multiple passes of thecable being tested in contrast a single time-related cap-ture. The result of these digitizing operations of multiplepasses are averaged together to produce a compositeimage of the pulse activity. An additional benefit of a dig-itizing oscilloscope is that it allows the possibility of re-solving the presence of multiple sites within the cable.[0007] To provide the performance of a digital oscillo-scope in a PDM system by mimicking the operation ofa digital oscilloscope using pulse triggering is not a via-ble option since the PDM system would need to be re-designed to include features that may not be currentlyincluded in the PDM system hardware such as a setta-ble threshold detection operation on the incoming datastream with an additional data path to process thethreshold signals. Such a redesign is not desirable sinceit may compromise the partial discharge detection andmeasurement facilities that are already included in thePDM system. The PDM system is designed to operateon a pulse-by-pulse basis to identify individual pulseevents rather than relying on a repetitive signal, as istypically used for PDSL measurement. The redesignwould involve considerable effort and expense to in-clude the triggering system of an oscilloscope, as wellas a philosophical change to the operation of the PDMsystem. The PDM system would change from an essen-tially free-running system to a triggered system so that,rather than looking for events in terms of peak heights(i.e., the factor of interest), the system would detectevents crossing a defined threshold. This presents prob-lems for PDM because of a possibility that events willbe overlooked for failure to meet the trigger criteria.[0008] Accordingly, a need exists for a PDM systemwhich can perform pulse averaging for PDSL withouttriggering and other operations associated with oscillo-scopes.

Summary of the Invention

[0009] In accordance with the present invention, a sin-gle digitization system is provided which allows both PD-SL and PDM. Software processing of the informationgenerated by a PDSL system is enhanced to performaveraging functions such as those provided by a digitaloscilloscope, as well as provide automation of the PDSL

1 2

EP 1 153 308 B1

3

5

10

15

20

25

30

35

40

45

50

55

measurement process. An enhanced PDSL system isimplemented in accordance with the present inventionto provide a combination of accuracy, tolerance to noiseand ease of use to the PDSL process.[0010] In accordance with another aspect of thepresent invention, the software processing of the en-hanced PDSL system is adapted for use with conven-tional time domain reflectometry (TDR) systems withouta PD measurement capability to enhance functionalityin the same way.[0011] In accordance with the present invention, aPDSL system is provided to determine cable propaga-tion velocity, as well as a method for using same. ThePDSL system introduces a calibration pulse (i.e., a pulseof insufficient power to cause discharge at faults) into acable to be tested, and data capture operations are per-formed to relate the time required for a pulse to travelthe full length of a cable being tested to the length ofthat cable.[0012] In accordance with yet another aspect of thepresent invention, a method is provided for applying anexcitation pulse to a conductor being tested and for ob-taining a statistical sum of pulse activity on the conduc-tor in response to the excitation pulse. The statisticalsum is obtained from buffering whereby pulse propaga-tion time, cable length traveled by pulses and buffer po-sitions for storing data relating to the pulses are corre-lated such that pulse activity including reflections andinterference occurring in a conductor being tested is rep-resented in predictable buffer positions.[0013] In accordance with the present invention, amethod is provided for determining the location of faultsites in a conductor. A PDSL system comprises a pulsedischarge measurement (PDM) system and is pro-grammed to store the data captured by the PDM systemin a reference buffer corresponding in size to the prop-agation time of a pulse along the length of the conductor.Samples of the captured pulses are scanned to locatethe peaks of pulses above a selected noise level. Thesepulses are stored into a temporary working buffer, alongwith a selected number of samples, normalized and thenadded to the reference buffer. The reference buffer pro-vides a statistical average of pulse activity. Primary orexcitation pulses and their reflections, as well as tran-sient interference pulses and radio frequency interfer-ence, are indicated at the beginning and the end of thereference buffer. Reflections of pulses from fault sitesare represented in the remaining portion of the refer-ence buffer. Fault site distances from the end of the ca-ble can be determined due to the proportional dimensionof the reference buffer with respect to the length of thecable and cable propagation time.

Brief Description of Drawings:

[0014] The various aspects, advantages and novelfeatures of the present invention will be more readilycomprehended from the following detailed description

when read in conjunction with the appended drawings,in which:

Fig. 1 illustrates the configuration of a conductorand a PDSL system constructed in accordance withan embodiment of the present invention to locatefaults in the conductor;Fig. 2 illustrates pulses and reflected pulses ana-lyzed using a PDSL system during length calibra-tion in accordance with an embodiment of thepresent invention;Fig. 3 is a block diagram of a PDSL system con-structed in accordance with an embodiment of thepresent invention;Fig. 4 illustrates waveforms in the PDSL system de-picted in Fig. 3;Fig. 5 illustrates the capture of data by the PDSLsystem in accordance with an embodiment of thepresent invention;Fig. 6 is a flow chart depicting a sequence of oper-ations for calibrating a PDSL system and for deter-mining a distance scaling factor in accordance withan embodiment of the present invention;Fig. 7 is a flow chart depicting a sequence of oper-ations for data acquisition via a PDSL system in ac-cordance with an embodiment of the present inven-tion;Fig. 8 depicts a buffer constructed in accordancewith an embodiment of the present invention; andFigs. 9 and 10 illustrate pulses detected on a con-ductor by a PDSL system and their representationin a reference buffer in accordance with an embod-iment of the present invention.

[0015] Throughout the drawing figures, like referencenumerals will be understood to refer to like parts andcomponents.

Detailed Description Of The Preferred Embodiments

[0016] With reference to Fig. 1, an exemplary electri-cal conductor 10 is illustrated, which is hereinafter re-ferred to as a cable for illustrative purposes. It is to beunderstood that the system and method of analyzingpulse activity of the present invention can be used in aconnection with the testing of different media and differ-ent types of test signals. In the illustrated example, highvoltage signals are used as test signals. A PDSL system12 is connected to the near end of the electrical conduc-tor 10 and is operable to introduce low voltage pulses(CP) therein for calibration. The PDSL system 12 alsoenergizes the electrical conductor 10 using high voltageAC or DC signals. The electrical conductor 10 is there-fore energized to the point where any faults 16 in theelectrical conductor 10 discharge and generate pulsesthat are measured by the PDSL system 12. A cable thatis undergoing testing such as PDSL in a real environ-ment is subject to a number of forms of pulse activity

