Optimizing Groundwater Seepage Prediction in Tunnels Using the Human Mental Search (HMS) Algorithm: A Cognitive-Inspired Approach to Complex Geotechnical Challenges

Abstract

The paper investigates the application of the Human Mental Search (HMS) algorithm as an advanced optimization technique for predicting groundwater seepage in tunnel environments. Groundwater infiltration into tunnels poses a significant geotechnical challenge, potentially leading to structural instability, environmental degradation, and increased construction and operational costs. Traditional methods, including analytical, numerical, and empirical approaches, often encounter difficulties in addressing the complexities of subterranean water flow, particularly in heterogeneous geological conditions. Although machine learning techniques and optimization algorithms, such as Artificial Neural Networks (ANN) and Particle Swarm Optimization (PSO), have enhanced seepage prediction accuracy, they typically demand substantial computational resources and pre-processed data. The HMS algorithm, inspired by human cognitive processes, incorporates memory recall, adaptive clustering, and strategic selection, allowing it to refine solutions in an efficient and effective manner. This study demonstrates that HMS excels in predicting groundwater seepage, achieving remarkable accuracy and computational efficiency. The results show that HMS outperforms established algorithms in every metric tested, with an R² value of 0.9988, indicating an almost perfect fit between predicted and observed inflow values. Additionally, HMS achieves an exceptionally low Mean Squared Error (MSE) of 0.0002 and Root Mean Squared Error (RMSE) of 0.0137, underscoring its precision in minimizing prediction errors. In comparison, the Whale Optimization Algorithm (WOA) attains an R² value of 0.9951, MSE of 6.65, and RMSE of 1.71, while Genetic Programming (GEP) and ANN-PSO exhibit even higher error values. Furthermore, HMS records a Variance Accounted For (VAF) of 99.88% and a Mean Absolute Error (MAE) of 0.0041, further validating its ability to deliver highly reliable predictions with minimal error. These exceptional results highlight the HMS algorithm’s superior performance in optimizing groundwater seepage predictions, making it a highly effective and computationally efficient tool for geotechnical engineering applications. In addition to improving predictive modeling for tunnel seepage, HMS holds significant potential for real-time groundwater monitoring and other broader geotechnical applications.