Hybrid Machine Learning Approach for Accurate Bearing Capacity Prediction of Geogrid-Reinforced Soils: Integrating ANN and Optimization Algorithms

Abstract In geotechnical engineering, identifying the bearing capacity of soil is a crucial and fundamental aspect. The strength and stability of any structure depend significantly on this bearing capacity, and it has a direct and considerable impact. There are many traditional estimation methods and formulas available to determine bearing capacity. Often, traditional methods don’t work well when dealing with complicated soil conditions because soil behavior can change a lot depending on various factors. That’s why using hybrid learning techniques can be really helpful. These advanced methods provide us a better chance of predicting the bearing capacity of soil more accurately. In this approach, a dataset of one thousand entries is taken, which is then divided into training and testing sets. In this study, we take into account several key factors such as cohesion (c), internal friction angle (φ), unit weight (γ), footing width (B), foundation depth (D), and whether the load is static or dynamic. To improve the accuracy of estimating the soil bearing capacity, here we make use of Genetic Algorithms (GA) and Particle Swarm Optimization (PSO). These hybrid, intelligent models have performed well not just in how they function but also in terms of their simplicity and the reliability of their results. In recent years, there has been growing interest in using artificial intelligence (AI) to estimate the bearing capacity of soils. Although these techniques do come with some limitations, they are particularly effective when it comes to working with complex or large datasets. A major strength of AI methods is their ability to Identify hidden patterns and Connections that traditional calculation methods often overlook.