DISCUSSION ON THE APPLICATION OF WIRELESS ACTIVE SENSING UNIT

Transcription

1 4 th Worl Conference on Structural Control an Monitoring 4WCSCM-261 DISCUSSION ON HE APPLICAION OF WIRELESS ACIVE SENSING UNI FOR SRUCURAL CONROL C.H. Loh an K.C. Lu National aiwan University, aipei ,aiwan J.P. Lynch an Y. Wang University of Michigan, Ann Arbor, MI , USA Stanfor University, Stanfor, CA , USA P.Y. Lin National Center for Research on Earthquake Engineering, aiwan Abstract Performance aspects of the wireless active sensing unit, inclue the reliability of the wireless communication channel for realtime ata elivery, an its application to builing structural control is explo in this stuy. he application of the MR-amper to actively control a half-scale 3-story steel builing excite by shaking table is stuie using the wireless active sensors. With three MR ampers installe in each floors, structural responses uring seismic excitation are measu by wireless sensors an communicate to the MR amper s wireles active sensing unit where an H 2 controller has been implemente in the wireless sensing an control unit. he wireless active sensor is responsible for the reception of response ata, etermination of optimal control forces, an issuing of comman signals to the amper. Finally, the control performance is iscusse by using both the wi communication system an the wireless active sensing unit. Introuction raitional structural control technology has employe wire-base systems to o the communication for structural control an structural ata collection. However, the installation of these wire-base control systems can be expensive in labor, time an price. In orer to uce these monetary an time expenses for the installation of wire-base systems, new technologies in embee systems an wireless communication have been aopte in acaemic an inustry research for wireless sensing an monitoring. he use of wireless communication for SHM ata acquisition was illustrate by Strser an Kiremijian [1998]. More recently, Lynch et al. extene the work by embeing amage ientification algorithms into wireless sensing unit [Lynch et al. 2004] an has alreay proven reliable when use in lieu of coaxial wiring in structural monitoring systems [Wang et al. 2005; Lynch et al. 2006]. he avantages of using wireless sensing units for structural health monitoring have been verifie while in the structural control area, many challenges must still be explo in greater etail. o capitalize on low-cost semi-active actuators installe in high ensity in a single structure, wireless communication is propose to minimize the high-cost of coaxial wires. A prototype Wireless Structural Sensing an Control (WiSSCon) system has been propose [Lynch et al. 2006] for structure response mitigation. he software written to operate the wireless sensors uner the real-time requirement of the control problem is presente in this stuy. he promising performance in applying wireless communication an embee computing technology into a real-time feeback structural control system was presente. his paper presents the experimental verification of using both fully centralize control an fully ecentralize control strategies with the implementation of semi-active control evices an aopting wireless control an sensing system. wo major research irections are emphasize: (1) Develop the theoretical basis of fully centralize an fully Loh, Lu, Lynch, Wang, an Lin 1

