INNOVATIVE AI-DRIVEN APPROACHES FOR ENHANCED ROCK MASS CHARACTERIZATION: A COMPREHENSIVE COST ANALYSIS AND PRACTICAL APPLICATION IN CIVIL ENGINEERING

Abstract

This study investigates the potential of innovative AI-driven approaches for rock mass characterization, aiming to enhance accuracy, consistency, and cost-effectiveness in civil engineering projects. Traditional rock mass classification methods like the Q-system and Rock Mass Rating (RMR) often rely on subjective judgment and extensive fieldwork, leading to inconsistencies and high costs. This research demonstrates significant improvements in classification performance by integrating advanced AI models, including logistic regression, decision trees, and support vector machines (SVM). The study utilizes comprehensive borehole data, rock joint set data, lab test results, and geological maps.

Key findings reveal that the logistic regression model achieved near-perfect accuracy, precision, recall, and F1 scores, establishing it as the most reliable model for rock mass classification. Decision tree and SVM models also performed strongly, capturing complex relationships between features and rock mass quality. The research underscores the importance of selecting relevant features, such as Rock Quality Designation (RQD), joint set number (Jn), joint roughness number (Jr), joint alteration number (Ja), joint water reduction factor (Jw), stress reduction factor (SRF), elevation, and depth.

These AI models have extensive practical applications, enhancing site investigations, foundation design, tunnel and underground construction safety assessments, slope stability, and landslide risk management. Implementing AI models leads to substantial cost savings by reducing the need for extensive fieldwork, optimizing resource allocation, and minimizing over-engineering.

This study contributes to the field by providing a framework for integrating AI into geotechnical engineering, highlighting the benefits of AI models in improving classification accuracy and reliability. Future research should focus on integrating larger datasets, exploring advanced AI techniques, and assessing the scalability and adaptability of AI models to different geological settings. The findings underscore the transformative potential of AI in rock mass characterization, promising improved safety, efficiency, and outcomes for civil engineering projects.