Wireless Structural Sensing and Feedback Control with Embedded Computing

11

Wireless Structural Sensing and Feedback Wireless Structural Sensing and Feedback Control with Embedded ComputingControl with Embedded Computing

Yang Wang, Prof. Kincho H. Law

Department of Civil and Environmental Eng., Stanford Univ.

Andrew Swartz, Prof. Jerome P. Lynch

Department of Civil and Environmental Eng., Univ. of Michigan

Kung-Chun Lu, Prof. Chin-Hsiung Loh

Dept. of Civil Eng., National Taiwan Univ.

SPIE. San Diego, CA. Feb 27, 2006

22

OutlineOutline

Research BackgroundResearch Background

Hardware and Software Design of the Hardware and Software Design of the Wireless Structural Wireless Structural SensingSensing System System

Validation Tests: Geumdang Bridge, KoreaValidation Tests: Geumdang Bridge, KoreaLaboratory Steel Frame at NCREE, TaiwanLaboratory Steel Frame at NCREE, TaiwanGi-lu Bridge, TaiwanGi-lu Bridge, Taiwan

Hardware and Software Design of the Hardware and Software Design of the Wireless Feedback Wireless Feedback Structural ControlStructural Control System System

Half-scale Laboratory Steel Frame with MR Damper at Half-scale Laboratory Steel Frame with MR Damper at NCREE, TaiwanNCREE, Taiwan

33

From Wire-based Sensing to Wireless SensingFrom Wire-based Sensing to Wireless Sensing

DISADVANTAGES OF WIRED CHANNELS

E. G. Straser, and A. S. Kiremidjian (1998): Installation of wired system can take about 75% of testing time for large structures

M. Celebi (2002): Each sensor channel $5,000, half of the cost on installation (cabling, labor, etc.)

I. Solomon, J. Cunnane, P. Stevenson (2000): over 1000 sensors on Tsing Ma Bridge, Kap Shui Mun Bridge, and Ting Kau Bridge. 36 km of copper cable and 14 km of fiber optic cable. 1 year installation.

Traditional DAQ: Wire-based

Future Wireless DAQ System

Wireless SHM prototype system Jointly developed by researchers in Stanford Univ.

and the Univ. of Michigan

44

CHALLENGES

Limited power consumption

Restricted wireless communication range, bandwidth, reliability

Difficulty for data synchronization and real-time data delivery

SYSTEM DESIGN PRINCIPLES

Judicious hardware component selection

Simple, efficient, and robust software design

Challenges in Wireless Structural SensingChallenges in Wireless Structural Sensing

55

Sensor SignalDigitization

4-channel 16-bitAnalog-to-Digital

Converter ADS8341

Computational Core

WirelessCommunication

Wireless Transceiver:20kbps 2.4GHz

24XStream, or 40kbps900MHz 9XCite

StructuralSensors

128kB ExternalSRAM

CY62128B

8-bit Micro-controller

ATmega128

SPIPort

ParallelPort

UARTPort

Wireless Sensing Unit

Functional Diagram of Wireless Sensing UnitFunctional Diagram of Wireless Sensing Unit

66

Final Package of the Latest Prototype UnitFinal Package of the Latest Prototype Unit

Antenna Length: 5.79” (14.7cm)

Container Dimension4.02” x 2.56” x 1.57”(10.2 x 6.5 x 4.0 cm)

Total power consumption with MaxStream 9XCite modem

75 – 80 mA when active; 0.1 mA standby. (5 VDC)

Wireless communication MaxStream transceiver

9XCite: 90 m indoor, 300 m outdoor, 38.4 kbps

24XStream: 150 m indoor, 5 km outdoor, 19.2 kbps

Total unit cost using off-the-shelf components

$130 for small quantity assembly (2004)

77

Wireless Sensing NetworkWireless Sensing Network

Simple star topology network

Near-synchronized and reliable data collection from all wireless sensing units

Communication protocol design using state-machine concept

Firmware for wireless sensing units

Server-side computer software

88

Geumdang Bridge Test, KoreaGeumdang Bridge Test, Korea

Collaboration with Prof. Chung Bang Yun, Prof. Jin Hak Yi, and Mr. Chang Geun Lee, Korea Advanced Institute of Science and Technology (KAIST)

12.6 m

2.6 m

2o

SECTION A-A

ElastomericPad

AccelerometerLocation

99

Wire-based System vs. Wireless SystemWire-based System vs. Wireless System

Sensor PropertyPCB393 Piezoelectric(Cabled System)

PCB3801 MEMS Capacitive(Wireless System)

