應用隨機子空間識別法於結構健康診斷：結合穩態圖穩定標準與頻域分解法

Application of Stochastic Subspace Identification in Structural Health Monitoring： Combining the Stabilization Diagram and Frequency Domain Decomposition

郭采蓉(Tsai-Jung Kuo)

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(2018年)

https://doi.org/10.6342/NTU201801252

摘要

本研究依據前人的研究成果，基於隨機子空間識別法（SSI）發展出一套唯輸出系統識別方法（Output-Only System Identification）應用於結構健康診斷（SHM）當中。以協方差型隨機子空間識別法（SSI-COV）做為識別方法的基礎，引入多個穩定標準（criteria）去除穩態圖（stabilization diagram）中不穩定的系統極點，並運用奇異值決定系統階數的選取範圍。如此一來，便可以增加識別結果的穩定性並減少由操作者主觀意識的不同而產生的誤差。此外，結合精緻頻域分解法（rFDD）區分結構模態與諧波（harmonic）。協方差型隨機子空間識別法由前述過程去除虛假模態與諧波模態後，即可獲得準確且穩定的模態參數。最後，再利用識別出來的模態參數建立柔度矩陣來偵測損壞位置。另外，再結合奇異譜分析法（SSA）與隨機遞減法（RDM）提出一種阻尼比的識別方式。本研究使用振動台試驗的試體（八層樓鋼構架）以及兩個實際結構（中保雲端數據中心大樓、關渡大橋）來進行驗證，由分析結果顯示，本文提出之方法能夠有效的識別出結構的模態參數並成功偵測損傷位置。

關鍵字

結構健康診斷 ； 唯輸出系統識別 ； 隨機子空間識別法 ； 頻域分解法 ； 奇異譜分析法 ； 隨機遞減法

並列摘要

This study is to develop an output-only system identification method using covariance-driven stochastic subspace identification (SSI-COV) for structural health monitoring (SHM). In applying SSI-COV for structural system identification the method does not yield exact values for the parameters but only estimates with uncertainties. These uncertainties are responsible for the appearance of spurious modes. One of the important challenge is to remove the spurious modes from which the stabilization diagram was introduced to remove unstable system poles. The quality of the stabilization diagram depends on the values of the input parameters of the algorithm and the noise ratio of the time series under analysis. Criteria to remove the spurious modes were proposed which include: examine the physical poles under a fix model order and check stabilization criteria between different modal order. Besides, the discussion on the identification of correct modes are presented by in cooperating the refined frequency domain analysis (rFDD) with SSI-COV method. Harmonic mode can also be detected from this examination. In cooperated with the identified accurate modal parameters the system flexibility matrix can be constructed and applied for structural damage assessment. In addition, combined with single spectrum analysis (SSA) and random decrement method (RDM), a more stable system damping ratio can be estimated. To verify the proposed algorithms, data from the shaking table of an eight-story steel structure and two actual structures (SIGMU Building and Guan-Du Bridge) were used for verification. The analysis results show that the method proposed in this paper can effectively and accurately identify the dynamic characteristics of structure under operating condition.

並列關鍵字

structural health monitoring ； output-only system identification ； stochastic subspace identification ； frequency domain decomposition ； singular spectrum analysis ； random decrement method

參考文獻

[1] J.N. Juang (1994). Applied system identification. Englewood Cliffs, NJ, USA﹕Prentice Hall.

[2] L. Ljung (1999). System Identification: Theory for the User. Second edition, Prentice Hall, Upper Saddle River, NJ, USA.

[3] L. Zhang, and R. Brincker (2005), An overview of operational modal analysis: major development and issues, in: Proceedings of First International Operational Modal Analysis Conference, Copenhagen, Denmark.

[4] G.H. Golub, and C.F. Van Loan (1996). Matrix computations. Johns Hopkins University Press.

[5] P. Van Overschee, and B. De Moor (1991). Subspace Algorithm for the Stochastic Identification Problem. In Proceedings of the 30th IEEE Conference on Decision and Control, pp. 1321-1326.