3 4

EP 1 153 308 B1

4

5

10

15

20

25

30

35

40

45

50

55

which all have an influence on the measurement of theposition of the discharge source.[0017] A primary pulse (PP) is a discharge pulse froma fault site that has propagated down the cable towardthe measuring end without being reflected. The path 18of the PP is depicted in Fig. 1. The first reflection (FR)is the pulse PP' that has propagated along the cable tothe far end 22 (i.e. the end that is furthest from the meas-uring point 24), has been reflected at the far end 22 and,consequently, has traveled toward the measuring end24 (i.e., as exemplified by path 20). The time differencebetween the two pulses PP and FR is proportional to thedistance X between the far end 22 of the cable and thefault site 18. This time difference is useful when perform-ing PDSL measurements.[0018] In accordance with the present invention, thePDSL system 12 performs PDSL in an optimal mannerby measuring the PP and the FR while rejecting essen-tially all other forms of pulse activity detected along thecable. The FR pulse ordinarily has a lower magnitudethan the PP because of the attenuation of the cable. Theeffect is offset somewhat by the positive reflection coef-ficient at the far end 22 of the cable 10. When the atten-uation is low, the FR pulse is generally larger in magni-tude than the primary pulse. The PDSL system 12 of thepresent invention is preferably a measurement systemwhich takes this difference in magnitude into account.[0019] Second and higher order reflections (SHOR),as illustrated in Fig. 2, result when the discharge pulse(e.g., the PP) from a fault 16 reflects along the cable 10from both the near end 24 and the far end 22, decreasingin size as the cable 10 attenuates. Accordingly, multiplereflection pulses can be detected following the FR pulse.SHORs are therefore characterized by equally spacedpulses. The time difference between SHOR pulses isproportional to the length of the cable 10. These SHORpulses are spaced from the PP by the same amounts tor l which are related as described below in connectionwith Fig. 9.[0020] Transient interference pulses (TIPs) can becoupled into the measurement system of the PDSL sys-tem 12 from the external environment via the cable 10or the air, for example. These TIPs manifest themselvesas pulses on either the near end 24 or the far end 22 ofthe cable 10. TIPs consist of a pulse with one or morereflections. The time spacing of these reflections is alsoproportional to the length of the cable and therefore thesame as that for SHORs, as illustrated in Fig. 2. TheTIPs are distinguishable from SHORs, however, in thatthey are not correlated to the PP and the FR. The TIPsoccur randomly with respect to the PP and the FR, asshown in phantom in Fig. 2 for illustrative purposes.[0021] Radio frequency interference (RFI) can becoupled onto the PDSL system as a result of operatingin an open, or an only partially shielded environment. Asa result, spot frequencies or frequencies with sidebandscan be coupled onto the cable 10. This has the effect ofincreasing the background noise level of the system,

which may result in reflections being obscured, espe-cially when the attenuation of the cable 10 is high. Be-cause of the non-stochastic nature of discharge activity,even if the RFI is correlated with the line frequency, itwill not be correlated to the PP and the FR. Thus, theRFI can be eliminated using averaging.[0022] The PDSL system 12 comprises a digital peakdetection measurement system 30 and means for gen-erating high voltage signals 32. Alternatively, the PDSLsystem 12 is operable with an external high voltage pow-er supply. A high voltage power supply which can beincorporated into or connected to the PDSL system is,for example, any of the power supplies based on the970 Series system controller available from Hipotronics,Inc., Brewster, New York. A block diagram of a digitalpartial discharge measurement (PDM) system 30 whichimplements digital peak detection in the PDSL 12 andoperates in accordance with the present invention is de-picted in Fig. 3. Signal waveforms A, B and C at the out-puts of various components in the PDM system 30 areillustrated in Fig. 4. Applications for the PDM system 30include, but are not limited to, testing and monitoringpower cable, distribution and power transformers, me-dium and high voltage switch gear, power circuit break-ers, gas insulated switch gear, bushings, shunt reactors,potential and current transformers, power factor correc-tion capacitors, line insulator products, lightening arre-stors, among other high voltage components and insu-lating materials of all types.[0023] A sample (e.g., an insulation system samplesuch as the cable 10) which is to undergo partial dis-charge detection using the PDM system 30 is connectedto a coupling impedance 34. The PDM system 30 is pref-erably not steady-state and pulses are frequently super-imposed on the high voltage waveform conductedacross the sample 10 from a high voltage source 32.With reference to Fig. 4, the waveform provided to thecoupling impedance 34 is illustrated as waveform A. Themagnitude of the pulse 38 superimposed on the highvoltage waveform A has been exaggerated for illustra-tive purposes. The output of the coupling impedance 34is depicted as waveform B in Fig. 4. Following process-ing by amplifier 36, the pulse can appear as the wave-form C in Fig. 4.[0024] With continued reference to Fig. 3, digital peakdetection is performed by a peak detection circuit 40comprising a digitizer 42, and peak detection andprocessing logic 44. The output of the peak detectionlogic 44 is passed to a buffer memory 46 and subse-quently to a computer 48. The computer 48 is preferablyconnected to a display device 50 and performs otherprocessing and display functions.[0025] The PDM system 30 preferably provides atleast two basic modes of operation for use in differentapplications. The basic modes are: (1) general purposemeasurement and pulse display; and (2) time-depend-ent pulse capture. A pulse capture and analysis modeusing varying windows is also described in the afore-

5 6

EP 1 153 308 B1

5

5

10

15

20

25

30

35

40

45

50

55

mentioned co-pending applications. When operating ingeneral purpose measurement and display mode, thePDM system 30 most closely mimics the operation of atraditional instrument. This mode is optimized to providethe fastest possible update rate on the system display50 to allow the PDM system 30 to mimic the charaaer-istics of an analog cathode-ray oscilloscope, for exam-ple.[0026] With reference to Fig. 5, individual pulses (e.g., pulse 54) are captured in fixed windows (e.g., window56), taking into account the positive and negative peakmagnitudes. The pulses are each stored in a capturememory based on their position in the cycle of a clocksignal 58, and the number of cycles that have elapsedsince the last time the capture memory was read. Hav-ing the phase implicit in the position of a pulse in thecapture memory simplifies the process of writing the in-dividual pulses onto the system display 50 by minimizingthe calculation required. Where pulses occur so closetogether that they occupy a single phase position 56,the PDM system 30 records the highest pulse and indi-cates the highest pulse on the system display 50. Thisavoids the PDM system 30 having to write a pulse to thedisplay 50, only to draw over the current display with alarger pulse. This mode guarantees that the highest dis-charge magnitude pulse is measured, but does notguarantee to resolve all pulses under all situations. Inother words, multiple pulses occurring in one window 56yields one result, while a single pulse crossing two win-dows yields two results. This mode does, however, pro-vide a fast update rate (e.g., typically 25 times per sec-ond) combined with an accurate discharge magnitudemeasurement. The display produced looks like that onan analog display, in that it is bipolar and can display theovershoot on a pulse (i.e., a pulse occurring concurrent-ly with the tail of another pulse). This makes the PDMsystem 30 comfortable for a user used to traditional in-struments. The pulse capture and analysis mode is pre-ferred over the general operation mode when it is re-quired to look at the statistics of discharge activity. Forexample, when discharge fingerprinting is to be per-formed, all of the pulses in a defined interval can be cap-tured via the pulse capture and analysis mode. None-theless, the general purpose mode is advantageous forcapturing the overshoot of the pulses to provide a morerealistic display.[0027] The second mode of operation, that is, thetime-dependent pulse capture mode, is the simplestmode of operation. Once the PDM system 30 is trig-gered in this mode (i.e., using fixed time windows asshown in Fig. 5), the PDM system 30 fills up a pulsecapture memory with successive samples taken fromthe digitizer 42. In this mode, no attempt is made to cap-ture the peaks of the pulses. Accordingly, no measure-ment of discharge magnitude is made. The primary useof this mode is for fault location in cables. The positionof pulses within the cable 10 can be found by measuringthe time interval between a pulse and its reflection. By