2 ecentralize control algorithms an implement to WiSSCon system for structural control, (2) Experimental verification on the wireless communication for structural control is mae. Comparison on both the wi control system an the wireless control system is examine. he Experimental Setup of WiSSCon System WiSSCon (Wireless Structural Sensing an Control System) is a prototype system esigne for real-time wireless structural sensing an feeback control [Wang et al an Lynch et al. 2006]. In the WiSSCon system, wireless communication is use for the feeback of structural response ata to wireless sensors serving as the control kernel (i.e. to calculate control solutions base on receive ata). It consists of a wireless sensing unit with the embee control software in the microcontroller an the action boar. Fig. 1 shows the basic architecture of the wireless sensing an control unit. he wireless sensing unit is responsible for measuring the ynamic response of the structure an the action boar is to convert the igital signal (8 bit) to analog signal (16 bit) for control purpose. For the calculation of control forces at each time-step, the wireless sensing unit was also esignate as the control kernel (terme the wireless control unit) utilizes its local embee computing resources to quickly process sensor ata, generate control signals, an apply control commans to structural actuators within the esignate time-step uration. On each floor a WiSSCon system is installe an connecte to the MR-amper via VCCS (voltage to current converter). he operation of WiSSCon system an the program flow can be explaine from server sie an the embee coe sie which is summarize into five steps: Step 1: Boot up the system from both PC server an embee system in the WiSSCon system; Step 2: he server checks all the wireless sensing an control units in the network through the wireless transceiver (as shown in Fig. 2a for etail escription of step 2) an reset clock/counter the synchronization process is verifie; Step 3: From the embee coe, each wireless control unit broacasts a beacon to all other units in the network sequentially announcing that a new time step begins. he communication latency of each wireless transmission nees to be carefully consie. For the centralize control it is set 20 millisecons for each sensing unit to receive the signal from other unit in this stuy (as shown in Fig. 2b). For case of centralize control a total of 60 ms is neee for all units receive all the signals from other units. hen to calculate the control force which 35 millisecons may nee. So a total of 95 millisecons is requi to conuct the centralize active control. his is the reason to set the sampling rate of 10Hz for centralize control; Step 4 an Step 5: Feeback the results to PC server an exit the control program. he builing control system inclues: Sensors (velocity meters an accelerometers), MR-ampers, VCCS (voltage converts to current system), an WiSSCon system, as shown in Fig. 3. Because two ifferent 0 ~ 5V Analog Sensors Sensing Interface Sensor Signal Digitization 4-channel 16-bit Analog-to-Digital 4-channel 16-bit Analog-to-Digital Converter ADS8341 Converter ADS8341 SPI Port Control Interface Computational Core 128kB External SRAM CY62128B Parallel Port 8-bit Micro-controller Amega128 SPI Port 16-bit Digital-to-Analog 16-bit Digital-to-Analog Converter AD5542 Converter AD5542 UAR Port Control Signal Generation Wireless Communication 20kbps 24XStream Wireless Moule with a 2.4GHz ransceiver Structural Actuators that Accepts -5V ~ 5V Comman Signal Figure 1: Har ware architecture of wireless control unit (Lynch et al. 2005). Loh, Lu, Lynch, Wang, an Lin 2

3 (a) (b) Figure 2: (a) Synchronization of ata communication an check start from sever coe in Step 2, (b) Data broacast an time sharing among sensing units for centralize control in Step 3. control strategies are use in this stuy, the setup of the control system is also ifferent for using ifferent control strategy. For the e-centralize control three iniviual sub-systems are consie an each sub-system has its own control evice. he velocity meters are installe at both sies of the amper to measure the relative isplacement between two ajacent floors (using wi system). his signal will fee into the WiSSCon system to calculate the control force. he floor acceleration ata from each particular floor (or sub-system) is collecte wirelessly by the WiSSCon system an Kalman estimator will be use to estimate the full state which incorporate with the measu floor relative velocity to calculate the control voltage. Base upon a measure of the shaft velocity of the amper an the estimate control force from WiSSCon system, esi commen voltage for control can be achieve. hrough VCCS (Voltage-Current Converter system) the control current will be estimate an fee into the amper for control purpose. As for the centralize control each WiSSCon system will collect all the floor acceleration responses wirelessly instea of just single floor response. Both centralize an e-centralize control algorithms will be introuce in the following section. Formulation he Control Algorithm Sensing unit Accelerometer (i+1)-th floor Velocity meter Power Supply (24 Volt) i-th floor VCCS (convert voltage to current : 0~2amp Amp) Recor Relative velocity (i-th floor) WiSSCon Unit Calculate Control Force (MR-Moel) Convert Force to Voltage Broacast Floor Acceleration Action Boar (16 bit D/A converter) 0.0 V ~ 1.2 V Figure.3: Control setup using wireless sensing an control unit an its connection with the MR-amper. Loh, Lu, Lynch, Wang, an Lin 3