Maximum Range 0.5g 3g

Sensitivity 10 V/g 0.7 V/g

RMS Resolution (Noise Floor) 50 g 500 g

Minimal Excitation Voltage 18 VDC 5 VDC

Sampling Frequency 200Hz 200Hz / 70Hz

1010

Comparison Between Two SystemsComparison Between Two Systems

155 160 165 170 175 180 185 190-0.04

-0.02

0

0.02

0.04Wire-based DAQ, Sensor #13

Time(s)Acc

eler

atio

n (g

)

155 160 165 170 175 180 185 190-0.04

-0.02

0

0.02

0.04Wireless DAQ, Sensor #13

Time(s)Acc

eler

atio

n (g

)

0 5 10 150

1

2

FFT - Wire-based DAQ, Sensor #13

Frequency (Hz)Mag

nitu

de

0 5 10 150

0.2

0.4

0.6

0.8

FFT - Wireless DAQ, Sensor #13

Frequency (Hz)

Mag

nitu

de

Accelerometer Location

Abutment

Pier 5 Pier 6Pier 4

14

1 13

16

12

1817

2

19

3

26

4

25

5

24

6

15

7

23

8

22

9

21

10

20

11

36m 36m46m

N

1111

Sensor SignalDigitization

4-channel 16-bitAnalog-to-Digital

Converter ADS8341

Computational Core

WirelessCommunication

Wireless Transceiver:20kbps 2.4GHz

24XStream, or 40kbps900MHz 9XCite

StructuralSensors

128kB ExternalSRAM

CY62128B

8-bit Micro-controller

ATmega128

SPIPort

ParallelPort

UARTPort

Wireless Sensing Unit

Sensor SignalConditioning

Amplfication, Filtering,and Voltage-offsettng

Functional Diagram with Sensor Signal Conditioning ModuleFunctional Diagram with Sensor Signal Conditioning Module

1212

Latest Bridge Tests with Sensor Signal ConditioningLatest Bridge Tests with Sensor Signal Conditioning

Mean shifting: any analog signal to 2.5V mean

Amplification: 5, 10 or 20

Anti-alias filtering: band pass 0.02Hz – 25Hz

Sensor Allocation for Tests at Geumdang Bridge, Jul 2005

Abutment

Pier 5 Pier 6Pier 4

8 10 1211

1

13

2 3 4

9

5 6

14

7

38m 38m46m

Accelerometer Location

N

Printed circuit board of the signal conditioning module

(5.0 × 6.5 cm)

1313

0 5 10 150

2

4

FFT - Wire-based DAQ

Frequency (Hz)Mag

nitu

de

0 5 10 150

2

4

FFT - Wireless DAQ

Frequency (Hz)

Mag

nitu

de

15 20 25 30 35-0.04

-0.02

0

0.02Wire-based DAQ

Time (s)

Acc

eler

atio

n (g

)

15 20 25 30 35

-0.02

0

0.02Wireless DAQ

Time (s)Acc

eler

atio

n (g

)

Comparison for Wireless DAQ with Signal Conditioning Comparison for Wireless DAQ with Signal Conditioning

28 29 30 31 32 33-0.03

-0.02

-0.01

0

0.01

0.02Comaprison between Wire-based and Wireless DAQ

Time (s)

Acc

eler

atio

n (g

)

Wire-basedWireless

2.5 3 3.5 4 4.5 50

1

2

3

4

5Comparison between FFT to the Acceleration Data

Frequency (Hz)

Mag

nitu

de

Wire-basedWireless

Prof. J.P. Lynch, University of Michigan. Presentation at 10:50 am, Tue. Session 3, Room: Sunset.

1414

Collaboration with Prof. C. H. Loh, National Taiwan University &

National Center for Research on Earthquake Engineering (NCREE)

Laboratory 3-Story Structure on a 6-DOF Shaking TableLaboratory 3-Story Structure on a 6-DOF Shaking Table

WSU6

A3

A2A1

WSU5

A5

A4

WSU4

A6

A7

WSU1

A11

A10

A8

A12

A9WSU2

WSU3

S41 S42

S43S44

1515

Field Validation Tests at Gi-lu Bridge, Chi-chi, TaiwanField Validation Tests at Gi-lu Bridge, Chi-chi, Taiwan

Gi-lu Cable-Stayed Bridge, Chi-chi,

Taiwan

Span: 120m (L) + 120m (R)

Prof. C.-H. Loh, National Taiwan University. Presentation at 9:20 am, Tue. Session 10, Room: Towne

Vertical Dir.

