類神經網路及振動訊號在結構破壞檢測之應用

Abstract

結構物設計及施工完成後,常因許多原因而導致結構物發生破壞,當結構發 生破壞時,如果有適當的檢測設備,檢驗出其破壞,並發出警訊,可使工作人 員作適當的補強與修護,以減少結構之損害,達到預防勝於治療的目的。本 文主要是以含裂縫之金屬桿件進行模態振動實驗,將量測到的時域訊號轉 換為頻域訊號,並將訊號做特徵化處理,以類神經網路建立金屬桿件受損型 式 (位置.深度) 與振動訊號之關連性,而後一旦獲取未知受損金屬桿件之 振動訊號時,即可藉類神經網路做桿件之裂縫位置及受損程度之判別，以 此建立受損結構之整體診斷架構。 The structures are damaged, due to its enviromental chang and overloading, after their completion. If they are maintained properly with routingly examination and repairation, eir service life will be extened. This research ultilized the modal testing of cracked steel beam to explore the vibrational characteristics of the crackedeam. The time domain signal was transformed into frequencymain by FFT. These data were used to setup a diagnose theition and depth of the cracked structure from its vibrational signal.