Reliability Prediction of SMRF Based on the Combination of Neural Network And Incremental Dynamic Analysis

Abstract

This paper conducts a comprehensive vulnerability analysis of steel structures, taking into account the stringent HAZUS restrictions. The demand distribution for each mode of failure takes the form of a normal logarithm after extracting the fragility chart. Thus, the two parameters of mean and standard deviation can be used to construct the fragility chart. A total of five modes of failure were used in this paper. Therefore, 10 unknown values were used to derive the fragility curves. Afterward, Incremental Dynamic Analysis (IDA) was used under 40 natural records to obtain the fragility curve. To save time in the analysis and prediction of structural responses, a neural network method was used to select records more efficiently. It was observed that this method is better than the analytical method in considering random uncertainty in steel structures when several acceleration values are used.