Presenting innovative ensemble model for prediction of the load carrying capacity of composite castellated steel beam under fire

Abstract

The main aims of this study are related to present nonlinear analysis and prediction model for evaluating and predicting the performance of composite castellated steel beam (CCSB) under fire and static load. This study includes two parts, including parametric study and the prediction models. The beam size, concrete thickness, temperature, and beam length are considered as the independent variables in the parametric study. Among considered independent variables in this study, the increasing concrete thickness has a negligible positive effect on the load carrying capacity. In addition, gene expression programming (GEP), three regression models such as multiple linear regression (MLR), principal component regression (PCR), and multiple Ln equation regression (MLnER), the combination of GEP and MLR, and the combination of GEP, MLnER, and MLR are used to predict the load carrying capacity of the castellated beam. The maximum positive and negative errors, mean of absolute percentage error (MAPE), and statistical parameters such as the coefficient of determination, root mean square error (RMSE), normalized square error (NMSE), and fractional bias are utilized to evaluate and compare the performance of the models. According to the results of these parameters, the two best models specify. In addition, a distribution error table is used to specify the best model. Based on the results of this table and statistical parameters, the combination of GEP and MLR is considered as the best model for predicting the load carrying capacity of CCSBs.