comparison with the time for a pulse to travel the fulllength of the cable 10 and return to an originating point(e.g., measuring point 24), the position of the fault 16from the far end 22 of the cable 10 can be determined.This mode also provides diagnostic facilities because itallows the pulse shape to be studied to optimize thebandwidth of the system. Also, because of the time-based capture, this mode allows the measurement ofinterference frequencies such that suitable filtering canbe implemented.[0028] The computer 48 in the digital discharge de-tection PDM system 30 is preferably a personal compu-ter, for example. The computer 48 is figured to allow flex-ible test recording and data exporting to different soft-ware programs such as Word™ and Excel™ The com-puter 48 is programmed to provide a flexible analysistool for digital partial discharge detection. Pulse captureis achieved against phase or time coordinates. Differentmodes of operation are provided for full control over gat-ing of pulses in both the vertical and horizontal axes.FPGA technology is preferably used for peak detectionand operations (i.e., peak detection logic 44) in conjunc-tion with the central processing unit board of the com-puter 48. Pulses from the amplifier 36 are provided tothe FPGA peak detector 40 which comprises a digitizer42 (hereinafter referred to as an analog-to-digital con-verter (ADC)). The ADC 42 is preferably a 10-bit ADCto provide 9-bit resolution a sign bit. As described in theco-pending applications, the output of the ADC 42 is pro-vided to a two-stage pipeline comprising an ADC bufferand a peak buffer. The values in these buffers representtwo stages in a pipeline and are provided to a magnitudecomparator. A state machine controller in the peak de-tection logic 44 provides gate control to determine thetime window within which peak detection is performed.The state machine also controls the clocking of the pipe-line and resetting of values in the pipeline.

Automated Cable Length Calibration

[0029] Before PDSL measurements are taken, thepropagation time of a pulse travelling the full length ofthe cable is related to the length of the cable 10, in ac-cordance with the present invention. A calibration pulseis injected into the near end 24 of the cable 10 (i.e., atthe measuring point). The calibration pulse (CP) prefer-ably does not energize the cable 10 sufficiently for dis-charge activity to occur at faults, should any faults existin the cable. The time difference between the injectionof the CP and the first reflection thereof (CP') is meas-ured. The factor to relate the time difference between aPP and its reflection (i.e., FR) to the distance of the faultsite 16 from the far end 22 of the cable 10 is obtainedfrom the time difference between the calibration pulse(CP) and its reflection (CP') from the far end 22 of thecable 10.[0030] To automatically calibrate, the PDSL system12 can be used to provide calibration pulses to the near

7 8

EP 1 153 308 B1

6

5

10

15

20

25

30

35

40

45

50

55

end 24 of the cable 10 and to trigger the PDSL captureof the system. Since the PDSL capture in the PDM sys-tem 30 does not occupy a full cycle of the power source32, the PDM 30 is operable to ensure that the PDSLcapture has a calibration pulse in it. Thus, the PDM sys-tem 30 is advantageous in that is it configured to allowthe starting phase of the PDSL capture to be set and thephase position to be known. An important aspect of thecalibration of the PDSL system 12 is that the pulse isvisible above the noise.[0031] To perform automatic length calibration, thePDSL system 12 runs in the general measurement andpulse display mode, using the amplifier 36 to measurethe background noise level of the system. The level isrecorded, as indicated in block 60 in Fig. 6. The PDSLsystem 12 is programmable to commence calibration af-ter the computer 48 has determined that a CP is intro-duced with a peak height of at least 50% greater thanthe recorded background noise level (block 62). Withreference to block 64, the PDSL system 12 then prefer-ably switches to the time-dependent pulse capturemode (hereinafter referred to as the PDSL mode) andtriggers a series of captures (block 66). The data frommultiple captures is combined as described below inconnection with Fig. 6.[0032] In accordance with the present invention, thePDSL system 12 is programmed via the peak detectionlogic 44 to establish a primary reference buffer in thebuffer memory 46) and to clear the primary referencebuffer (block 68). PDSL measurement is commenced inthe PDSL mode. Accordingly, data is captured into asecond buffer in the buffer memory 46 (blocks 70 and72). The data represents a time-based series of datapoints that provide ADC output at a particular instant.[0033] The PDSL system 12 searches the data in thesecondary buffer until the peak of the calibration pulseis located, which is the highest value point recorded inthe secondary buffer (block 74). The data in the second-ary buffer is shifted until this peak value is stored at apredetermined position or register in the buffer (e.g.,10% into the buffer), as indicated in block 76. This elim-inates problems due to jitter or synchronization drift byrepositioning the peak to a known point. The data in thesecondary buffer is added to that in the primary buffer(block 78). The process is repeated a number of times(e.g., between 50 and 100 times) to ensure that all non-correlated noise sources in the pulse stream (e.g., TIPsor RFI) are eliminated. With reference to Fig. 2, the CP,its reflection CP', as well as TIPs and SHORs, occur atsimilar time intervals t corresponding to the length l ofthe cable 10.[0034] With continued reference to Fig. 6 and the neg-ative branch of decision block 66, the PDSL system 12searches for the zero point of the CP in the secondarybuffer and notes the CP position (block 80). The CP inFig. 2 is illustrated as the first pulse φ[A]. The PDSL sys-tem 12 subsequently scans the secondary buffer for theCP reflection CP', assuming the reflection CP' is the sec-

ond highest pulse (i.e., neglecting the first highest pulseCP). The PDSL system 12 determines the zero point ofthe reflection CP' and the corresponding position of thezero value in the secondary buffer (block 82). The re-flection CP' is illustrated in Fig. 2 as the second pulse φ[B]. The time difference between the CP and its reflec-tion CP' is calculated (block 84), and a distance scalingfactor is calculated from that time difference (block 86).For example, the distance scaling factor can be l/t wherethe length l of the cable is known and the time differenceis calculated. The distance scaling factor l/t is alsoequivalent to (l x f)/n wherein f is the sampling frequencyin the PDSL mode and n is the number of samples takenbetween pulses φ[A] and φ[B]. The automatic calibrationdescribed in connection with Fig. 6 need not be used toscale measurements if the cable propagation velocity isknown.

Data Acquisition and Processing

[0035] Once the PDSL system 12 has been calibrat-ed, the PDSL system 12 is prepared to perform data ac-quisition and processing operations, which are de-scribed below in connection with Figs. 7 and 8. The PD-SL system 12 preferably captures data over a prede-fined number of acquisitions, set by the user or somesuitable default setting. The amount of data that is gath-ered is a compromise between obtaining a sufficientamount of data to ensure that the full length of the con-ductor 10 is measured, and avoiding the capture of anexcessive amount of data and increased processingtime. The computer can determine the amount of databased on the amount of memory left after the PP is de-tected. The user can override the computer-determinedamount if the user can estimate how much data is nec-essary, based on experience of the transit time or prop-agation speed of a particular type of conductor or cable.For example, the PDM system 30 can capture 0.25 Meg-abytes (MB) of information on each capture. Since, at aline frequency of 60 Hz, an whole entire power mainscycle may not be covered, it is necessary for the PDSLsystem 12 to know where on the mains cycle the datacapture process is to start to ensure digitizing of validdischarge pulses. The starting point can be determinedfrom the pulse information gathered while operating inthe pulse display mode.[0036] After calibration, a high voltage signal is ap-plied to the cable 10 which is sufficient for the cable 10to discharge at any fault sites 16 therein (block 90 of Fig.7). This high voltage energization can be generated, forexample, using the aforementioned Series 970 control-ler connected to, or incorporated in, the PDSL system12. If the PDSL system 12 is using a Series 970 control-ler or similar device, there are at least two possibilitiesfor applying voltage. First, a voltage level defined by theuser or the PDSL system 12 can raise the voltage untila level of activity defined by the user is achieved. Oncea voltage is achieved whereby the sample discharge oc-

