Control and Signal Processing in a Structural Laboratory

Control and Signal Processing in a Structural Laboratory Authors: Weining Feng, University of Houston-Downtown, Houston, Houston, TX 7700 FengW@uhd.edu Alberto Gomez-Rivas, University of Houston-Downtown, Houston, TX 7700 Gomez-RivasA@uhd.edu George Pincus, University of Houston-Downtown, Houston, TX 7700 PincusG@uhd.edu Abstract Structural Analysis and Design (SAD) and Control and Instrumentation Electronics (CIE) are two University of Houston-Downtown Engineering Technology B.S. 4-years degree programs that are ideally related in the field of active control of civil structures. SAD deals with the analysis and design of structures, their loads, and failure modes. CIE deals with the design of systems for control of processes. The study of how to design active controls for structures that respond favorably to imposed loads and deformations is a problem that naturally connects the two disciplines. This paper describes how two seemingly different engineering technology majors can function and learn from each other in a synergistic environment for the benefit of students in both programs. Index Terms structures, controls, technology instruction, synergy, engineering technology INTRODUCTION Two engineering technology degree programs, Control Instrumentation Electronics Design and Structural Analysis and Design, in the Engineering Technology Department, University of Houston-Downtown, have started an initiative to develop interdisciplinary senior projects in structural monitoring and control. Using a structural beam testing bench and a real-time data acquisition system, senior students in both degree programs have the opportunities to work on a number of collaborative projects, for example, sensor characteristics and specifications, multi-sensor and data fusion, signal conditioning systems, measurement data analysis via Fourier Transformations, structural vibration analysis and monitoring, actuators for active structural control, and control systems design and implementation. An important part of the effort is to emphasize modern technology in engineering technology education. The paper presents a detailed account of the work done so far in this joint activity. Future plans are also discussed. The need for structural vibrations measurement/monitoring and control exists in a number of different environments. In space exploration, micro-gravity scientific experiment platforms in space stations have very stringent specifications on the vibration that is allowed. In all of these experiment platforms, some kind of vibration control system exists and the control systems used can be generally categorized into passive, semi-active, and active control systems [1]-[]. Similarly, in civil structures such as long suspended bridges and ultra high buildings, structural vibrations need to be closely monitored and contained for comfort and safety [3]-[4]. Among various issues related to structural monitoring and control, selection of the measurement and instrumentation system should be the first one to be addressed. The measurement unit is essential for any active control system. Furthermore, the quality of measurement unit has a critical effect on the quality of the overall control system. This paper describes the instrumentation system that has been used for the study of structural monitoring in the Engineering Technology (ET) Department, University of Houston-Downtown (UHD). Within the ET department of UHD, two ET programs, the established Structural Analysis and Design (SAD) Engineering Technology, and the relatively new Control and Instrumentation Electronics Design (CIED) program, have started working together in the interdisciplinary area of structural monitoring and control. This paper serves to provide an overview of the work carried out so far, and also identifies areas for further future development. STRUCTURAL MONITORING SYSTEM The first structural system studied consists of an aluminium flexible beam, excitation source, sensor, and a PC-based data acquisition system. 1

The testing bench: Currently, the structure that is being monitored is a thin aluminum beam, Figure 1, with a span of 8.0 ft simply supported at both ends. The theoretical value of the resonance frequency of the beam is 1.4 Hz. The modes of vibration can be calculated by the following formula based on a lumped-mass model. n π EI f n = (1) L m where n = Mode of vibration, n = 1,, 3, L =Beam span E =Modulus of elasticity I = Moment of inertia of the cross section m =Unit mass When n = 1, f 1 is the natural frequency of the beam. It is worth noting that the frequency of second mode is four times of natural frequency (f = 4 f 1 ), and the third mode frequency is 9 times of natural frequency (f 3 = 9 f 1 ). The above model was tested extensively by one of the authors in the experimental determination of vibration frequencies of curved beams [5]. Excitation source: The excitation to the beam is the electromagnetic force generated by a coil that has a fixed position in the test space, above the center of the beam and with no contact with the beam. The coil is directly connected to a signal generator typically used in an electronic laboratory. Due to the light weight of the beam, and the vibration being induced around the resonance frequency, the signal current from the signal generator is sufficient to energize the coil without any additional current amplification. The frequency of the sine wave is carefully adjusted and the amplitude of beam vibration is recorded. When the amplitude of vibration reaches its maximum, then the frequency of excitation signal coincides with the resonance frequency of the beam. Sensors: Two different types of vibration sensors have been tested. One is a piezo sensor (by Piezotronics, model 353B34, http://www.davidson.com.au/products/vibration/pcb/accels/353b34.asp) powered by a current source (model 480C0, Piezotronics, http://www.pcb.com/products/gep/gep480c0.html). According to the manufacturer s specifications, the piezo sensor has a sensitivity of 100.0 mv/g and a frequency response range of 1 to 4000 Hz. The other is a basic mechanical accelerometer (seismic sensor) with a mass-spring-damper internal construction. The mechanical sensor is used here for comparative study of sensor properties. The sensor is mounted at the middle of the beam. The data acquisition (DAQ) system: This consists of a National Instrument (NI) DAQ plug-in card PCI-604E, a terminal block with BNC connections. The software used is LabView/Virtual Bench developed by NI, and a PC, which hosts the DAQ card and the software. The DAQ card provides 8 channels of analog (differential) input signals and two channels of analog output signals. At the frequency range for structural monitoring, the DAQ system supports simultaneous data acquisition up to 8 sensors. This feature is useful for the study of multi-sensor data-fusion. LabView is a graphic programming language and offers the possibility to create sophisticated customer user interfaces. Virtual Bench is DAQ application software, which requires no programming effort except configuring some essential system settings. The software tool transforms the host PC together with the DAQ card into a digital oscilloscope with spectrum analyzer (FFT) and a huge data storage capability (PC hard disc). Assuming the specifications of the PC (in terms of CPU speed, memory size, etc.) being above average, then the bandwidth of the oscilloscope is only limited by the specification of the DAQ card used. MEASUREMENT DATA AND SIGNAL PROCESSING The screen capture in Figure illustrates the piezo sensor signal when the beam was vibrating at its resonance frequency, with both time-domain (upper trace) and frequency-domain (lower trace) responses. Based on the frequency analysis (FFT, lower trace), it is seen that the actual resonance frequency is around 1 Hz. Figure 3 displays the measurement signal obtained from the mechanical sensor. It is important to see the complementary nature of the information presented in time-domain and frequency-domain responses. It is apparent that, based on the time-domain response (upper trace in Figure ), the natural frequency of the beam is around 1 Hz since there are equivalently 1 cycles within one second in the upper trace. However, it is not so simple to determine the frequencies of other modes of vibration based on the time-domain response, although it is observed that there are some high frequency elements residing on the 1 Hz wave. The frequency domain response (lower trace in Figure ) confirms the natural frequency (the first peak). The harmonic frequencies also show up clearly in the frequency response, each about 1 Hz apart. The 3 rd mode of vibration at around 108 Hz is much pronounced, and this can be explained by the

