Mining Technology Evaluation for Steep Coal Seams Based on a GA-BP Neural Network

Area of Science:

• Mining Engineering
• Artificial Intelligence in Mining

Background:

• Guizhou Province's steep coal mines require renovation for safety and longevity.
• Appropriate mining technology selection is critical for successful mine renovation.

Purpose of the Study:

• To develop a framework for selecting optimal mining technologies for steep coal mines.
• To evaluate the performance of artificial neural networks, specifically the genetic algorithm-optimized backpropagation (GA-BP) network, for predicting mining methods.

Main Methods:

• Spearman correlation analysis to identify key input variables (geological conditions, working face parameters).
• Orthogonal experiments to optimize neural network hyperparameters.
• Testing of five backpropagation (BP) algorithm variants and a genetic algorithm-BP (GA-BP) model.
• Evaluation using coefficient of determination (R²) and mean squared error (MSE).

Main Results:

• The variable step-size algorithm with a momentum term (VSS + MT) was optimal for the BP network.
• The GA-BP network achieved higher prediction accuracy (90% success rate, R²=0.85, MSE=0.04) compared to the standard artificial neural network (70% success rate, R²=0.79, MSE=0.06).
• Industrial application at Zhong-Yu coal mine recommended fully mechanized mining, validating the model's reliability.

Conclusions:

• The GA-BP neural network is a superior model for selecting mining technologies in steep coal mines.
• The developed evaluation framework and model are reliable for practical industrial applications.
• Optimized technology selection ensures mine safety, operational harmony, and extended lifespan.