Smart processing of data from permanent monitoring systems

Smart processing of data from permanent Smart processing of data from permanent monitoring systems: innovations and needsmonitoring systems: innovations and needs

(BRIMOS (BRIMOS –– EMASS)EMASS)

Herman Van Herman Van derder AuweraerAuweraer, Bart , Bart PeetersPeeters55--thth SAMCO Workshop, 26SAMCO Workshop, 26--27 Jan. 200427 Jan. 2004

5-th SAMCO WorkshopVienna, January 26-27, 2004

ContextBridge Monitoring

•• Permanent monitoring systemsPermanent monitoring systems•• LongLong--span cablespan cable--stayed and suspension bridgesstayed and suspension bridges

–– Large investmentLarge investment–– Motivated by insurance companiesMotivated by insurance companies

•• PurposePurpose–– Design verificationDesign verification–– Event recordingEvent recording

–– ArchiveArchive–– ServiceabilityServiceability

–– Health monitoringHealth monitoring•• MassiveMassive amounts of dataamounts of data

–– Smart processingSmart processing–– ManagementManagement


Operational Modal AnalysisVasco da Gama Bridge

z

x y

1D

1U

4U

4D

8D

8U9D

9U10D

10U11D

11U12,13,14U27U

27D12,13,14D

15D

15U16D

16U17D

17U18D

18U19D

19U 28D

28U

29D29U

20D

20U21D

21U22D

22U23D

23U24D

24U25D

25U

26U

26D

P1P2

P3PN

PS

P4

P5P6

(North)LISBOA

SETÚBAL(South)

7U

7D

2U

2D

3U

3D

5U

5D6D

6U


Ambient AccelerationsModes BV1 and T2

Vertical accelerationVertical accelerationTransversal accelerationTransversal acceleration

0.00 2.50Hz0.00

0.00

Logg2

0.00 300.00s-0.02

0.02

Rea

lg2

0.00 960.60s-0.52

0.59

Rea

lgTime

Corr.

Spec.


Øresund Bridge: Continuous Monitoring System Cable Forces

CM-5

00

CM-5

xxCM

-5xx

Powe

r CR4-31Galvanic

Separation

CR4-30PIC-

Board

CR4-20DSP-Board

CM-5

xxCM

-5xx

CR4-20DSP-Board

CR4-20DSP-Board

Computer

CM-5

xxCM

-5xx

CM-5

00

Sensor3-axis

Sensor1-axis

Sensor1-axis

Sensor1-axis


Future: Stonecutters Bridge in Hong Kong

•• Stonecutters Bridge (2007)Stonecutters Bridge (2007) •• WASHMS: wind and structural WASHMS: wind and structural health monitoring systemhealth monitoring system•• WindWind•• TemperatureTemperature•• AccelerationsAccelerations•• StrainsStrains•• TrafficTraffic loadingloading•• HHumidityumidity, rainfall, rainfall•• DDisplacementisplacement (by GPS)(by GPS)•• CCorrosionorrosion•• TOTAL: 1114 channels !!!TOTAL: 1114 channels !!!


From Real-time Displacements to FE Model validation


Challenges for structural monitoringand damage identification

•• Signal capturing and processingSignal capturing and processing•• SensorsSensors•• Sensor networksSensor networks•• Data transmissionData transmission•• Data processingData processing

•• Change detectionChange detection•• Signal based methodsSignal based methods•• Model based methodsModel based methods

–– Model comparisonModel comparison–– Statistical TestStatistical Test

•• Discrimination of environmental effectsDiscrimination of environmental effects•• Data managementData management



•• Smart SensorsSmart Sensors•• Use of MEMSUse of MEMS•• Optical sensorsOptical sensors•• Local digitizationLocal digitization•• Local processing (correction, calibration, preLocal processing (correction, calibration, pre--

processing, FFT, ….)processing, FFT, ….)•• Fusion of data types: vibration, displacement, Fusion of data types: vibration, displacement,

corrosion, GPS, video, environmental, static … corrosion, GPS, video, environmental, static … •• Gradual transition to advanced sensors (needGradual transition to advanced sensors (need--toto--

have basis) have basis) --> mixed mode support approach

© 2000, Stanford University© 2000, Stanford University

> mixed mode support approach


SensorsOptical Fibers

•• TechnologyTechnology•• Interferometry Interferometry using 2 fibersusing 2 fibers

–– Mechanically coupledMechanically coupled–– Free reference fiber Free reference fiber

(temperature)(temperature)•• Bragg gratingsBragg gratings

–– MultiplexingMultiplexing•• OtherOther

•• Can offer added valueCan offer added value•• Quality / stabilityQuality / stability•• LongLong--term reliabilityterm reliability•• InstallationInstallation

•• ParametersParameters•• Strain (static + dynamic)Strain (static + dynamic)•• TemperatureTemperature•• HumidityHumidity•• Corrosion

©© SmartecSmartec

©© SmartecSmartecCorrosion



•• BusBus--systems and wireless transmissionsystems and wireless transmission•• Sensor network architecturesSensor network architectures•• Data bus systems (e.g. CDMA, SPDIF,…)Data bus systems (e.g. CDMA, SPDIF,…)•• Wireless transmission (from analog to WLAN…)Wireless transmission (from analog to WLAN…)

