Decision tree–based seismic damage prediction for reinforcement concrete frame buildings considering structural micro-characteristics

Abstract

A decision tree–based seismic vulnerability method for reinforcement concrete frames is proposed. Structures with stories equal to 3, 6, 9, and 12 were considered herein and 45,360 (4 × 11,340) reinforcement concrete frame damage samples with different micro-characteristic values were simulated using the capacity spectrum method. Afterward, with the adoption of CART algorithm, a decision tree was derived to visualize the relationship between the structural characteristics and damage states according to training samples. Damage prediction can then be made for unseen structures according to their characteristic values directly using the configured decision trees. Ten training and testing sets were established randomly from the sample library and their seismic vulnerabilities under three earthquake intensity levels were assessed to verify the proposed method. The results show that the decision tree predictor is efficient for seismic vulnerability assessment of reinforcement concrete frames, and the predictor shows high prediction accuracy and stability.