建築物微振量測之最佳感測器配置

Abstract

台灣位屬多地震區域，由於地震頻繁，為了結構物之安全性，需於定期進行結構物健康檢測，尤其是災害發生後能確保結構物之安全。設置感測器極為耗成本且耗時，故本文所提出之最佳化配置為減少再次檢測同棟結構物時所需之感測器顆數，並且提高判斷結構物模態頻率的精度。為此目的，本文結合了四種方法進行最佳化配置，分別為隨機子空間識別法、三次樣條內插法、K-means演算法與基因演算法等方法判斷出加速度感測器配置之最佳樓層位置。提出各種不同模型，並與模態頻率解析解相互比較，驗證此法能以少數層樓之時間歷時，得到整體結構物的模態頻率，以確保結構物之安全性。為了驗證此方法之可行性，於數值模擬上，本研究建立幾個不同勁度折減之模型，再次驗證結構物災害發生前以及災害發生後，最佳配置差異；在現地實驗方面，本研究以六棟真實結構物進行實驗，分別為國立台灣大學土木研究大樓、國立台灣大學醫學院附設癌醫中心、淡水施工中20層樓建築物、板橋兩棟20與21層樓建築物及國家地震工程研究中心等建築，並以傅立葉快速轉換法來驗證此本文方法之可行性，由各種數值模擬以及實驗上證明。故本文最主要的貢獻在於提出之感測器最佳配置法，可以大幅減少第二次檢測時所使用之感測器數量與獲得較為精準之中高模態頻率，作為日後檢測之重要相關依據，當建築物受到損害時，即時檢測以瞭解建築物之受損情形。

Taiwan is located in the seismic zone with high frequency earthquake occurrences. In order to increase structure safety, it needs to monitor the structural health before and after disaster occurs. This study proposes a method to obtain the optimal sensors placements (OSP), which could reduce the number of sensors for building monitoring. In additions, the method could find out the higher modal frequencies for structures. First, collect the real time-histories and use Cubic spline interpolation method to obtain simulated time-histories for each floor. Second, use Stochastic Subspace Identification to generate stabilization diagrams. Third, K-means clustering method is used to obtain modal frequencies. Finally, use Genetic Algorithm method to find OSP. In this study, created some model to proved this method that can obtained the high accuracy modal frequency, and then created some model which stiffness decrease to proved that used OSP method to obtain modal frequency before and after stiffness decrease which similar with real modal frequency. There are six in-situ experiments for the method verifying. The six in-situ building are NTU Civil Engineering Research Building, NTU Cancer Center, Tamsui building (under construction), two apartment complex in Banqiao and National Center for Research on Earthquake Engineering, respectively. The OSP method can obtained great result from these in-situ experiments.