狀態空間DLV法在剪力構架之地震損傷探測分析與實驗驗證

Abstract

本研究結合SRIM系統識別分析與狀態空間DLV損傷識別法，發展出一套以結構地震反應監測訊號為基礎之結構破壞偵測技術。DLV損傷識別法係以結構受損前、後之柔度矩陣為基礎，由其差異矩陣之奇異值分解找出損傷定位向量，凡在此特定荷載作用下應力為零之桿件，即為潛在之受損桿件。由於柔度矩陣對於高階模態較不敏感，因此僅以部份低階模態即可有效建構出柔度矩陣，靈敏反映結構之受損狀態。 本研究以平面結構為分析對象，經由數值模擬及一系列振動台地震模擬試驗，充分驗證本研究提出之狀態空間DLV損傷識別法可由結構之全域反應(樓層絕對加速度)定位出局部受損。試驗與分析結果顯示，在完全觀測條件下，無論是單一或複數樓層受損，皆可精準定位出受損樓層；在觀測不足之情況下，若受損樓層有觀測時，無論是單一或複數樓層受損，在所考慮之各種觀測組合下，都能將受損樓層成功定位出來；惟在受損樓層未作觀測時，本研究提出之方法皆無法成功辨識受損位置。建議未來實際建置結構健康監測系統時，應優先考慮於結構之軟弱層配置感應器。

In this study, a method for structural damage detection based on seismic response data is developed using the state-space damage localization technique and system realization of information matrix (SRIM). The method of damage locating vectors (DLV) is based on the flexibility matrix that the DLVs are obtained by performing singular value decomposition on the flexibility differential matrix of the structure before and after the damage state. Members with zero stress under the DLVs as external loads to the structure are considered potentially damaged. As the flexibility matrices are less sensitive to higher modes, they can be effectively constructed by considering only part of the lower modes, yet sensitively reflecting the damaged states of the structures. In this study, a planner shear-type structure is considered as the objective to verify, via a series of the numerical simulation and shaking table tests, the feasibility of identifying local structural damages from seismic global responses (specifically the absolute accelerations of the floors ) using the state-space damage localization method. Both analytical and experimental results indicate that, under the condition of full observation, localized storey damages can be precisely identified, regardless of single or multiple damages. In the case of insufficient observation, if the damaged floors are observed (co-located observation), then the damaged stories can still be identified, regardless of single or multiple damages. However, if the damaged floors are not observed (not co-located observation), then the proposed scheme fails to identify the damaged stories. Therefore, in real applications, response monitoring of the soft-and-weak stories in the structural health monitoring system are suggested with the highest priority.