9 10

EP 1 153 308 B1

7

5

10

15

20

25

30

35

40

45

50

55

curs, the voltage is preferably held until testing is com-pleted.[0037] Once the sample 10 is at a selected voltage fordischarging, the starting point for digitization is deter-mined (block 92). The starting position can be deter-mined automatically if the discharge activity is visibleover the background noise. It is to be understood thatonly the primary pulses (PPs) need to be visible. Alter-natively, the user can indicate the starting position of thedata capture sequence. For example, the user can setthe starting position by observing the position of dis-charge activity on the normal phase-related dischargedisplay and choosing an appropriate value. Once astarting point is established, the PDSL system 12 com-mences the data acquisition process (block 94). Dataacquisition preferably involves a defined number of cap-ture phases, the number of which can be set by the user.The higher the number of captures, the more statisticallyrelevant the captured data is.[0038] The PDSL system 12 establishes a referencebuffer 93 for the final gathered data which is illustratedin Fig. 8. The reference buffer 93 is sized according tothe time difference between a calibration pulse and itsreflection as measured during the calibration phase(Fig. 6). The use of the distance scaling factor deter-mined during calibration ensures that the data buffer 93corresponds to the length of the cable 10, in addition tosmall guard bands 97 and 99 of registers 95 added ontoeach end of the buffer 93. The contents of this referencebuffer are set to zero, as indicated in block 96 of Fig. 7.For each of the data captures, the PDSL system 12 per-forms a sequence of operations. After the PDSL system12 commences the capture mode, as indicated in thedecision block 98, the PDSL system 12 stores outputvalues from the digitizer 42 (e.g., 10-bit representationsof a waveform processed via an analog-to-digital con-verter) and continues to store values until 256K sampleshave been captured in the buffer memory 46 (block100). The data from the PDSL capture is transferred intoa working buffer in the buffer memory 46 (block 102).The system scans along the working buffer until it findsa pulse that exceeds the noise floor of the system (block104). The PDSL system 12 locates the peak of thatpulse. Once the peak has been found, the correspond-ing pulse is copied into a temporary buffer, along with anumber of samples following the pulse (block 106). Thenumber of the samples following the pulse, which arestored in a temporary buffer in the buffer memory 46,corresponds to the size of the reference buffer andtherefore the length of the cable. The entries in the tem-porary buffer are normalized to make the magnitude ofthe first pulse unity (block 108). The entries in the tem-porary buffer are added to the corresponding entries inthe reference buffer (block 110). The data in the workingbuffer continues to be scanned and processed as de-scribed with reference to blocks 104, 106, 108, 110 untilthe entire working buffer is scanned (block 112).[0039] Once the data acquisition process has been

performed, the reference buffer holds a sequence of da-ta that represents the activity within the length of the ca-ble.

Analysis of the Reference Buffer Data

[0040] Once the reference buffer (e.g., buffer 93 inFig. 8) has been filled, it can be analyzed to identify ac-tivity within the cable. The contents of the reference buff-er represent a statistical average of the history of pulsesover the time that elapsed while the pulses traveled thefull length of the cable 10. With reference to Fig. 9, anumber of pulses A, B, C and D traveling along the cable24 and measured by the PDSL system 12 at the meas-uring end 24 are depicted in Fig. 9. The pulses A, B, Cand D are also depicted individually in Fig. 10, alongwith the history of the pulses 120 as represented in thereference buffer. For illustrative purposes, the pulse Boccurred at a time interval corresponding to the distanced on the cable, as depicted in Fig. 1. The pulses C andD each occurred at a time interval corresponding to l orthe length of the table.[0041] The data detected at the near end or measur-ing point 24 of the cable 10 is depicted on the left of theraw data shown in Figs. 9. The contents of the referencebuffer provide a plot of activity along the length of thecable, as indicated at 120 in Fig. 10. The contents in thereference buffer corresponding to data at the measuringend 24 of the cable 10 essentially always comprises apulse 122 which is the statistical sum of all of the pulses(e.g., pulses A, B, C and D) recorded during the dataacquisition phase. The shaded area 124 in Fig. 10 rep-resents the environment external to the cable, while theunshaded area corresponds to activity in the cable 10.[0042] If the attenuation of the cable is low, there isthe possibility of SHORs occurring. SHORs appear onthe final result 120 in the reference buffer as a contribu-tion to the pulse 122 at the start of the reference bufferand a corresponding pulse 128 at the far end of the buff-er. This is so because, as stated previously, SHORs arespaced in accordance with the time corresponding topropagation of a pulse along the length of the cable.SHORs are therefore spaced apart or separated in timeby this same amount. TIPs appear in the same way asthey manifest on either one end 22 of the cable 10 orthe other end 24.[0043] The effect of SHORs and TIPs, therefore is toaccentuate the height of the pulse 122 at the start of thereference buffer and to produce a pulse 128 on the endof the buffer, thereby indicating the start and the finishpoints of the cable 10, respectively. Because RFI, whichdoes not correlate to the pulse activity on the sample, isaveraged over the length of the cable for a large numberof captures, RFI is seen as an offset of the data in thereference buffer. Thus, the effect of interferences aresuppressed by the present invention.[0044] The data of interest to most users are the PPsand FRs. The PPs (e.g., pulse A in Figs. 9 and 10) occur

11 12

EP 1 153 308 B1

8

5

10

15

20

25

30

35

40

45

50

55

at the start of the cable 10 and therefore accentuate thefirst pulse 122. The FRs (e.g., pulse B) differ from theinterferences and other reflections in that they are notspaced from the PP by the length of the cable, but ratherby some lesser amount d. As the statistical averagesare obtained, the effect of the PPs and FRs is to placepulses (e.g., pulse E in Fig. 10) inside the cable and notjust on the ends as with SHORs and TIPs. Accordingly,data corresponding to FRs, for example, is containedsomewhere in the reference buffer other than at theends of the buffer. The FRs indicate fault sites 16. Thelocation of the faults within the cable 10 therefore areindicated. Since the reference buffer length corre-sponds to the time elapsing as a pulse propagates alongthe length l of the cable, the time difference between theleft side of the reference buffer and the pulse E indicatesthe distance d from the near end 24 of the cable or thedistance x from the far end 22 of the cable. If no pulsesare indicated between the ends of the buffer, a user caninfer that a problem is close to the ends 22 and 24 of thecable 10, in the termination system at the far end 22 orexternal to the cable 10.[0045] If the PDSL system 12 is operating with a Se-ries 970 system controller, the user needs only enter thevoltage at which the cable is to be tested, the numberof capture cycles to be performed and either the lengthof the cable or its velocity of propagation. These param-eters can be embedded in a test specification loadedfrom a file into the PDSL system 12. Once the PDSLsystem 12 has the required information, a test can beperformed by pressing a button and waiting for the datato be produced. Thus, no intervention by the user is re-quired. If the number of captures is sufficient to averageout the non-correlated noise on PDSL system 12, thePDSL system can automatically locate the position ofmultiple faults within the cable 10, allowing the systemto be used by unskilled operators with minimal training.[0046] When multiple faults exist within the cable 10,it is likely that their respective discharge rates differ.Thus, the relative frequency of occurrence of the pulsesis different and therefore the heights of the correspond-ing averaged pulses are different. Because the data arenormalized based on the height of the primary pulse,there is no effect on the height of the first reflection puls-es seen on the trace of the PDSL system display 50 asa result of differences in discharge magnitude. In the ref-erence buffer, the height of pulses within the cable 10are dependent on two factors, that is, the relative fre-quency of occurrences and the attenuation of the cable.In accordance with the PDSL system of the present in-vention, the effect of cable attenuation is proportional tothe distance of the fault from the far end of the cable.Thus, it is possible to determine the attenuation per unitlength of cable from the relative height of the calibrationpulse and its reflection. By applying this correction tothe reference buffer, the height of pulses correspondingto faults within the cable is proportional to the relativefrequency of occurrence, indicating the level of activity

of the faults.[0047] Although the present invention has been de-scribed with reference to a preferred embodiment there-of, it will be understood that the invention is not limitedto the details thereof. Various modifications and substi-tutions have been suggested in the foregoing descrip-tion, and others will occur to those of ordinary skill in theart. All such substitutions are intended to be embracedwithin the scope of the invention as defined in the ap-pended claims.

