拉力裂縫鋼棒之動態特性分析及損傷檢測

Abstract

結構物於建造完成之後，可能因許多原因開始產生損壞，若能有效地診斷結構物桿件所存在的裂縫與受力狀況，就可以有效地防止整體結構毀損，杜絕桿件的斷裂。本研究即在於以有限元素法模擬拉力裂縫桿件，配合振動模態實驗，分析其動態之頻率與模態特性。 動態特性與裂縫之位置、深度、桿件內含拉力間有許多的關連性，利用這些關連性，可以定義裂縫與拉力的資訊，再以類神經網路學習訓練。本研究成果顯示使用訓練的成果來檢定裂縫的位置、深度與桿件拉力，而建立一套完善的桿件拉力裂縫檢測模組。

The constructed Structures exist lots of damages existing, due to thousands of reasons. If we could find out the cracks and the forces in the members, we would prevent that the whole structure members fractured and destroyed effectively. The dynamic characteristics of frequencies and modeshapes were analyzed bye the finite element approach and verified by model testing in this research. These dynamic characteristics have closed relations with crack locations, depths, and forces in the members. By these relations, we could identify the crack locations, depths and forces inside the member. These relationships were used to learning and training by Artificial Neural Networks to identify the crack locations, depths, and forces inside the members. According to the results of this research, we could identify the crack locations, depths, and forces inside, and accurately the detecting measure can be setup.