Shieh-Kung Huang

Structural damage diagnosis based on on-line recursive stochastic subspace identification

This paper presents a recursive stochastic subspace identification (RSSI) technique for on-line and almost real-time structural damage diagnosis using output-only measurements. Through RSSI the time-varying natural frequencies of a system can be identified. To reduce the computation time in conducting LQ decomposition in RSSI, the Givens rotation as well as the matrix operation appending a new data set are derived. The relationship between the size of the Hankel matrix and the data length in each shifting moving window is examined so as to extract the time-varying features of the system without loss of generality and to establish on-line and almost real-time system identification. The result from the RSSI technique can also be applied to structural damage diagnosis. Off-line data-driven stochastic subspace identification was used first to establish the system matrix from the measurements of an undamaged (reference) case. Then the RSSI technique incorporating a Kalman estimator is used to extract the dynamic characteristics of the system through continuous monitoring data. The predicted residual error is defined as a damage feature and through the outlier statistics provides an indicator of damage. Verification of the proposed identification algorithm by using the bridge scouring test data and white noise response data of a reinforced concrete frame structure is conducted.

Development of a portable electrical impedance tomography data acquisition system for near-real-time spatial sensing

The main goal of this study was to develop and validate the performance of a miniature and portable data acquisition (DAQ) system designed for interrogating carbon nanotube (CNT)-based thin films for real-time spatial structural sensing and damage detection. Previous research demonstrated that the electrical properties of CNT-based thin film strain sensors were linearly correlated with applied strains. When coupled with an electrical impedance tomography (EIT) algorithm, the detection and localization of damage was possible. In short, EIT required that the film or “sensing skin” be interrogated along its boundaries. Electrical current was injected across a pair of boundary electrodes, and voltage was simultaneously recorded along the remaining electrode pairs. This was performed multiple times to obtain a large dataset needed for solving the EIT spatial conductivity mapping inverse problem. However, one of the main limitations of this technique was the large amount of time required for data acquisition. In order to facilitate the adoption of this technology and for field implementation purposes, a miniature DAQ that could interrogate these CNT-based sensing skins at high sampling rates was designed and tested. The prototype DAQ featured a Howland current source that could generate stable and controlled direct current. Measurement of boundary electrode voltages and the switching of the input, output, and measurement channels were achieved using multiplexer units. The DAQ prototype was fabricated on a two-layer printed circuit board, and it was designed for integration with a prototype wireless sensing system, which is the next phase of this research.

A damage detection algorithm integrated with a wireless sensing system

In this study, the authors propose a frequency response function change method (FRFCM) which can be integrated with a wireless sensing system to detect damage of a building structure. The FRFCM was derived based on motion equations under a ground excitation both before and after a structure is damaged. The advantage of FRFCM is that only the frequency response functions of some frequency ranges around natural frequencies of a structure are needed to detect the location and extent of a damage. On the other hand, the wireless sensing units have the calculation ability to transform the measured time series to the frequency spectrum using the fast Fourier transform (FFT) algorithm. Therefore, only a few frequency bands of the frequency spectrum in the wireless sensing units are necessary to be delivered to the wireless server, instead of the whole measured time series. By doing so, the transmit power consumption of a wireless sensing unit is greatly reduced, hence increasing the feasibility of on-line damage detection using wireless sensing system based on structural vibration signals. The proposed idea was validated in a shaking table test of a 6-story steel building structure in a laboratory. In order to detect damage on-line automatically via a wireless sensing system, a FFT algorithm and a automatic peak-peaking algorithm for selecting natural frequencies of a structure were imbedded into the wireless sensing units. The damage extent of each story of the structure was displayed on the screen of the host computer automatically after the transmit of fragments of Fourier spectrum from wireless sensing units was done.

