應用隨機子空間系統識別方法探討橋梁結構健康診斷

Application of Stochastic Subspace Identification in Bridge Structural Health Monitoring

陳明徹(Ming-Che Chen)

指導教授 ： 羅俊雄

國立臺灣大學/工學院/土木工程學研究所/碩士(2012年)

https://doi.org/10.6342/NTU.2012.02956

摘要

本研究的目的在於應用唯輸出(output-only)系統識別方法-隨機子空間識別法(Stochastic Subspace Identification, SSI)於橋梁結構健康診斷。在離線分析的應用上，將不同矩陣維度大小所識別出的系統極點(system pole)繪至於穩態圖上，以得到正確的結構振態；為達成建立長期監測系統目的，本研究採用一套適用於更新SVD分解(SVD Decomposition)的演算方法:延伸工具變項-投影逼近子空間追蹤法(Extended Instrumental Variable-Projection Approximation Subspace)達成線上更新子空間的目的。將遞迴式協方差型隨機子空間識別法(Recursive Covariance-driven Stochastic Subspace Identification, RSSI-COV)應用於關渡大橋(Guan-du Bridge)動態參數識別，並做連續時間的資料蒐集以探討動態參數與環境因子間之關係。接著對隨機子空間識別法的識別參數做研究，在不同的參數條件如:列區塊數(Block row)、視窗長度與系統階數(System order)的設定結果作一系列之敏感性分析，討論及比較其差異並從中獲得較佳的參數值，此外奇異譜分析法(Singular Spectrum Analysis, SSA)做為前置濾波器之概念與隨機子空間識別法結合，可有效低提升資料解析能力並消除噪訊的干擾，最後，考慮以上敏感性分析因子可得較佳的識別結果，此識別結果透過非線性主成份分析-自相關類神經網路(Auto-associative Neural Network, AANN)可得到頻率和溫度及車載間之關係，觀察頻率值受環境因子變化而改變，期許作為往後識別損壞基準。

關鍵字

隨機子空間識別 ； 遞迴式隨機子空間識別 ； 關渡大橋 ； 系統識別 ； 敏感性分析 ； 奇異譜分析 ； 自相關類神經網路

並列摘要

In this research application of output-only system identification technique, known as Stochastic Subspace Identification (SSI) algorithms in bridge health monitoring. With the aim of finding accurate modal parameters of the structure in off-line analysis, a stabilization diagram is constructed by plotting the identified poles of the system with increasing the size of data matrix. For the purpose of continuous monitoring, in this study a new technique for updating SVD decomposition: Extended Instrumental Variable version of Projection Approximation Subspace Tracking algorithm (EIV-PAST) is taking charge of the system-related subspace updating task. In the following, the identification task of a real large scale structure: Guan-du bridge, a benchmark problem for structural health monitoring of arch steel bridge is carried out, for which the capacity of Recursive Covariance-driven Stochastic Subspace Identification (RSSI-COV) will be demonstrated. In order to study the influence of loading from environments, consecutive measurement of the dynamic response had been carried out. A sensitivity study of the parameters of SSI is carried out, different parameter conditions such as block row, window length and system order are considered. The introduction of a pre-processing algorithm known as Singular Spectrum Analysis (SSA) can greatly enhance the identification capacity and reduce noise interference. Finally, consider the parameter sensitivity analysis we obtain good identification results, the uncertainty of system dynamic characteristics of the experimental target due to traffic loading and temperature is investigated through nonlinear principal component analysis known as Auto-associative Neural Network (AANN), observed the modal frequencies of the structure change due to environmental changes, as a benchmark for the future to identify the damage detection.

並列關鍵字

Stochastic subspace Identification ； Recursive Stochastic Subspace Identification ； Guan-du Bridge ； System Identification ； Sensitivity Analysis ； Singular Spectrum Analysis ； Auto-associative Neural Network

參考文獻

[29] 劉勛仁, “依唯結構反應量測之系統模態參數識別”, 國立中興大學土木工程研究所碩士論文, 2008.

[1] Chang, Peter C.; Alison, Flatau and Liu, S.C.. Review Paper: Health Monitoring of Civil Infrastructure. Structural Health Monitoring 2003 2:257.

[2] Doebling, Scott; Farrar, Charles and Prime, Michael. A summary review of vibration-based damage identification methods. Shock and Vibration Digest, 30(2), 91-105, 1998.

[5] Soderstrom, Torsten and Stoica, Petre. System identification. Prentice Hall International, 1989.

[6] Golub, Gene H.; Van Loan, Charles F. Matrix computations. Johns Hopkins University Press, 1996.