location where the sensor is placed. Note that there is a bias about 0.86V in the time-domain signal; hence a large spike is seen at zero frequency in the frequency response. The screen capture of data acquisition from the mechanical sensor is shown in Figure 3. It is seen that the sensor picked up the natural frequency, but it failed to detect other modes of vibration, despite the fact that the sensor output has a much larger range (around ± 1.0V), i.e. the sensor has a small frequency response range compared with the piezo sensor. It is useful for students to see the comparison of the measurement results from different sensors to gain a better understanding of the different properties and specifications of sensors. CONCLUSIONS AND FUTURE PLANS The work carried out on the testing bench of Figure 1 has involved the joint effort of the faculty and students of both SAD and CIED programs of ET department. It has been an excellent learning experience for all participants of the project. Based on the results obtained so far, it is felt that further work can be continued on a number of areas: To introduce multiple sensors for distributed structural monitoring. Data-fusion techniques will be adopted in order to retrieve meaningful information from the multiple sensor data. Alternative sensors may be used that offer reasonable trade-offs between sensor cost and signal quality. Signal filtering can be introduced to enhance the quality of signal measurement. For active vibration control, it is necessary to carry out research for suitable actuating devices to effect the control action. More advance features of the DAQ system will be explored to investigate the feasibility of the DAQ system for structural control. It is likely that a faster and more sophisticated real-time control system has to be used in order to generate satisfactory performance of the control system. The above list of further research will provide more and wider learning opportunities to SAD and CIED students. It is expected that this kind of interdisciplinary effort will continue to be a successful strategy for growth of both programs. ACKNOWLEDGEMENT The Texas Higher Education Coordination Board (THECB) under Grant No. 00365-0096-001funded part of the work presented in this paper. The authors would also like to acknowledge the support and funding from the College of Science and Technology, UHD. REFERENCES [1] Grodsinsky, C. M., and Whorton, M. S., Survey of Active Vibration Isolation Systems for Microgravity Applications, J. of Spacecraft and Rockets, Vol. 37, No.5, September-October 000. [] Chandra, R. S., Grigoriadis, K. S., and Fialho, I. J., LPV Control Synthesis for Active Rack Isolation System, Institute for Space Systems Operations, 001 Annual Report. [3] Lee, G. C., Liang, G. C., and Tong, M.Z., Development of Variable Passive Control System for Structural Response Reduction, Proceedings of the Second World Conference on Structural Control, Kyoto, Japan, 1998. [4] Hansen, H. I., Thoft-Christensen, P., Mendes, P. A., and Branco, F. A., Wind-Tunnel Tests of a Bridge Model with Active Vibration Control, Structural Engineering International, Vol. 10, No. 4, 000 [5] Gomez-Rivas, A., Natural Frequencies of Transverse Vibration of Curved Beams, PhD dissertation, the University of Texas, Austin, Texas, 1968. 3

FIGURES AND TABLES FIGURE 1 TEST SETUP FIGURE TIME AND FREQUENCY RESPONSES FROM PIEZO SENSOR (SAMPLING FREQUENCY FS = 300 HZ) 4

FIGURE 3 DATA ACQUISITION FROM THE MECHANICAL SENSOR (SAMPLING FREQUENCY FS = 300 HZ) 5