Claims

1. A method for determining the location of faults alongan electrical conductor (10) having a near end (24)and a far end (22) comprising the steps of:

applying a voltage signal to the near end of saidelectrical conductor to energize said electricalconductor sufficiently for discharging at anysaid fault therein, said electrical conductor be-ing operable to propagate at least one of a plu-rality of pulses comprising a primary pulse orig-inating at any said fault in response to the ex-citation of the electrical conductor by said volt-age signal and traveling toward the near end ofsaid electrical conductor, a first reflected pulsecorresponding to the reflection of said primarypulse at the far end of said electrical conductor,a second order pulse and a higher order pulsecorresponding to the reflection of said first re-flected at a corresponding one of the near endand the far end of said electrical conductor, anda transient interference pulse coupled to saidelectrical conductor via the environment sur-rounding said electrical conductor;storing data relating to the length of said elec-trical conductor, said data being selected fromthe group consisting of the propagation time forat least one of said first pulse and said pluralityof pulses to travel along said electrical conduc-tor, the length of said electrical conductor, andvelocity of said at least one of said primarypulse and said plurality of pulses travelingalong said electrical conductor;establishing a reference buffer correspondingto the length of said electrical conductor usingsaid data;initializing said reference buffer to zero;obtaining a plurality of samples from said elec-trical conductor corresponding to the amplitudeof any of said plurality of pulses;storing said plurality of samples in a workingbuffer;scanning said plurality of samples in said work-ing buffer to locate a first one of said pluralityof pulses having an amplitude level that is

13 14

EP 1 153 308 B1

9

5

10

15

20

25

30

35

40

45

50

55

greater than a predetermined signal level;storing selected ones of said plurality of sam-ples as respective entries in a temporary buffer,said selected samples comprising samples cor-responding to said first one of said plurality ofpulses and subsequent ones of said plurality ofsamples stored in said working buffer, thenumber of said selected samples correspond-ing to the length of said electrical conductor;normalizing said entries in said temporary buff-er such that the magnitude of said first one ofsaid pulses is unity;adding said entries in said temporary buffer tocorresponding entries in said reference buffer;andrepeating said obtaining step, said storing stepfor storing said plurality of samples, said scan-ning step, said storing step for storing said se-lected samples, said normalizing step and saidadding step, said entries in said reference buff-er representing a statistical average of saidpulses during a period of time corresponding tothe amount of time required for one of said puls-es to travel the length of said electrical conduc-tor.

2. A method as claimed in claim 1, wherein said tem-porary buffer comprises entries for storing data re-lating to said plurality of samples, said storing stepfor storing said selected samples in said temporarybuffer comprises the step of providing said selectedsamples occurring at the near end and the far endof said electrical conductor in said temporary bufferat corresponding ones of entries located at the be-ginning and at the end of said temporary buffer andthe remaining said selected samples being storedin consecutive order using said entries between thefirst and the last of said entries.

3. A method as claimed in claim 2, wherein said pro-viding step comprises the step of providing a select-ed number of entries prior to and after said entrieslocated at the beginning of and at the end of saidtemporary buffer, respectively, to operate as guardbands with respect to said selected samples.

4. A method as claimed in claim 1, wherein said re-peating step is performed a predetermined numberof times to define a selected data acquisition period,and further comprising the step of analyzing saidelectrical conductor using said entries in said refer-ence buffer, one of said entries in said referencebuffer corresponding to the statistical sum of all ofsaid pulses captured via each said repeating stepduring said data acquisition period.

5. A method as claimed in claim 4, wherein said oneof said entries corresponding to said statistical sum

is located at least proximally with respect to said en-tries located at the beginning of and at the end ofsaid reference buffer, said analyzing step compris-ing the step of analyzing said entries in said refer-ence buffer as representing different ones of saidplurality of pulses at respective portions along thelength of said electrical conductor.

6. A method as claimed in claim 1, further comprisingthe steps of:

identifying said first reflected pulse correspond-ing to each said fault on said electrical conduc-tor as a corresponding one of said entries thatis not proximal with respect to the beginning ofand the end of said reference buffer and has aselected magnitude; anddetermining the location of each said fault onsaid electrical conductor by correlating the lo-cation of the corresponding said entries indicat-ing each said first reflected pulse with respectto a point along the length of said electrical con-ductor.

7. A method as claimed in claim 6, wherein said elec-trical conductor can be subjected to radio frequencyinterference, further comprising the step of disre-garding different ones of said plurality of pulses re-lating to either of transient interference pulse andradio frequency interference since they are indicat-ed at the beginning of and at the end of said refer-ence buffer.

8. A method as claimed in claim 6, wherein said elec-trical conductor can be subjected to radio frequencyinterference, further comprising the step of identify-ing radio frequency interference with respect to saidelectrical conductor as an offset value with respectto said selected samples in said reference buffer.

9. A method as claimed in claim 6, further comprisingthe step of determining if any of said second orderpulse, said higher order pulse, and said transientinterference pulse occurred using said entries at thebeginning of and at the end of said reference buffer.

10. A partial discharge site location system for locatingfaults along the length of an electrical conductorcomprising:

a power supply device for energizing said elec-trical conductor;a partial discharge measurement system (12)connected to said electrical conductor and op-erable to identify individual pulses occurring onsaid electrical conductor, said pulses corre-sponding to discharge at any fault sites alongsaid electrical conductor as a result of ener-

15 16

EP 1 153 308 B1

10

5

10

15

20

25

30

35

40

45

50

55

gization by said power supply device, said par-tial discharge system comprising a processor(44), a memory device (46), and an analog-to-digital converter (42), said processor being pro-grammed to perform a pulse capture operationto represent said pulses as digitized samplesof pulse amplitudes from said analog-to-digitalconverter, said digitized samples being storedin said memory device; anda processing device (48) programmed to per-form at least one of a plurality of operationscomprising a calibration operation to relate thepropagation time of one of said pulses travelingthe entire length of said electrical conductor tothe length of said electrical conductor, and a da-ta acquisition operation wherein said digitizedsamples from said pulse measurement systemare obtained, said processing device being pro-grammed in accordance with said data acqui-sition operation to create a reference buffer cor-responding to the length of said electrical con-ductor, to store said digitized samples from onesaid pulse capture operation in a working buff-er, to scan said working buffer to locate said dig-itized samples corresponding to peaks of puls-es above a selected noise level and store saiddigitized samples corresponding thereto into atemporary buffer along with a selected numberof said digitized samples acquired thereafter,normalizing said digitized samples in said te-morary buffer, and adding said digitized sam-ples in said temporary buffer to said referencebuffer.