Chi-chi side

Lu-Ku sideU1 U2 U3 U4 U5

U6 U7 U8 U9

1616

Sensor SignalDigitization

4-channel 16-bitAnalog-to-Digital

Converter ADS8341

Computational Core

WirelessCommunication

Wireless Transceiver:20kbps 2.4GHz

24XStream, or 40kbps900MHz 9XCite

StructuralSensors

128kB ExternalSRAM

CY62128B

8-bit Micro-controller

ATmega128

SPIPort

ParallelPort

UARTPort

Wireless Sensing Unit

Sensor SignalConditioning

Amplfication, Filtering,and Voltage-offsettng

Actuation SignalGeneration

16-bitDigital-to-Analog

Converter AD5542

SPIPort

StructuralActuators

Functional Diagram with Actuation Signal Generation ModuleFunctional Diagram with Actuation Signal Generation Module

1717

Integrated Switching Regulator PT5022

Command Signal Output

Digital Connections to ATmega128 Micro-

controller

Analog Connections to ATmega128 Micro-

controller

Digital-to-Analog Converter AD5542

Operational Amplifier LMC6484

Control Signal Generation ModuleControl Signal Generation Module

Supply voltage: 5 VDC

Output signal: -5 ~ 5 VDC

Output settling time: 1 s

Size: 5.5 x 6.0 cm

1818

20 kN Magneto-Rheological (MR) Damper

Wireless Feedback Structural Control TestsWireless Feedback Structural Control Tests

Collaboration with Prof. C. H. Loh, National Taiwan University &

National Center for Research on Earthquake Engineering (NCREE)

Floor: 3m x 2m

Floor weight: 6,000 kg

Inter-Story height: 3m

Shaking table: 5m x 5m

1919

S3

S2

S1

C1

Ci: wireless control unit

Si: wireless sensing unit

Ti: wireless transceiver

T1

Lab experimentcommand server

Actuator

Wireless Sensing and Control System OverviewWireless Sensing and Control System Overview

Time step length:

20 ms with 9XCite transceiver

80 ms with 24XStream transceiver

4. S1 sends out sensor data toC1; S2 and S3 back off a fewmilli-seconds respectively,and then send data to C1

Major Program Flow for a WiSSCon Laboratory Test

1. The server checks thewireless sensing and controlunits in the network throughthe wireless transceiver T1

2. The shaking tablestarts an earthquakerecord

3. C1 broadcasts a beaconsignal to all the units in thenetwork, announcing that anew time step begins

5. C1 analyzes all the sensordata, decides the controlsignal and applies the signalto the structural controller

6. The server overhears all thewi re les s c om m un i c a t i onthrough T1, and logs the datain the computer hard disk

Loop at each time step

2020

kuBkzAkz ddddd 1

kx

kxkz

d

dd

0 and 0 where

,1

0

RQ

kRukukQzkzkQzkzuJfk

kd

Tdd

Tdfd

Tfdd

kGzku dd

Embedded Computing (1)Embedded Computing (1)

Discretized Linear Quadratic Regulator (LQR) Control Algorithm:

Minimize index:

Optimal control force:

LQR ComputingReal-time sensor data

Desired control force

2121

5

1

0 1 1 2

3 21

2

( ) ( ) ( )

( ) ( 1) ( 1) ( 1)

( ) [ ( ), ( ) ( ) ( ), ( ) ( ) , ( ) ( ) ( ), ( ) ( ) ]

( ) (0.0083 ( ) 0.005) ( )

-13.2924V 22.9678V +1.0297 V - 1.0762

-161

d

i ii

T

d

F t F t z t

z k z k k k dt

k x k x k z k z k x k z k x k z k z k x k z k

F t V t x k

2

23

24

25

.6060V - 88.7154V - 389.2721

-5.0428V -169.2379V-160.4490

-0.6433V - 8.0282V- 0.7757

0.3452V - 6.775V- 0.316

Embedded Computing (2)Embedded Computing (2)

Modified Bouc-Wen MR damper model developed by researchers at NCREE:

MR Damper Model Computing

Desired control force

Appropriate actuation voltage to the MR damper

2222

0 0.002 0.004 0.006 0.008 0.01 0.0120

1

2

3

Drift (m)

Flo

or

Maximum Inter-story Drifts

Damper Volt = 0VDamper Volt = 1VWireless SystemWired System

Preliminary Structural Control TestsPreliminary Structural Control Tests

2323

Future Research in Wireless Sensing and ControlFuture Research in Wireless Sensing and Control

Collaboration with Prof. C. H. Loh, National Taiwan University &

National Center for Research on Earthquake Engineering

Explore more wireless communication technologies

Multiple MR dampers

Decentralized Control

2424

AcknowledgementAcknowledgement

Prof. Chung Bang Yun, Prof. Jin Hak Yi, and Mr. Chang Geun Lee, Korea Advanced Institute of Science and Technology (KAIST)

Prof. Anne Kiremidjian from Civil Engineering, and Prof. Ed Carryer from Mechanical Engineering at Stanford University

National Science Foundation CMS-9988909 and CMS-0421180

Office of Naval Research Young Investigator Program awarded to Prof. Lynch at University of Michigan.

The Office of Technology Licensing Stanford Graduate Fellowship

2525

The EndThe End