A seismic damage index monitoring system is presented in this paper. The method is based on artificial neural networks and measured responses of the structure. Park & Ang damage index is selected for monitoring the damage at each story of the structure. The system is consisted of two neural networks. The input to the first NN is the maximum drift of each story and the output is the damage due to maximum drift of the story. Also the input to the second NN is the differential drift and the sum of the accelerations of stories above each story. The Output of the second NN is the cumulative damage of the story. The data needed for training of neural networks are collected from analysis of a simulated model of the structure under different earthquake records. The damage index in each analysis is computed from analytical method. The performance of damage detection system is determined from comparison of damage index computed from analytical method to one determined by neural networks in a three story benchmark building. Through this study, it is shown that the proposed approach has the potential of being a practical tool for damage monitoring methodology applied to smart civil structures.

Keywords