11. A partial discharge site location system as claimedin claim 10, wherein said processing device is pro-grammed to repeat said data acquisition operationa selected number of times to obtain a statistical av-erage of pulse activity along said electrical conduc-tor in said reference buffer, first reflections of a dis-charge pulse from any fault along said electricalconductor being identified in between the beginningand the end of said reference buffer.

Patentansprüche

1. Verfahren zum Bestimmen des Orts von Fehlernentlang eines elektrischen Leiters (10), der ein na-hes Ende (24) und ein entferntes Ende (22) besitzt,das die Schritte aufweist:

Anlegen eines Spannungssignals an das naheEnde des elektrischen Leiters, um den elektri-schen Leiter ausreichend für eine Entladung anirgendeinem solchen Fehler darin mit Energiezu beaufschlagen, wobei der elektrische Leiterso betreibbar ist, um mindestens einen einer

Vielzahl von Impulsen, aufweisend einen pri-mären Impuls, der an irgendeinem solchenFehler, in Abhängigkeit der Anregung des elek-trischen Leiters, durch das Spannungssignalausgeht und zu dem entfernten Ende des elek-trischen Leiters hin läuft, einen ersten, reflek-tierten Impuls entsprechend zu der Reflexiondes primären Impulses an dem entfernten En-de des elektrischen Leiters, einen Impuls zwei-ter Ordnung und einen Impuls höherer Ord-nung, entsprechend zu der Reflexion des er-sten reflektierten, an einem entsprechenden ei-nen des nahen Endes und des entfernten En-des des elektrischen Leiters, und einen Über-gangs-Interferenz-Impuls, gekoppelt zu demelektrischen Leiter über die Umgebung, die denelektrischen Leiter umgibt, zu propagieren;Speichern von Daten, die sich auf die Längedes elektrischen Leiters beziehen, wobei dieDaten aus der Gruppe ausgewählt sind, die ausder Propagationszeit für mindestens entwederdes ersten Impulses oder der Vielzahl der Im-pulse, die entlang des elektrischen Leiters lau-fen, der Länge des elektrischen Leiters und derGeschwindigkeit des mindestens einen des pri-mären Impulses und der Vielzahl von Impulsen,die entlang des elektrischen Leiters laufen, be-steht;Einrichten eines Referenzpuffers entspre-chend zu der Länge des elektrischen Leitersunter Verwendung der Daten;Initialisieren des Referenzpuffers auf Null;Erhalten einer Vielzahl von Abtastungen vondem elektrischen Leiter entsprechend zu derAmplitude irgendeines der Vielzahl der Impul-se;Speichern der Vielzahl der Abtastungen in ei-nem Arbeitspuffer;Abtasten der Vielzahl der Abtastungen in demArbeitspuffer, um einen ersten einen der Viel-zahl der Impulse zu lokalisieren, der einen Am-plituden-Pegel besitzt, der größer als ein vor-bestimmter Signalpegel ist;Speichern ausgewählter solcher der Vielzahlder Abtastungen als jeweilige Eintritte in einemtemporären Puffer, wobei die ausgewähltenAbtastungen solche Abtastungen aufweisen,die dem ersten einen der Vielzahl der Impulseund darauf folgenden solchen der Vielzahl derAbtastungen, gespeichert in dem Arbeitspuffer,entsprechen, wobei die Zahl der ausgewähltenAbtastungen der Länge des elektrischen Lei-ters entspricht;Normieren der Eintritte in dem temporären Puf-fer so, dass die Größe des ersten einen der Im-pulse eine Einheit ist;Addieren der Eintritte in dem temporären Pufferzu entsprechenden Eintritten in dem Referenz-

17 18

EP 1 153 308 B1

11

5

10

15

20

25

30

35

40

45

50

55

puffer; undWiederholen des Schritts des Erhaltens, desSpeicherschritts zum Speichern der Vielzahlder Abtastungen, des Abtastschritts, des Spei-cherschritts zum Speichern der ausgewähltenAbtastungen, des Normierungsschritts und desAddierschritts, wobei die Eintritte in den Refe-renzpuffer einen statistischen Durchschnitt derImpulse während einer Zeitperiode entspre-chend zu der Zeitdauer darstellen, die für einender Impulse erforderlich ist, um entlang derLänge des elektrischen Leiters zu laufen.

2. Verfahren nach Anspruch 1, wobei der temporärePuffer Eintritte zum Speichern von Daten aufweist,die sich auf die Vielzahl der Abtastungen beziehen,wobei der Speicherschritt zum Speichern der aus-gewählten Abtastungen in dem temporären Pufferden Schritt eines Bereitstellens der ausgewähltenAbtastungen, die an dem nahen Ende und an dementfernten Ende des elektrischen Leiters auftreten,in dem temporären Puffer an entsprechenden sol-chen von Eintritten, die an dem Anfang und an demEnde des temporären Puffers angeordnet sind, undwobei die verbleibenden der ausgewählten Abta-stungen in einer aufeinander folgenden Reihenfol-ge, unter Verwendung der Eintritte zwischen demersten und dem letzten der Eintritte, gespeichertwerden, aufweist.

3. Verfahren nach Anspruch 2, wobei der Bereitstel-lungsschritt den Schritt eines Bereitstellens einerausgewählten Anzahl von Eintritten vor und nachden Eintritten, angeordnet an dem Anfang und andem Ende des temporären Puffers, jeweils, um alsSchutzbänder in Bezug auf die ausgewählten Ab-tastungen zu arbeiten, aufweist.

4. Verfahren nach Anspruch 1, wobei der Wiederho-lungsschritt eine vorbestimmte Anzahl von Malendurchgeführt wird, um eine ausgewählte Daten-Ak-quisitions-Periode zu definieren, und weiterhin denSchritt eines Analysierens des elektrischen Leitersunter Verwendung der Eintritte in den Referenzpuf-fer aufweist, wobei einer der Eintritte in den Refe-renzpuffer entsprechend zu der statistischen Sum-me aller der Impulse über jeden solchen Wiederho-lungsschritt während der Daten-Akquisitions-Peri-ode erfasst wurde.

5. Verfahren nach Anspruch 4, wobei der eine der Ein-tritte entsprechend zu der statistischen Summe zu-mindest proximal in Bezug auf die Eintritte, ange-ordnet an dem Beginn und an dem Ende des Refe-renzpuffers, angeordnet ist, wobei der Analysie-rungsschritt den Schritt eines Analysierens der Ein-tritte in den Referenzpuffer als unterschiedliche sol-che der Vielzahl der Impulse an jeweiligen Berei-

chen entlang der Länge des elektrischen Leitersdarstellend aufweist.

6. Verfahren nach Anspruch 1, das weiterhin dieSchritte aufweist:

Identifizieren des ersten reflektierten Impulsesentsprechend zu jedem solchen Fehler an demelektrischen Leiter als ein entsprechender ei-ner der Eintritte, der nicht proximal in Bezug aufden Anfang und das Ende des Referenzpuffersvorhanden ist und eine ausgewählte Größe be-sitzt; undBestimmen des Orts jedes solchen Fehlers andem elektrischen Leiter durch Korrelieren desOrts des entsprechenden der Eintritte, die je-weils den ersten reflektierten Impuls in Bezugauf einen Punkt entlang der Länge des elektri-schen Leiters anzeigen.

7. Verfahren nach Anspruch 6, wobei der elektrischeLeiter einer Funkfrequenz-Interferenz unterworfenwerden kann, das weiterhin den Schritt eines Ver-nachlässigens unterschiedlicher solcher der Viel-zahl der Impulse, die sich auf entweder einen Über-gangs-Interferenz-Impuls oder eine Funkfrequenz-Interferenz beziehen, da sie an dem Beginn und andem Ende des Referenzpuffers angezeigt sind, auf-weist.

