Shear Strength of Cellular Steel Beams Predicted by Hybrid ANFIS-ECBO Model

Cellular steel beams, also known as castle beams, are structural members designed to optimize steel structures while maintaining structural integrity. These beams are commonly used in buildings and bridges to support heavy loads and/or long spans. The purpose of this paper is to predict the shear strength of cellular steel beams using artificial intelligence methods using data obtained from experimental results and finite element analysis. The number of data used is 96. The number of inputs is 6 variables, which include beam web thickness, beam flange thickness, web cell height, web cell width, hole distance and number of holes in the beam web, and the output is the cell beam shear strength. To evaluate the model, quality evaluation criteria such as MSE, RMSE, MAE, MAPE, NMAE, NRMSE and correlation coefficient (R) have been used. In this study, the Adaptive Neural Fuzzy Inference System (ANFIS) is improved with ECBO meta-heuristic algorithm and a new hybrid computational model (ANFIS-ECBO) is used to predict the shear strength of cellular steel beams. This paper performs a sensitivity analysis on the prediction of shear strength of cellular steel beams. System identification (SI) is performed using ANFIS to find the most sensitive combinations of input variables. Six different models have been developed based on SI results. The results showed that model 5 gave better answers in both experimental and finite element methods. The highest correlation value and the lowest errors were related to model 5. Model 5 showed that the distance between Holes were the least important input in influencing the results. Also, the number of optimal membership rules has been identified with ANFIS. It was concluded that the ANFIS-ECBO model with 10 membership rules can accurately predict the shear strength of cell beams. Hence, the ANFIS-ECBO-based model can be used as a design tool.