Investigation of ANN Architecture for Predicting Load-Carrying Capacity of Castellated Steel Beams

Castellated steel beams (CSB) are an attractive option for the construction industry thanks to outstanding advantages, such as ability exceed large span, lightweight, and allowing flexible arrangement of technical pipes through beams. In addition, complex localized global failures characterizing these structural members have led researchers focus on development efficient design guidelines. This paper aims propose artificial neural network (ANN) model with optimal architecture predict load-carrying capacity CSB a scheme simple beam bearing load located at center beam. The ANN is built 9 input variables, which essential parameters equivalent geometrical properties mechanical material, overall depth castellated beam, vertical projection inclined side opening, web thickness, flange width, width post middepth, horizontal minimum yield stress, stress. output variable CSB. With [9-1-1] containing one hidden layer, performance evaluated based statistical criteria R2, RMSE, MAE. results show that highly effective predictor best value R2 = 0.989, RMSE 3.328, MAE 2.620 testing part. seems be algorithm machine learning predicting capacity.