子空間識別法於系統識別及結構損壞診斷之應用

Application of Subspace Identification in System Identification and Structural Damage Detection

翁健煌(Jian-Huang Weng)

指導教授 ： 羅俊雄

國立臺灣大學/工學院/土木工程學研究所/博士(2010年)

https://doi.org/10.6342/NTU.2010.02940

摘要

本研究的目的在於探討子空間識別法(Subspace Identification)於系統識別及結構損壞診斷上的應用範疇，並根據實際的分析結果來決定適合發展於土木結構物長期健康監測系統的相關技術。首先，根據使用資料的類型將子空間識別法分為兩種並介紹各自的理論推導：(1)單獨以輸出資料為基礎進行系統識別的隨機子空間識別法(Stochastic Subspace Identification, SSI)，其實際分析時所使用的資料為結構微振動下所量測得的反應；(2)以輸入及輸出資料為基礎的子空間識別法(Subspace Identification, SI)，其使用資料為地震時所記錄到的地表加速度歷時與結構反應。在應用上，子空間識別法會面臨如何決定系統參數及挑選真實模態的問題，本研究將以實際案例的分析經驗來說明相關問題的解決之道。在結構損壞診斷的技術發展上，本研究將介紹三種不同的方法：(1)以零子空間為基礎的損壞診斷法(Null Subspace-Based Damage Detection)；(2)正規化勁度與阻尼矩陣的逆算法(Inverse Methods)；(3)二階段有限元素更新法(Two-State Finite Element Model Updating)，利用子空間識別法的分析結果，上述三種方法皆可準確地識別出結構的實際損壞狀況。為達成建立長期監測系統的目的，本研究提出一套適用於更新LQ分解(LQ Decomposition)的演算方法，有助於加速傳統遞歸式子空間識別法的計算時間，由數值模擬與真實實驗之分析結果顯示，該方法理論推導無誤並可有效地追蹤識別結構特性的改變，最後，此方法已經被使用在台大土木研究大樓的監測系統開發上，經由分析結果可知，該結構隔震層在地震時的非線性行為確實會導致結構自然頻率與阻尼比的改變，並與隔震層的層間位移量有直接關係。

關鍵字

子空間識別 ； 系統識別 ； 損壞識別 ； 健康診斷 ； 中間層隔震 ； 監測系統 ； LQ分解法 ； 遞歸式系統識別

並列摘要

This study applies the subspace identification to system identification and damage detection. Many of the damage detection methods are surveyed in order to establish a continuous monitoring system of the civil engineering structures. First, according to the use of the collection data two theoretical derivations are introduced: (1) the stochastic subspace identification (SSI) using output-only data, or the ambient vibration data for continuous monitoring; and (2) the subspace identification (SI) using input/output data, or the seismic responses. The user-defined parameters in the subspace identification are studied through the laboratory and field tests and also three noise-rid methods are employed to distinguish the noise modes from the results of analysis. For damage detection three methods are introduced: (1) null subspace-based damage detection; (2) inverse methods for the identification of the normalized stiffness and damping matrices; and (3) two-stage finite element model updating. These methods can accurately detect the location or quantification of the structural damage from the results of identification. For the purpose of continuous monitoring, in this study a new technique for updating LQ decomposition is proposed to achieve the recursive computation of the subspace identification. From the results of the numerical simulation and the laboratory experiment it is proven that the proposed method of the recursive subspace identification can accurately detect the abrupt damage of the structure. Finally, the proposed methods are applied to the long-term monitoring of the new Civil Engineering Department Research Building in the campus of National Taiwan University, which is a mid-story isolation building. The relationship between the natural frequencies and the drift of the isolation system is established to provide a reference for future research.

並列關鍵字

Subspace Identification ； Damage Detection ； Structural Health Monitoring ； Mid-Isolation ； LQ Decomposition ； Recursive Subspace Identification

參考文獻

Akaike, H. (1975). Markovian representation of stochastic processes by canonical variables, Siam J. Control, 13(1) 162-173.

Alonso, F. J., Castillo, J. M. Del, and Pintado, P. (2005). Application of singular spectrum analysis to the smoothing of raw kinematic signals, Journal of Biomechanics, 38, 1085-1092.

Alves, S. W. and Hall, J. F. (2006) System identification of a concrete arch dam and calibration of its finite element model, Earthquake Engineering and Structural Dynamics, 35, 1321-1337.

Amani, M. G., Riera, J. D., and Curadelli, R. O. (2007). Identification of changes in the stiffness and damping matrices of linear structures through ambient vibrations, Structural Control and Health Monitoring, 14, 1155-1169.

Baruch, M. and Bar-Itzhack, I. Y. (1978). Optimal weighted orthogonalization of measured modes, AIAA J., 16(4), 346-351.