Control of equipment isolation system using wavelet-based decentralized sliding mode control

Critical non-structural equipment, including high-precision equipment in the technology facilities, life-saving equipment in the hospitals, data storage equipment in the communication, computer, and data centers, etc., is vulnerable to vibration, and on top of that, the failure of these vibration-sensitive equipment will cause severe economic loss. In recent years, a lot of research has been conducted towards evaluation of semi-active control strategy for earthquake protection of vibration-sensitive equipment. Various innovative control algorithms have been studied to compensate the inertial loading, and these new or improved control strategies, such as the control algorithms based on the linear-quadratic regulator (LQR) and the sliding mode control (SMC), are also developed as a key element in smart structure technology. However, except to the advantage of simplicity and variability, both LQR and SMC are (centralized) full-state feedback controller and the state vector needs to be presented through a state estimator or compensator if it is not measured during earthquake excitation. On the other hand, considering decentralized control strategy, the controller is only evaluated using the response in the vicinity of control devices, thus minimizing the wiring and sensor communication requirements. However, the very limited information obtained and used by decentralized control strategy also restrains the control capability from reducing the response where without installing sensors. The aim of this paper is to develop a decentralized control algorithm on the control of both structure and non-structural equipment simultaneously to overcome the limitations of decentralized control through combining the advantage of wavelet analysis. This wavelet-based decentralized control algorithm will be simulated based on a frame with an equipment isolation system and a magnetorheological (MR) dampers located on the top of first floor numerically. The performance and robustness of wavelet-based decentralized control algorithm as well as the response of primary structure are evaluated and discussed through simulation study to demonstrate the efficiency of proposed control algorithms.

Extracting ground motion characteristics of distant earthquakes for mitigating displacement-sensitive equipment

Ground motions induced by strong, distant earthquakes may contain extremely long-period seismic waves that have a dominant period of up to 10 s. In general, these long-period seismic waves have small amplitudes and do not endanger the safety of building structures and civil infrastructures. However, they may bring unexpected shutdown of some vibration (displacement)-sensitive equipment (such as the wafer scanners in high-tech fabs), which can cause production loss. In this study, seismic waveforms collected from broadband seismometers distributed in Taiwan were used to investigate the ground motion characteristics of the collected distant earthquakes (with an epicenter distance of over 1000 km). The time-variant dominant frequency was extracted using moving window wavelet packet transform to monitor significant long-period seismic waves from the preevent data of each seismic event. The slope of the Arias intensity and the slope index of the recorded seismic waves were also developed to detect the potential accumulation of vibration energy increasing with respect to time and amplitude. The proposed index was used to detect the features of significant distant earthquakes, and it provides a mechanism to prevent unexpected shutdown of displacement-sensitive equipment. Finally, the proposed approach is discussed in relation to the damage severity of high-tech fabs.

Control of equipment isolation system using wavelet-based hybrid sliding mode control

Critical non-structural equipment, including life-saving equipment in hospitals, circuit breakers, computers, high technology instrumentations, etc., is vulnerable to strong earthquakes, and on top of that, the failure of the vibration-sensitive equipment will cause severe economic loss. In order to protect vibration-sensitive equipment or machinery against strong earthquakes, various innovative control algorithms are developed to compensate the internal forces that to be applied. These new or improved control strategies, such as the control algorithms based on optimal control theory and sliding mode control (SMC), are also developed for structures engineering as a key element in smart structure technology. The optimal control theory, one of the most common methodologies in feedback control, finds control forces through achieving a certain optimal criterion by minimizing a cost function. For example, the linear-quadratic regulator (LQR) was the most popular control algorithm over the past three decades, and a number of modifications have been proposed to increase the efficiency of classical LQR algorithm. However, except to the advantage of simplicity and ease of implementation, LQR are susceptible to parameter uncertainty and modeling error due to complex nature of civil structures. Different from LQR control, a robust and easy to be implemented control algorithm, SMC has also been studied. SMC is a nonlinear control methodology that forces the structural system to slide along surfaces or boundaries; hence this control algorithm is naturally robust with respect to parametric uncertainties of a structure. Early attempts at protecting vibration-sensitive equipment were based on the use of existing control algorithms as described above. However, in recent years, researchers have tried to renew the existing control algorithms or developing a new control algorithm to adapt the complex nature of civil structures which include the control of both structures and non-structural components. The aim of this paper is to develop a hybrid control algorithm on the control of both structures and equipments simultaneously to overcome the limitations of classical feedback control through combining the advantage of classic LQR and SMC. To suppress vibrations with the frequency contents of strong earthquakes differing from the natural frequencies of civil structures, the hybrid control algorithms integrated with the wavelet-base vibration control algorithm is developed. The performance of classical, hybrid, and wavelet-based hybrid control algorithms as well as the responses of structure and non-structural components are evaluated and discussed through numerical simulation in this study.