4 wo ifferent control algorithms are use for the control experiment: the fully centralize control an fully ecentralize control. In eveloping the fully centralize control algorithm the entire structural system an all measurements are consie to generate the control gain together. Since the full-state measurement is impossible an can not be implemente in the realistic structural control, therefore, the Kalman estimator, which transforms the measu vector into the full-state vector, will be use to estimate the full-state responses of the structural system an to calculate the control gain. Consiering the limite sensors in the structure, the Kalman estimator nees to be selecte to transform the limite measurements into the full-state measurement estimation. he H 2 control algorithm will be use. First the objective function, incluing the system matrix an the term of control evices, efine as: [ ] 66 [ ] [ ] 33 [ ] (1) k 0 J z k Q z k u k R u k Here, the weighting matrices, Q an R, are relate to the full state an the control forces, respectively. After transforming the function by the proceures of variation metho, the new form of the function can imply the Ricatti equation an be solve as A P A A P B (2 R B P B ) B P A 2 Q P (2) in which the matrix P is the solution of the Ricatti equation, obtaine by iterative calculation. Consiering the limite sensors in the structure, the Kalman estimator nee to be selecte to transform the limite measurements into the full-state measurement estimation, as shown below: z [ k 1] A z [ k ] B u[ k ] L ( y [ k ] C z [ k ] F u[ k ]) (3), s Here, L is the solution of the Ricatti equation from the Kalman formulation [Dyke et al. 1996]. he control force can then be replace by the following form u[ k 1] Gz [ k 1] G ( A GAˆ z[ˆ GL cs ms y[ B G LC c LF G) z [ GLy where Âcs is the moifie system matrix in relating to control. Finally, Eq. (7) provies the estimate fullstate response ( z ) to estimate the control forces in the real response. Because the control force was generate using the full-state response vector, it is calle the fully centralize control. he fully ecentralize control emphasizes on the control of local system aroun the location of control evice by using the response measurements from sensors aroun the local sub-system so as to generate the control force instantly an efficiently. From this point of view, a complete structural system can be separate into many sub-systems an each sub-system contains its own sensor measurements an control evices. n this stuy, consier three inepenent sub-systems in the structure, the state-space equation an the measurement equations become: [ (4) 3 z[ k 1] A z[ k ] ( B ) u [ k ] E x [ k ] i i g i 1 y [ k ] ( C ) z[ k ] ( D ) u [ k ] ( D ) u [ k ] ( F ) x [ k ] 1,1 16, ,1 j 11 j,1 11 g j 2 y [ k ] ( C ) z[ k ] ( D ) u [ k ] ( D ) u [ k ] ( F ) x [ k ] y 2,3 16, ,2 j 11 j,2 11 g j 1,3 3 ( 3 [ ( C,3 ) 16 z[ ( D,33) 11 u1[ D,3 j ) 11 u j[ ( F,3 ) 11 xg[ j1,2 (5) Loh, Lu, Lynch, Wang, an Lin 4

5 Here, R is not a matrix any more an may vary with ifferent control forces because every objective function focuses on one control force of its own sub-system. Since there are three control evices an three acceleration measurements at each floor as feeback, therefore only three subsystems for this ecentralize control metho are involve an the objective functions are shown below: k 0 J z [ k ] Q z[ k ] u [ k ] R u [ k ] k 0 J z [ k ] Q z[ k ] u [ k ] R u [ k ] k 0 J z [ k ] Q z[ k ] u [ k ] R u [ k ] Here, each objective function focuses on each control force so that the number of control gains is also three. he Ricatti equation, which is erive from each objective function, can be presente he control gain can be escribe as: A PA A P( B ) (2 R B P( B ) ) ( B ) PA 2 Q P (7) 1, i 61 11, i 61, i 61 u [ (2 R B PB ) B PAz[ G z[ (8) 1 i, i, i, i 16 hen the Kalman estimator nees to be moifie as: z [ k 1] A z [ ( B ) u [ L (( y ) [ ( C ) z [ ( F ) u [ ) i i j, s j j j, i i It means that base on the measurement from iniviual sub-system the estimation of full-state is requi. where i inicates the i-th estimator which is with the same length of the full-state vector. An the control force can be obtaine from i i i u [ k 1] G ( z ) [ k 1] (10) (6) (9) Analysis of he Experimental Results From the shaking table test of the 3-story steel frame with the implementation of MR-amper in each floor, the control effectiveness using both wi an wireless control systems are examine. he MRamper is installe in the mile of the floor system with V-type bracing system. Verification on the communication of control comments is stuie first, an then the control effectiveness using ifferent control strategies is iscusse. In this control experiment three ifferent communication systems are use: (1) NCREE ata acquisition system (Pacific Series 5500 Digital Conitioning System with sampling rate of 200Hz)with Simulink for control, (2) WiSSCon Centralize control system (with sampling rate of 10Hz), (3) WiSSCon De-centralize control system (with sampling rate of 50Hz). Detail escription of the three system is escribe as follows: (a). NCREE Laboratory Control System: Combine the NCREE igital conitioning system, Simulink an the DSpace I/O boar, the traitional laboratory control test using wi system will be use an the results will be use as the benchmark moel for comparison. he sampling rate for this control experiment (either centralize or ecentralize control) is 200Hz an H2 control algorithm will be use. he result from this control system will be serve as a benchmark for comparison with other wireless communication control system. (b). Centralize control with sampling rate of 10 Hz: In the wireless centralize control the sensing unit on each floor will collect all the acceleration ata broacaste from all other sensing units on ifferent floor. If high sampling rate is chosen then ata will loss uring the communication, so a sampling rate of 10 Hz is use in this case. (3). De-centralize control with sampling rate of 50 Hz: In the ecentralize control, each wireless sensing an control unit only receives signal from sensor in each specific subsystem. here is no nee to wait the signals transmitte from other Loh, Lu, Lynch, Wang, an Lin 5