8. Verfahren nach Anspruch 6, wobei der elektrischeLeiter einer Funkfrequenz-Interferenz unterworfenwerden kann, das weiterhin den Schritt eines Iden-tifizierens einer Funkfrequenz-Interferenz in Bezugauf den elektrischen Leiter als einen Offset-Wert inBezug auf die ausgewählten Abtastungen in demReferenzpuffer aufweist.

9. Verfahren nach Anspruch 6, das weiterhin denSchritt eines Bestimmens, ob irgendeiner des Im-pulses zweiter Ordnung, des Impulses höherer Ord-nung und des Übergangs-Interferenz-Impulses auf-trat, unter Verwendung der Eintritte an dem Anfangund an dem Ende des Referenzpuffers, aufweist.

10. Teilentladungs-Ortungssystem zur Fehlerortungentlang der Länge eines elektrischen Leiters, dasaufweist:

eine Energieversorgungsvorrichtung zur Ener-giebeaufschlagung des elektrischen Leiters;ein Teilentladungs-Messsystem (12), verbun-den mit dem elektrischen Leiter und so betreib-bar, um individuelle Impulse zu identifizieren,die an dem elektrischen Leiter auftreten, wobeidie Impulse einer Entladung an irgendwelchenFehlerstellen entlang des elektrischen Leitersals eine Folge einer Energiebeaufschlagung

19 20

EP 1 153 308 B1

12

5

10

15

20

25

30

35

40

45

50

55

durch die Energieversorgungsvorrichtung ent-sprechen, wobei das Teilentladungssystem ei-nen Prozessor (44), eine Speichervorrichtung(46) und einen Analog-Digital-Wandler (42)aufweist, wobei der Prozessor so programmiertist, um einen Impulserfassungsvorgang durch-zuführen, um die Impulse als digitalisierte Ab-tastungen von Impulsamplituden von dem Ana-log-Digital-Wandler darzustellen, wobei die di-gitalisierten Abtastungen in der Speichervor-richtung gespeichert werden; undeine Verarbeitungsvorrichtung (48), program-miert so, um mindestens einen einer Vielzahlvon Vorgängen, aufweisend einen Kalibrie-rungsvorgang, um die Propagationszeit einesder Impulse, der über die gesamte Länge deselektrischen Leiters läuft, zu der Länge deselektrischen Leiters in Relation zu setzen, undeinen Daten-Akquisitionsvorgang, bei dem diedigitalisierten Abtastungen von dem Impuls-messsystem erhalten werden, durchzuführen,wobei die Verarbeitungsvorrichtung so ent-sprechend zu dem Daten-Akquisitionsvorgangprogrammiert ist, um einen Referenzpuffer ent-sprechend zu der Länge des elektrischen Lei-ters zu erzeugen, um die digitalisierten Abta-stungen von einem solchen Impulserfassungs-vorgang in einem Arbeitspuffer zu speichern,um den Arbeitspuffer abzutasten, um die digi-talisierten Abtastungen entsprechend zuPeaks von Impulsen oberhalb eines ausge-wählten Rauschpegels zu lokalisieren und umdie digitalisierten Abtastungen entsprechenddazu in einem temporären Puffer zusammenmit einer ausgewählten Anzahl der digitalenAbtastungen, die danach erhalten sind, zuspeichern, wobei die digitalisierten Abtastun-gen in dem temporären Puffer normiert werdenund wobei die digitalisierten Abtastungen indem temporären Puffer zu dem Referenzpufferaddiert werden.

11. Teilentladungs- und Ortungssystem nach Anspruch10, wobei die Verarbeitungsvorrichtung so pro-grammiert ist, um den Daten-Akquisitionsvorgangeine ausgewählte Anzahl von Malen zu wiederho-len, um ein statistisches Mittel einer Impulsaktivitätentlang des elektrischen Leiters in dem Referenz-puffer zu erhalten, wobei erste Reflexionen einesEntladungsimpulses von irgendeinem Fehler ent-lang des elektrischen Leiters zwischen dem Anfangund dem Ende des Referenzpuffers identifiziertwerden.

Revendications

1. Procédé pour localiser les pannes le long d'un con-

ducteur électrique (10) ayant une extrémité proxi-male (24) et une extrémité distale (22), comprenantles étapes consistant à :

appliquer un signal de tension à l'extrémitéproximale dudit conducteur électrique de façonà alimenter ledit conducteur électrique d'unemanière suffisante pour produire une déchargeau niveau d'une quelconque panne dans celui-ci, ledit conducteur électrique pouvant servir àpropager au moins une impulsion d'une plura-lité d'impulsions comprenant une impulsion pri-maire apparaissant au niveau d'une quelcon-que panne précitée en réponse à l'excitation duconducteur électrique par ledit signal de ten-sion et se déplaçant vers l'extrémité proximaledudit conducteur électrique, une première im-pulsion réfléchie correspondant à la réflexionde ladite impulsion primaire à l'extrémité distaledudit conducteur électrique, une impulsion dedeuxième ordre et une impulsion d'ordre supé-rieur correspondant à la réflexion de ladite pre-mière impulsion réfléchie à une extrémité cor-respondante parmi l'extrémité proximale et l'ex-trémité distale dudit conducteur électrique, etune impulsion d'interférence transitoire cou-plée audit conducteur électrique par l'intermé-diaire de l'environnement entourant ledit con-ducteur électrique ;mémoriser des données relatives à la longueurdudit conducteur électrique, lesdites donnéesétant choisies dans le groupe comprenant letemps de propagation pour qu'au moins uneimpulsion parmi ladite première impulsion et la-dite pluralité d'impulsions se déplace le longdudit conducteur électrique, la longueur duditconducteur électrique et la vitesse de ladite aumoins une impulsion parmi ladite impulsion pri-maire et ladite pluralité d'impulsions se dépla-çant le long dudit conducteur électrique ;établir un tampon de référence correspondantà la longueur dudit conducteur électrique en uti-lisant lesdites données ;initialiser ledit tampon de référence à zéro ;obtenir une pluralité d'échantillons à partir duditconducteur électrique, correspondant à l'ampli-tude de l'une quelconque de ladite pluralitéd'impulsions ;mémoriser ladite pluralité d'échantillons dansun tampon de travail ;parcourir ladite pluralité d'échantillons dans le-dit tampon de travail de façon à localiser unepremière de ladite pluralité d'impulsions ayantun niveau d'amplitude qui est supérieur à un ni-veau de signal prédéterminé ;mémoriser les échantillons choisis de laditepluralité d'échantillons à titre d'entrées respec-tives dans un tampon temporaire, lesdits

21 22

EP 1 153 308 B1

13

5

10

15

20

25

30

35

40

45

50

55

échantillons choisis comprenant des échan-tillons correspondant à ladite première de laditepluralité d'impulsions et aux échantillons sui-vants de ladite pluralité d'échantillons mémori-sés dans ledit tampon de travail, le nombre des-dits échantillons choisis correspondant à la lon-gueur dudit conducteur électrique ;normaliser lesdites entrées dans ledit tampontemporaire de telle manière que l'amplitude deladite première desdites impulsions est l'unité ;ajouter lesdites entrées dans ledit tampon tem-poraire aux entrées correspondantes dans ledittampon de référence ; etrépéter ladite étape d'obtention, ladite étape demémorisation destinée à mémoriser ladite plu-ralité d'échantillons, ladite étape de parcours,ladite étape de mémorisation destinée à mé-moriser lesdits échantillons choisis, ladite éta-pe de normalisation et ladite étape d'ajout, les-dites entrées dans ledit tampon de référencereprésentant une moyenne statistique desditesimpulsions pendant une période de temps cor-respondant à la quantité de temps nécessairepour qu'une desdites impulsions se déplace surla longueur dudit conducteur électrique.

