Evaluation of load-settlement behavior of shallow footings using hybrid MLP-evolutionary AI approach with ER-WCA optimization

In this study, the multilayer perceptron (MLP) model underwent optimization using evolutionary artificial intelligence algorithms. This optimization was further enhanced by integrating the Evaporation Rate-based Water Cycle Algorithm (ER-WCA). This integrated approach resulted in a refined technique employed to forecast the load-settlement behavior of shallow footings located near slopes. Addressing this intricate engineering challenge necessitates a comprehensive approach, considering various input variables such as unit weight (UW) (kN/m³), elastic modulus (EM) (kN/m²), friction angle (FA), dilation angle (DA), Poisson's ratio (PR) (v), and setback distance (SD) (m). To construct the requisite dataset, finite element analysis was conducted. Throughout the model's implementation, it became apparent that the hybrid model’s performance was notably influenced by the population size parameter in ER-WCA (R² = 0.9964 and 0.99631, RMSE = 20.4937 and 19.53741). Consequently, the proposed hybrid model demonstrated significant potential in accurately predicting the vertical load necessary to achieve a specific footing settlement.