Azarshahr travertine compression strength prediction based on point-load index (I_s) data using multilayer perceptron

Area of Science:

• Geotechnical Engineering
• Artificial Intelligence in Mining
• Rock Mechanics

Background:

• Azarshahr County's geology is dominated by travertine, leading to extensive open-pit mining.
• Rock drillability and excavation resistance are critical mining factors linked to compressive strength.
• Traditional rock strength assessment methods lack precision, hindering reliable excavation planning.

Purpose of the Study:

• To develop an artificial intelligence model for enhanced prediction of Azarshahr travertine's compressive strength.
• To utilize the Multilayer Perceptron (MLP) for accurate strength estimation.
• To improve excavation methodologies and drillability assessments in travertine mining.

Main Methods:

• Compiled a database of 150 point-load index (I_s) tests on Azarshahr travertine.
• Developed and trained a Multilayer Perceptron (MLP) model using the compiled dataset.
• Validated model accuracy using Receiver Operating Characteristic (ROC) curves and Area Under the Curve (AUC) metrics.

Main Results:

• The MLP model achieved high accuracy in predicting axial and diametral compressive strength.
• R-squared coefficients of 0.975 were obtained for both axial and diametral strength predictions.
• An overall accuracy of 0.968 (AUC) demonstrates the model's effectiveness.

Conclusions:

• The MLP-based model accurately predicts travertine compressive strength from point-load index data.
• This AI approach offers significant improvements over conventional methods for rock strength analysis.
• The predictive model provides valuable insights for optimizing excavation planning and drillability in Azarshahr travertine mines.