•• Critical Issues: Critical Issues: •• Absolute time synchronization (through WLAN, through Absolute time synchronization (through WLAN, through

GPS, by cable, by predictive correction…)GPS, by cable, by predictive correction…)•• Data transmission performanceData transmission performance

–– Distance versus Bit rateDistance versus Bit rate–– Accuracy, resolution, dropAccuracy, resolution, drop--out…out…

•• Power consumption and supply (cable, battery, power Power consumption and supply (cable, battery, power “scavenging”) [IMEC]“scavenging”) [IMEC]

•• Research topic at LMSResearch topic at LMS



•• Breakthrough research in highBreakthrough research in high--value sectorsvalue sectors•• Medical: IMEC Human++ programMedical: IMEC Human++ program•• Space: NASA shuttle and ISSSpace: NASA shuttle and ISS•• Aeronautics: onAeronautics: on--board SHM, flutter detectionboard SHM, flutter detection•• Earthquake alert and analysis (NEEES, EEarthquake alert and analysis (NEEES, E--defense…)defense…)

•• Develop synergies, reDevelop synergies, re--use results where possibleuse results where possible•• Other EC projectsOther EC projects•• EUREKA (FLITE…)EUREKA (FLITE…)•• NationalNational•• InternationalInternational


IMEC Human++ Program


IMEC Human++ Program


IMEC Micro-sensor and Human++ Program






Example: Aerospace Wireless Communication

•• MicroTAUMicroTAU, wireless tri, wireless tri--axial axial accelerometeraccelerometer•• NASA Space ShuttleNASA Space Shuttle•• TTo detect vibration of payloads o detect vibration of payloads

during launch and landingduring launch and landing•• fsfs = 250 Hz= 250 Hz•• 9 min memory9 min memory

•• Wireless?Wireless?•• WWireless data synchronization ireless data synchronization

between units between units (±30(±30µµss @ 250Hz@ 250Hz))•• Wireless data downloadWireless data download

•• Other programs: Other programs: •• IWIS (synch. 300 ns) IWIS (synch. 300 ns) --> ISS> ISS•• WAIS (synch. 300 ns) WAIS (synch. 300 ns) --> Naval Air> Naval Air•• WATS (synch. (±10WATS (synch. (±10µµs) s) --> USAF

© INVOCON© INVOCON

> USAF


Sensor Data Collection Topologies

•• Sequoia Sequoia TriaxialTriaxial Acceleration Computer (Acceleration Computer (SeTACSeTAC) [Italy]) [Italy]•• Adding data processingAdding data processing capabilitiescapabilities

© © SequoiaSequoia


Wireless Communication

•• WiMMSWiMMS•• Wireless Modular Wireless Modular

Monitoring SystemMonitoring System•• The John A.The John A. BlumeBlume

Earthquake EngineeringEarthquake EngineeringCenterCenter, Stanford University, Stanford University

•• New technologiesNew technologies•• MEMS accelerometerMEMS accelerometer•• Embedded systemsEmbedded systems

–– Data acquisitionData acquisition–– Computational powerComputational power

•• Wireless communicationWireless communication–– ISM bandISM band

© Stanford University© Stanford University


Wireless Communication

•• PlugPlug--andand--play sensors in wireless networksplay sensors in wireless networks

© © CrossbowCrossbow


Generic Sensor Network

Communications

- wireless

- copper

- fibre

-?

Internet

Evaluation

electronicsFibre sensor

electronics

Node processor

electronics

Temperature

Humidity

pH

Wind speed

Corrosion

Strain gauge

Accelerometer

etc

“Site”

Node

Storage

Actuators ?

Some nodes have sensors, some don’t Presentation© © CSLCSL



•• Distributed data processingDistributed data processing•• Local PreLocal Pre--processing: FFT, features ... processing: FFT, features ... •• Data reduction/compression (However, for modal Data reduction/compression (However, for modal

analysis: at least crossanalysis: at least cross--powers needed powers needed -- references) references) •• Local/global task distribution Local/global task distribution --> > local modal analysis?local modal analysis?•• Data request Data request –– data provision services data provision services --> Internet, LAN…> Internet, LAN…•• Use of Agent Technologies (LMS)Use of Agent Technologies (LMS)•• Specific Architectures ResearchSpecific Architectures Research

–– Berkeley: TinyBerkeley: Tiny--OS, “Motes”OS, “Motes”–– KarlsruheKarlsruhe: Smart: Smart--IT / IT / Ubicomp Ubicomp –– K.U.L. / V.U.B. in BelgiumK.U.L. / V.U.B. in Belgium–– …………



•• Research ApproachResearch Approach•• Test individual solutions on existing HW platform (Test individual solutions on existing HW platform (cfrcfr. ex.). ex.)•• Then prototype integration in platform to be decidedThen prototype integration in platform to be decided

© © VCE VCE -- BRIMOSBRIMOS

© © LMSLMS © © GeoSIGGeoSIG © © Larson DavisLarson Davis



•• Change detection: current research at LMS: Change detection: current research at LMS: Model based methods: Model based methods: •• Modal analysis Modal analysis --> requires automation > requires automation •• Statistical test Statistical test --> requires good nominal model and test data > requires good nominal model and test data

[INRIA approach, flutter detection][INRIA approach, flutter detection]•• Discrimination of environmental effects: importance of Discrimination of environmental effects: importance of

influence assessed, but approach to deal with it??influence assessed, but approach to deal with it??