2. Procédé selon la revendication 1, dans lequel ledittampon temporaire comprend des entrées desti-nées à mémoriser des données relatives à laditepluralité d'échantillons, ladite étape de mémorisa-tion destinée à mémoriser lesdits échantillons choi-sis dans ledit tampon temporaire comprenant l'éta-pe consistant à placer lesdits échantillons choisis,apparaissant à l'extrémité proximale et à l'extrémitédistale dudit conducteur électrique, dans ledit tam-pon temporaire à des entrées correspondantes desentrées situées au début et à la fin dudit tampontemporaire et le reste desdits échantillons choisisétant mémorisé en ordre successif en utilisant les-dites entrées entre la première et la dernière des-dites entrées.

3. Procédé selon la revendication 2, dans lequel laditeétape de placement comprend l'étape consistant àprévoir un nombre choisi d'entrées avant et aprèslesdites entrées situées au début et à la fin dudittampon temporaire, respectivement, afin qu'ellesfassent office de bandes de garde par rapportauxdits échantillons choisis.

4. Procédé selon la revendication 1, dans lequel laditeétape de répétition est exécutée un nombre prédé-terminé de fois de façon à définir une période d'ac-quisition de données choisie, et comprenant enoutre l'étape consistant à analyser ledit conducteurélectrique en utilisant lesdites entrées dans ledittampon de référence, une desdites entrées dans le-dit tampon de référence correspondant à la somme

statistique de l'ensemble desdites impulsions cap-turées par l'intermédiaire de chaque dite étape derépétition au cours de ladite période d'acquisitionde données.

5. Procédé selon la revendication 4, dans lequel laditeune desdites entrées correspondant à ladite som-me statistique est située au moins de manière proxi-male par rapport auxdites entrées situées au débutet à la fin dudit tampon de référence, ladite étaped'analyse comprenant l'étape consistant à analyserlesdites entrées dans ledit tampon de référencecomme représentant différentes impulsions de ladi-te pluralité d'impulsions au niveau de parties res-pectives sur la longueur dudit conducteur électri-que.

6. Procédé selon la revendication 1, comprenant enoutre les étapes consistant à :

identifier ladite première impulsion réfléchiecorrespondant à chaque dite panne sur leditconducteur électrique comme une entrée cor-respondante parmi lesdites entrées qui n'estpas proximale par rapport au début et à la findudit tampon de référence et a une amplitudechoisie ; etlocaliser chaque dite panne sur ledit conduc-teur électrique en corrélant la position desditesentrées correspondantes indiquant chaque ditepremière impulsion réfléchie par rapport à unpoint sur la longueur dudit conducteur électri-que.

7. Procédé selon la revendication 6, dans lequel leditconducteur électrique peut être soumis à une inter-férence de fréquences radio, comprenant en outrel'étape consistant à rejeter différentes impulsionsde ladite pluralité d'impulsions liées à l'une oul'autre cause parmi une impulsion d'interférencetransitoire et une interférence de fréquences radiopuisqu'elles sont indiquées au début et à la fin dudittampon de référence.

8. Procédé selon la revendication 6, dans lequel leditconducteur électrique peut être soumis à une inter-férence de fréquences radio, comprenant en outrel'étape consistant à identifier une interférence defréquences radio par rapport audit conducteur élec-trique comme une valeur décalée par rapportauxdits échantillons choisis dans ledit tampon deréférence.

9. Procédé selon la revendication 6, comprenant enoutre l'étape consistant à déterminer si une quel-conque impulsion parmi ladite impulsion de deuxiè-me ordre, ladite impulsion d'ordre supérieur et laditeimpulsion d'interférence transitoire est survenue en

23 24

EP 1 153 308 B1

14

5

10

15

20

25

30

35

40

45

50

55

utilisant lesdites entrées au début et à la fin dudittampon de référence.

10. Système de localisation de site de décharge partiel-le pour localiser les pannes sur la longueur d'unconducteur électrique comprenant :

un dispositif d'alimentation en courant adaptépour alimenter ledit conducteur électrique ;un système de mesure de décharge partielle(12) connecté audit conducteur électrique etpouvant servir à identifier des impulsions indi-viduelles apparaissant sur ledit conducteurélectrique, lesdites impulsions correspondant àune décharge au niveau de quelconques sitesde pannes le long dudit conducteur électriquedu fait d'une alimentation à l'aide dudit dispositifd'alimentation en courant, ledit système de dé-charge partielle comprenant un processeur(44), un dispositif de mémoire (46) et un con-vertisseur analogique-numérique (42), leditprocesseur étant programmé pour exécuterune opération de capture d'impulsions de façonà représenter lesdites impulsions sous la formed'échantillons numérisés d'amplitudes d'impul-sions provenant dudit convertisseur analogi-que-numérique, lesdits échantillons numérisésétant mémorisés dans ledit dispositif demémoire ; etun dispositif de traitement (48) programmépour effectuer au moins une parmi une pluralitéd'opérations comprenant une opération d'éta-lonnage, pour associer le temps de propaga-tion d'une desdites impulsions se déplaçant surtoute la longueur dudit conducteur électrique àla longueur dudit conducteur électrique, et uneopération d'acquisition de données pendant la-quelle sont obtenus lesdits échantillons numé-risés provenant dudit système de mesure d'im-pulsion, ledit dispositif de traitement étant pro-grammé en fonction de ladite opération d'ac-quisition de données afin de créer un tamponde référence correspondant à la longueur duditconducteur électrique, de mémoriser lesditséchantillons numérisés issus d'une opérationde capture d'impulsions précitée dans un tam-pon de travail, de parcourir ledit tampon de tra-vail de façon à localiser lesdits échantillons nu-mérisés correspondant à des crêtes d'impul-sions au-dessus d'un niveau de bruit choisi, demémoriser lesdits échantillons numérisés cor-respondant à celles-ci dans un tampon tempo-raire en même temps qu'un nombre choisi des-dits échantillons numérisés obtenus par la sui-te, de normaliser lesdits échantillons numéri-sés dans ledit tampon temporaire et d'ajouterlesdits échantillons numérisés contenus dansledit tampon temporaire audit tampon de réfé-

rence.

11. Système de localisation de site de décharge partiel-le selon la revendication 10, dans lequel ledit dis-positif de traitement est programmé pour répéter la-dite opération d'acquisition de données un nombrechoisi de fois afin d'obtenir une moyenne statistiquede l'activité d'impulsions le long dudit conducteurélectrique dans ledit tampon de référence, les pre-mières réflexions d'une impulsion de décharge àpartir d'une quelconque panne le long dudit conduc-teur électrique étant identifiées entre le début et lafin dudit tampon de référence.

25 26

EP 1 153 308 B1

15

EP 1 153 308 B1

16

EP 1 153 308 B1

17

EP 1 153 308 B1

18

EP 1 153 308 B1

19

EP 1 153 308 B1

20