A new integrated intelligent computing paradigm for predicting joints shear strength

Joints shear strength is a critical parameter during the design and construction of geotechnical engineering structures. The prevailing models mostly adopt the form of empirical functions, employing mathematical regression techniques to represent experimental data. As an alternative approach, this paper proposes a new integrated intelligent computing paradigm that aims to predict joints shear strength. Five metaheuristic optimization algorithms, including the chameleon swarm algorithm (CSA), slime mold algorithm, transient search optimization algorithm, equilibrium optimizer and social network search algorithm, were employed to enhance the performance of the multilayered perception (MLP) model. Efficiency comparisons were conducted between the proposed CSA-MLP model and twelve classical models, employing statistical indicators such as root mean square error (RMSE), correlation coefficient (R²), mean absolute error (MAE), and variance accounted for (VAF) to evaluate the performance of each model. The sensitivity analysis of parameters that impact joints shear strength was conducted. Finally, the feasibility and limitations of this study were discussed. The results revealed that, in comparison to other models, the CSA-MLP model exhibited the most appropriate performance in terms of R² (0.88), RMSE (0.19), MAE (0.15), and VAF (90.32%) values. The result of sensitivity analysis showed that the normal stress and the joint roughness coefficient were the most critical factors influencing joints shear strength. This paper presented an efficacious attempt toward swift prediction of joints shear strength, thus avoiding the need for costly in-site and laboratory tests.