6 sensing units, so higher sampling rate of 50Hz can be use for ecentralize control. In orer to verify the accuracy of the communication signal uring the control process comparison on the recore signal from ifferent stage will be collecte for comparison. Data collecte using WiSSCon system an NCREEwi system will be iscusse so as to verify the controlle capability of the wireless control system. Valiation of the WiSSCon system Comparison on the commen voltage nee to be verifie: one is signal generate from the microcontroller (calculate from embee program) of the wireless sensing unit in igital format an the other one is generate from the action boar in analog form. Comparison on the recore acceleration ata either irectly from sensor or broacaste from other sensing units is also verifie. Comparison on the commen voltage collecte from action boar an from numerical simulation is mae for both the centralize an e-centralize control. Comparison between the calculate amper force from the numerical theoretical moel (soli line) an the estimate amper force from WiSSCon wireless system (ash line) is also compa, as shown in Figure 4. From all these valiation stuy on WiSSCon system, it is prove that: the commen voltage is correct, there is no loss of ata uring the communication, an the action boar can generate the control voltage correctly. (a) (b) (c) Figure 4: (a) Example shows the measu relative velocity (1 st floor) of MR-amper, (b) Comparison between the calculate amper force between the theoretical moel (soli line) an the estimate amper force from wireless system (ash line), (c) Decie voltage to be fee into the MR-amper. Valiation of wireless control effectiveness Performance evaluation of MR-amper uring the control experiment plays an important rule on control effectiveness. In centralize control the amper s hysteretic behavior is examine by using both wireless system an wi system (NCREE system) uner the same sampling rate of 10Hz is compa, as shown in Figure 5a for centralize control an Figure 5b shows the comparison for ecentralize control with sampling rate of 50 Hz. o verify the control effectiveness using wireless sensing an control system the control results are compa with the wi control system with sampling rate of 200Hz. Comparison on the floor acceleration an isplacement using wi control system (200Hz) an wireless control system for case of centralize control (10Hz) an ecentralize control (50Hz) is also conucte. It is prove that the wireless sensing an control system can almost reach the same control effectiveness as the wi system. Comparison on the control effectiveness among ifferent control methos is also mae. Figure 6a shows the comparison among un-control case, numerical simulation of fully centralize control an wireless centralize control with 10 Hz, an Figure 6b is the comparison with the ecentralize control using wireless with 50 Hz. he results inicate that Loh, Lu, Lynch, Wang, an Lin 6

