Prediction of socketed pile settlement based on a hybrid form of multilayer perceptron via meta-heuristic algorithms

The settlement of pile foundations is a complex problem studied extensively in geotechnical engineering. However, accurately predicting the settlement amount remains challenging due to soil behavior's nonlinear and stochastic nature. This study proposes a novel approach that integrates machine learning, specifically multilayer perceptron (MLP), and optimization algorithms to predict pile settlement accurately. The MLP model is trained using a dataset of pile settlement measurements obtained from field tests. The MLP model is then optimized using three different algorithms: Coronavirus herd immunity optimizer (CHIO), sunflower optimization algorithm (SFA), and Killer Whale Algorithm (KWA). These optimization algorithms are chosen for their ability to solve complex optimization problems and improve the MLP model's accuracy. The study results indicate that the proposed approach improves the accuracy of predicting pile settlement compared to traditional methods. The MLP model's accuracy is improved by applying the optimization algorithms, with the CHIO outperforming the other two. The study also demonstrates the feasibility of using machine learning and optimization algorithms to solve geotechnical engineering problems. In conclusion, this study presents a new approach to predicting pile settlement using machine learning and optimization algorithms. The proposed approach improves the accuracy of predicting pile settlement and demonstrates the feasibility of using these tools in geotechnical engineering. Future research can focus on testing this approach on different datasets and evaluating its performance under various soil conditions.