非線性系統識別方法於結構健康監測之應用：非線性指標的探討

Nonlinear System Identification Method for Structural Health Monitoring: Techniques for the Detection of Nonlinear Indicators

毛建閎(Chien-Hong Mao)

指導教授 ： 羅俊雄

共同指導教授 ： 潘則建

國立臺灣大學/工學院/土木工程學研究所/碩士(2009年)

https://doi.org/10.6342/NTU.2009.00095

摘要

一個成功的結構物健康監測(Structural health monitoring)必須能夠經由分析結構物的反應訊號後，即能快速分辨破壞是否產生，左右其成功的關鍵在於結構破壞識別的運算方法。結構破壞識別主要可以區分為兩種：線性系統識別與非線性系統識別。本研究將考慮非線性系統識別方法(nonlinear system identification)的應用，其中以非參數方法(nonparametric method)為主，進行各種非線性指標的探討，包含了頻率域的非線性指標以及時間域的非線性指標。頻率域的非線性指標包含： (1) Hilbert transform of frequency response function, (2) coherence function, (3) Hilbert marginal spectrum, (4) wavelet packet transform component correlation coefficient, and (5) bispectral analysis; 時間域的非線性指標包含: (1) instantaneous frequency, (2) instantaneous phase difference, (3) Holder exponent, (4) discrete wavelet transform, and (5) singular spectrum analysis (SSA). 在介紹各個非線性指標的基本理論之後，本研究將會以一個一層樓的鋼筋混泥土架構的振動台試驗，進行各種非線性指標應用於真實結構物上的效果評估。分析的結果顯示，這些非線性指標可以有效探測出結構物是否有非線性行為，而這些非線性行為主要是由於結構物勁度折減、強度折減以及裂縫產生所致。另一方面，本研究也進行了SSA的延伸應用，包含：擷取結構物永久位移，去除噪音的影響，以及從加速度訊號推估結構物的永久位移量。從加速度訊號推估結構物的殘餘位移量是一個未來值得發展的主題，因為這個技術將有助於簡化結構健康監測的硬體設施，並使分析者能夠獲得更多結構破壞的資訊。此研究最終證實，經由適當的訊號分析以及非線性指標的應用，分析者能夠直接從量測資料中辨識出結構的損壞。

關鍵字

結構健康監測 ； 非線性系統識別 ； 訊號處理 ； 非線性指標 ； 奇異譜分析 ； 永久變位

並列摘要

With the progress of signal processing technologies, structural health monitoring (SHM) has received more and more attentions. The core algorithm in SHM is based on the detection of damage-sensitive indicator. In the recent decades, engineers already have the ability to deal with nonlinear problem. A literature survey of nonlinear indicators is firstly examined in the study. It is found that a successful SHM requires the monitoring technologies have their flexibility, simplicity, and, of course, accuracy. The nonparametric system identification method is a potential candidate which can meet these requirements. Therefore, several nonlinear indicators corresponding to the nonparametric system identification method are studied in this research, both from frequency and time domain analysis. In this research, the frequency-domain nonlinear indicators included: (1) Hilbert transform of frequency response function, (2) coherence function, (3) Hilbert marginal spectrum, (4) wavelet packet transform component correlation coefficient, and (5) bispectral analysis; and the time-domain nonlinear indicators included: (1) instantaneous frequency, (2) instantaneous phase difference, (3) Holder exponent, (4) discrete wavelet transform, and (5) singular spectrum analysis (SSA).
Test data from a series of shake table test to the 1-story 2-bay RC frame is generated from NCREE (National Center for Research on Earthquake Engineering), Taiwan. For these shake table tests data from two groups of specimens are analysed using the proposed nonlinear indicators. The first group of seismic response data is to consider the response from different specimen subjected to different level of seismic excitation (TCU082). The second group of data is to examine the damage level through a series of excitation back to back on a specimen. In cooperated with the experimental data, the result shows that nonlinear indicators can provide the identification of structural nonlinearities, which include stiffness degradation and cracks. Finally, the singular spectrum analysis (SSA) technique was used to extract structural residual deformation and to eliminate the noise effect. Furthermore, the SSA method can be used to derive residual displacement using the measured acceleration signal if there is no information on displacement measurement. In this thesis, a deeper realization to nonlinear indicators can be achieved. And it is possible to execute an effective SHM and nonlinear identification of the structure directly from the measurement by using appropriate nonlinear indicators.

並列關鍵字

structural health monitoring ； nonlinear system identification ； signal processing ； nonlinear indicator ； singular spectrum analysis ； permanent deformation

參考文獻

1. Doebling, S.W., C.R. Farrar, and M.B. Prime, A summary review of vibration-based damage identification methods. Shock and Vibration Digest, 1998. 30(2): p. 91-105.

2. Kerschen, G., K. Worden, A.F. Vakakis, and J.C. Golinval, Past, present and future of nonlinear system identification in structural dynamics. Mechanical Systems and Signal Processing, 2006. 20(3): p. 505-592.

3. Peng, Z.K. and F.L. Chu, Application of the wavelet transform in machine condition monitoring and fault diagnostics: A review with bibliography. Mechanical Systems and Signal Processing, 2004. 18(2): p. 199-221.

4. Reda Taha, M.M., A. Noureldin, J.L. Lucero, and T.J. Baca, Wavelet transform for structural health monitoring: A compendium of uses and features. Structural Health Monitoring, 2006. 5(3): p. 267-295.

5. Yen, G.G. and K.-C. Kuo, Wavelet packet feature extraction for vibration monitoring. IEEE Transactions on Industrial Electronics, 2000. 47(3): p. 650-667.