頻率反應函數於房屋結構健康監測的應用

Abstract

長期以來天然災害(如台灣常見的地震、颱風等)對大地上的結構物常造成不同程度的損壞，使得人們的生命財產受到嚴重威脅，故於災害後若能快速的對結構物進行健康診斷與修復評估，為目前國內外學界致力研究的趨勢之一，又近年來識別資料處理的技術和量測結構反應的儀器、方法等皆有長足的進步，如能善加利用將可大幅的提升結構健康監測所帶來的生命保障和經濟效益。而本研究主要為將頻率反應函數應用於結構健康監測中，首先使用SAP2000軟體建立數值模型，以獲得模型在受外力擾動下的反應訊號，接著利用頻率反應函數結合指標公式形成原始的頻率反應函數指標圖，並利用曲率法改良之，使改良後的指標圖能判斷任意損壞案例的損壞位置，而損壞程度方面則是以類神經網路學習單點案例的損壞程度，此方法在判斷任意單點損壞案例的損壞程度上亦有良好的效果，唯多點損壞案例的損壞程度無法由類神經網路整合建立，待數值模型的損壞監測流程建置完成後，最終再嘗試以實驗資料驗證系統應用於實際的可行性。

A building structure may sustain damage either when subjected to severe loading like a strong earthquake or when its material deteriorates. The monitoring of the structural health (SHM) of buildings and civil infrastructure has attracted considerable interest in the last decade. Monitoring the structural health of a particular structural system involves a damage identification process that comprises detecting, locating, identifying type of, and estimating severity of, damage. Hence, a reliable, feasible, and economic SHM system is practical for building structures. The aim of this study attempts to utilize frequency response function (FRF) as index in SHM for locating and estimating severity of damage to building structures. First, a numerical study, based on SAP2000 software, is employed to create simulated damage cases including single and multi site(s) damage scenarios. Following, a novel FRF curvature method is developed and applied to locate damage. An ANN model is then utilized to evaluate the severity of damage for single-site cases. Finally, an experimental study is employed to verify the feasibility of the proposed FRF method. The simulation and experiment results expose that the proposed FRF index can locate damage correctly in single as well as multi site(s) damage and the ANN model can estimate the severity of damage for single-site damage cases acceptably.