7 both centralize an ecentralize control using wireless sensing system can reach almost the same control effectiveness as the wire control system. (a) (b) Figure 5: Comparison on force-stroke an force-velocity relationship of MR-amper using both wi an wireless control system; (a) for centralize control with sampling rate of 10Hz, (b) for ecentralize control with sampling rate of 50 Hz. Conclusions his stuy examines the potential of wireless communication an embee computing technologies for real-time structural control application. Base on the implementation of the prototype WiSSCon system in the structure, both the centralize control metho an ecentralize control metho are applie to control a 3-story steel structure with MR-amper in each floor subject to EL-Centro earthquake excitation on shaking table. he following conclusions are rawn: 1. he WiSSCon concept has been emonstrate in this stuy, the performance of this system is shown to be superior to the case of semi-active control system an nearly comparable to the wi control system. he control effectiveness using either centralize or ecentralize control shows reliable performance of the system. 2. he latency limitation associate with wireless communication nees to be carefully examine. Different control algorithm to be embee in the WiSSCon system may have ifferent latency which has to be figure out in orer not to loss the ata in communication. 3. In the ecentralize control esign, only local sensor information has been use to generate the control signal that is sen to the ampers of each control subsystem. his e-centralize control algorithm can be carrie out successfully for a large-scale structural system. he simulation results emonstrate that ecentralize control provies almost the same control effectiveness (inices) as the centralize control.an avantage of the ecentralize control is the robustness in the face of failure of control system uring strong earthquake. Given one subsystem fails, the other subsystems can be capable of compensating accoringly an ensure suitable global performance of the system. 4. Finally, the wireless sensor networks are a promising technology capable of operation for a real-time communication an control system. he wireless feeback structural control with embee computing can uce the cost by eraicating cables in the control system, using low-cost Loh, Lu, Lynch, Wang, an Lin 7

8 microcontroller for the control kernel an highly flexible an aaptable system configuration because of wireless communication. Figure 6: Comparison on the acceleration responses in each floor for un-control case, numerical simulation an experimental results where experimental result is: (a) for centralize control, (b) for ecentralize control.. Acknowlegements his research is support by both National Science Council uner grant No. NSC Z an Central Weather Bureau. he authors woul also like to express their gratitue to NCREE technicians for their assistance when conucting the shaking table experiments. References Dyke, S.J., Spencer, B.F., Sain, M.K. an Carlson, J. D., Moeling an control of magnetorheological ampers for seismic response uction, Smart Materials an Structures, 5, 1996, Dyke, S.J., Spencer, B.F., Quast, P. Sain, M.K., Kaspaari, D.C. an Soong,.., Acceleration feeback control of MDOF structures, ASCE, J. of Engineering Mechanics, 122(9), 1996, Lin, P.Y., P.N. Roschke, C. H. Loh an C. P. Cheng, Semi-Active controlle base-isolation system with MR ampers an penulum system,, 13WCEE, Vancouver, August, 2004, paper #691. Lynch, J. P., Sunararajan, A., Law, K. H., Kiremijian, A. S., an Caryer, E., Embee amage etection algorithms in a Wireles Sensing Unit for Atainment of Operational Power eficiency, Smart Materials an Structures, IOP, Vol.13, No.4, 2004, Lynch, J. P., Wang, Y., Lu, K. C., Hou,. C., an Loh, C. H., Post-seismic Damage Assessment of Steel Structures Instrumente with Self-interogating wireles Sensors, Proceeings of the 8 th National Conference on Earthquake Engineering, San Francisco, CA, USA, April 18-22, Lynch, J. P., Wang, Y., Swartz, R. A., Lu, K. C. an Loh, C. H., Implementation of a close-llop structural control system using wireles sensor networks, submit to J. of Structural Control an Health Monitoring (2006). Renzi, B. an Serino, G, esting an moeling a semi-actively controlle steel frame structure equippe with MR ampers, Structural Control an Health Monitoring, 2004 Spencer, B.F., Dyke, S.J., Sain, M.K. an Carlson, J. D., Phenomenological moel for magnetorheological ampers, J. Engineering Mechanics, ASCE, 123, 1997, Straser, E. G. an Kiremijian, A. S., A moular, wireles amage monitoring system for structures, Report No.129, John A. Blume Earthquake Engineering Research Center, Department of Civil & Environmental Engineering, Stanfor University, CA Wang, Y., Lynch, L. P., Law, K. H., Design of a low-power wireless structural monitoring system for collaborative computational algorithms, Proceeings of SPIE 10 th annual Int. Symposium on Nonestructive Evaluation for Health Monitoring an Diagnostics, San Diego, CA, March 6-10, Loh, Lu, Lynch, Wang, an Lin 8