0.00 9.96Hz 2.91e-9

256e-9

Logg2

51.2e-9

71.6e-9

30.5e-9

109e-9

26.7e-9

464e-3 929e-3 1.36

1.86 2.29


Challenges for structural monitoringand damage identification: slat track

500 1000 1500 2000 2500 3000 350032.8

35.2

37.6

Number of fatigue cycles [-]

Freq

uenc

y[H

z]

500 1000 1500 2000 2500 3000 350032.8

35.2

37.6

Number of fatigue cycles [-]

Freq

uenc

y[H

z]

0 5 0 0 1 0 0 0 1 5 0 0 2 0 0 0 2 5 0 0 3 0 0 0 3 5 0 0 4 0 0 0-6 0

-4 0

-2 0

0

2 0

4 0

6 0

8 0

1 0 0

1 2 0

Nu m b e r o f fa tig ue c yc le s [-]

Cor

rect

ed s

tatis

tical

test

val

ue [-

]



•• Modal analysis automation research at LMSModal analysis automation research at LMS•• Robust and repeatable modal identificationRobust and repeatable modal identification•• Minimizing user interactionMinimizing user interaction•• Improved mathematical pole discriminationImproved mathematical pole discrimination•• I/OI/O PolyMAXPolyMAX method: discretemethod: discrete--frequency domain frequency domain

method with superior stabilization behaviour method with superior stabilization behaviour •• Now being researched for OMA applicationNow being researched for OMA application•• MultiMulti--patch data synthesispatch data synthesis•• Decision methodology:Decision methodology:

–– ClusteringClustering–– RuleRule--basedbased–– Neural NetworksNeural Networks


Polymax I/O: Porsche 911 Targa Carrera 4Fully Trimmed Car

LSCELSCE PolyMAXPolyMAX


Polymax I/O: Aircraft In-flight Testing

0 .0 0

1 .0 0

Log

( (m/s

2)/N

)

H z

- 1 8 0 .0 0

1 8 0 .0 0

Phas

e°

PolyMAXPolyMAXLSCELSCE


Stochastic Subspace Identification (BR)Z-24

OperationalOperationalCMIFCMIF


Operational PolyMAXZ-24

OperationalOperationalCMIFCMIF


Operational PolyMAXMulti-Patch Processing

2 3 4Reference sensorsReference sensors

1 5

•• Method 1: Independent processingMethod 1: Independent processing•• Merging shapes by computing LS modal scaling factors between Merging shapes by computing LS modal scaling factors between

reference sensorsreference sensors•• Averaging Averaging eigenfrequencieseigenfrequencies and dampingand damping

•• Method 2: Combined processing, scale shapes afterwardsMethod 2: Combined processing, scale shapes afterwards•• All half spectra are stacked (the reference sensors appear many All half spectra are stacked (the reference sensors appear many times)times)•• Only 1 set of modal parametersOnly 1 set of modal parameters

–– Rescale shapes as in method 1Rescale shapes as in method 1•• Method 3: Combined processing, scale data on beforehandMethod 3: Combined processing, scale data on beforehand

•• Scale half spectra based on averaged spectra between the referenScale half spectra based on averaged spectra between the referencesces•• No need to scale shapes after identificationNo need to scale shapes after identification


Z24-Bridge

Mode 1 Mode 2


Z24-BridgeMode 3

PolyMAXPolyMAX SSISSI


Z24-BridgeMode 4

SSISSIPolyMAXPolyMAX


Challenges for structural monitoringand damage identification: conclusions

•• Systems research on the level ofSystems research on the level of•• Smart sensors, heterogeneous sensorsSmart sensors, heterogeneous sensors•• Sensor networks, topologies, interfaces, busesSensor networks, topologies, interfaces, buses•• Wireless transmission & synchronizationWireless transmission & synchronization•• Power consumptionPower consumption

•• Analysis methods research:Analysis methods research:•• Various detection and localization methods (Test & FE)Various detection and localization methods (Test & FE)•• PolymaxPolymax method seen as key technology for automation of method seen as key technology for automation of

OMA and hence for monitoringOMA and hence for monitoring•• Methods for environmental effect discriminationMethods for environmental effect discrimination•• Methods for decisionMethods for decision•• Local/global processing (reduction & autonomy)Local/global processing (reduction & autonomy)

•• Leverage horizontal research results: Relevant input Leverage horizontal research results: Relevant input from Medical and from Medical and AeroSpaceAeroSpace research (IMEC, CSL, research (IMEC, CSL, INRIA, LMS...)INRIA, LMS...)


Thank you !Thank you !

Smart processing of data from permanent monitoring systems

Documents