| Summary: | 碩士 === 國立成功大學 === 建築學系碩博士班 === 92 === This research intends to identify a structural damage index and to establish a damage diagnosis system to detect building damage after a major earthquake so that the remedial process can be proceeded immediately in the post-earthquake recovery.
In the first part, SAP2000 is used to perform nonlinear time history analysis of plane frame and the space frame structures. Inter-story Drift Mode Shape (IDMS) is chosen as the key index in detecting damage conditions. The variation of IDMS before and during earthquakes is then compared to indicate the existence of the structural damage. Thereafter, the sensitivity of the IDMS variation to different degree of structural damage is also compared. It is concluded that IDMS is adequate to identify structural damage condition in which the floor stiffness is reduced above 30%. Additionally, the dynamic characteristics records of the existing building arrays instruments installed by the Central Weather Bureau in Taiwan are utilized to calculate the baselines of different buildings. It is found that baselines of S and SRC buildings are more accurate than that of RC buildings. In conclusion, the results obtained by using IDMS to analyze a shaking table test study of RC frame model show that the application of IDMS for damage detection is satisfactory.
Secondly, according to the superiority in coping with complex data and operation speed, we try to apply the artificial neural networks (ANNs) to identify the structural damage. IDMS is used as the input data in this case to train and test a back-propagation neural network. The training reveal that ANNs is effective to be a damage assessment technique for the diagnosis of structures. Using the computer model of an existing structure located in Kaohsiung has been proved that the network can discover the damage successfully. Meanwhile, establishing an integrated process to diagnose damage immediately by using IDMS and ANNs.
|
|---|