An integrated earthquake early warning system and its performance at schools in Taiwan

An earthquake early warning (EEW) system with integration of regional and onsite approaches was installed at nine demonstration stations in several districts of Taiwan for taking advantages of both approaches. The system performance was evaluated by a 3-year experiment at schools, which experienced five major earthquakes during this period. The blind zone of warning was effectively reduced by the integrated EEW system. The predicted intensities from EEW demonstration stations showed acceptable accuracy compared to field observations. The operation experience from an earthquake event proved that students could calmly carry out correct action before the seismic wave arrived using some warning time provided by the EEW system. Through successful operation in practice, the integrated EEW system was verified as an effective tool for disaster prevention at schools.

Identification of ground motion features for high-tech facility under far field seismic waves using wavelet packet transform

Seismic records collected from earthquake with large magnitude and far distance may contain long period seismic waves which have small amplitude but with dominant period up to 10 sec. For a general situation, the long period seismic waves will not endanger the safety of the structural system or cause any uncomfortable for human activity. On the contrary, for those far distant earthquakes, this type of seismic waves may cause a glitch or, furthermore, breakdown to some important equipments/facilities (such as the high-precision facilities in high-tech Fab) and eventually damage the interests of company if the amplitude becomes significant. The previous study showed that the ground motion features such as time-variant dominant frequencies extracted using moving window singular spectrum analysis (MWSSA) and amplitude characteristics of long-period waves identified from slope change of ground motion Arias Intensity can efficiently indicate the damage severity to the high-precision facilities. However, embedding a large hankel matrix to extract long period seismic waves make the MWSSA become a time-consumed process. In this study, the seismic ground motion data collected from broadband seismometer network located in Taiwan were used (with epicenter distance over 1000 km). To monitor the significant long-period waves, the low frequency components of these seismic ground motion data are extracted using wavelet packet transform (WPT) to obtain wavelet coefficients and the wavelet entropy of coefficients are used to identify the amplitude characteristics of long-period waves. The proposed method is a timesaving process compared to MWSSA and can be easily implemented for real-time detection. Comparison and discussion on this method among these different seismic events and the damage severity to the high-precision facilities in high-tech Fab is made.

Multi-subnet wireless sensing feedback for decentralized ℋ 2 control with information overlapping

This paper studies a time-delayed decentralized structural control strategy that aims to minimize the H2 norm of the closed-loop system. In a decentralized control system, control decisions are made based on data acquired from sensors located in the vicinity of a control device. Due to the non-convexity nature of the optimization problem caused by a decentralized architecture, controller design for decentralized systems remains a major challenge. In this work, a homotopy method is employed to gradually transform a centralized controller into multiple decentralized controllers. Linear matrix inequality (LMI) constraints are adopted in the homotopic transformation to ensure closed-loop control performance. In addition, multiple decentralized control architectures are implemented with a network of wireless sensing and control nodes. The sensor network allows simultaneous operation of multiple wireless subnets. Both the theoretical development and system implementation support the information overlapping between decentralized subnets. For validation, the wireless sensing and control system is installed on a six-story laboratory steel structure controlled by magnetorheological (MR) dampers. Shake-table experiments are conducted to demonstrate the performance of the wireless decentralized control strategies.

Theoretical and experimental study of vibration suppression for stayed cable

The objective of this study is to develop a numerical model of a stay cable interacted with deck, and to examine the vibration suppression technique of the stayed cable subject to external loading. First, a numerical model based on the finite difference method and the finite element method has been developed to simulate the effects of the bending stiffness and its sag-extensibility characteristics of the cable. Accurate vibration mode shapes and modal frequency of the interaction between stay cable and deck are examined. For the vibration control of cable, a MR-damper is used as control device. This damper can be achieved either through the passive control strategy or the semi-active control strategy employing decentralized sliding mode control (DSMC) and maximum energy dissipation (MED) on the staycable. To verify this study, a scaled-down cable structure is designed and constructed in NCREE, Taiwan. A small shaker is designed and mounted onto the cable to generate the sinusoid excitation with different amplitudes and frequencies. Dynamic characteristics of the cable-deck system are identified and the system model is developed for control purpose. The DSMC algorithm using MR damper was studied to reduce the cable vibration under different excitation frequencies.