近年來,近斷層地震引起學者廣泛的注意與重視,以921集集大地震為例,震央附近之車籠埔斷層錯動,使得鄰近斷層之建築物及橋梁產生毀壞和傾倒,斷層錯動產生永久地表位移並發生極大的速度脈衝為造成結構物破壞的主要原因之ㄧ。由於近斷層之地震紀錄並不如遠域地震者豐碩,因此本研究針對近斷層速度脈衝之波形特徵,建立一套近斷層人造地震歷時之模擬方法,採用最佳化理論且利用人工智慧 (AI,Artificial Intelligent) 中的基因演算法 (GA,Genetic Algorithm) 作為求解工具,進而製作近斷層人造地震波,以便快速地得到近斷層歷時供為結構分析與研究之用。 此外,依據現行耐震設計規範之規定在結構物進行動力歷時分析時,需採用與設計反應譜相符之水平地震紀錄,因此,本研究建立一套視窗化程式,內含建築物、公路橋梁與鐵路橋梁等耐震設計規範之設計反應譜並具有產生與反應譜相符 (response-spectrum-compatible) 之地震波的運算功能,使用者只需透過視窗選擇工址資料,並給予實測的地震紀錄,經由運算,就能方便且快速地得到符合設計反應譜的地表運動歷時。本文所得成果可供為工程師進行結構耐震設計之用。
Recently, more and more researches pay attention on the structural damages caused by near-fault earthquake. Since the near-fault earthquake can be characterized by a special wave form with a significant permanent ground displacement and a serious velocity pulse, it is not so popular in the past. As a result, its records of ground motion are far fewer than those of far-field earthquake. This thesis proposed an genetic-algorithm-based optimization model to simulate the important characteristics of near-fault earthquake. It can be seen that the results obtained could provide a satisfactory precision. Based on the design code, the ground motions with a spectrum compatible to that specified in the code are necessary for time history analysis in seismic design. This thesis developed a analysis program with visual operation system to establish the artificial ground motions required. The application to the design of building, highway bridge and railway bridge are included. By which, the complicated procedure for the seismic design can be simplified. The results obtained could betinit the practical design efforts for the engineers.