CA and RVM combined method for predicting collapse of vertical irregular mid-rise cold-formed steel structures under strong earthquake

The existing cellular automata (CA) method is improved to increase the prediction accuracy of vertical irregular mid-rise cold-formed steel framed structures with composite shear walls under strong earthquake. Firstly, the number of neighbor cells is increased, and the normalized inter-story drift is regarded as the cell state value through performing Pushover analysis on the structure. Then the combined method of CA and relevance vector machine (RVM) is developed by replacing the existing similar region matching criteria with RVM. Finally, the combined method and the existing CA method are used to predict the collapse inter-story drift of vertical irregular mid-rise cold-formed steel structure under strong earthquake (the collapse criterion is that the inter-story drift angle is more than 2.0%, 3.5% or 4.0%). The prediction results are compared with the results of nonlinear time history analysis.The results show that by taking the eigenvalue of the elastic-plastic stage as the cell state value, and selecting RVM to establish the nonlinear relationship between the cell state value and the inter-story drift when the structure collapses, the combined method can accurately predict the inter-story drift of vertical irregular mid-rise cold-formed steel structures when they collapse. Compared with the original CA method, the prediction accuracy is significantly improved.