Application of the metaheuristic algorithms to quantify the GSI based on the RMR classification

Area of Science:

• Geotechnical Engineering
• Rock Mechanics
• Computational Intelligence

Background:

• Accurate rock mass classification is crucial for earth sciences applications.
• The Rock Mass Rating (RMR) is quantitative, while the Geological Strength Index (GSI) is qualitative.
• Quantifying GSI from RMR is challenging due to limitations in existing empirical models.

Purpose of the Study:

• To develop accurate and generalizable predictive models for quantifying GSI directly from RMR.
• To compare the performance of metaheuristic optimization algorithms in developing GSI-RMR relationships.
• To enhance the accuracy and applicability of GSI estimation in engineering design.

Main Methods:

• Analysis of data from fourteen rock types.
• Development of five mathematical GSI-RMR equation types (linear, power, exponential, polynomial, logarithmic) using Particle Swarm Optimization (PSO), Simulated Annealing (SA), and Grey Wolf Optimization (GWO).
• Assessment of models using statistical indicators (R², RMSE, MAE, ASE, MAPE, MARE) and graphical analyses.

Main Results:

• Optimized GSI-RMR equations were derived using PSO, SA, and GWO.
• The SA-derived equation showed superior performance based on initial evaluations.
• Multi-criteria evaluation revealed that GWO and PSO models also offered advantages in specific aspects, highlighting the value of diverse optimization approaches.

Conclusions:

• The proposed metaheuristic-based GSI-RMR models offer improved accuracy and generalizability over traditional methods.
• These models facilitate more reliable rock mass strength parameter estimation and support system design.
• The study underscores the importance of multi-criteria evaluation for selecting optimal predictive models in geotechnical engineering.