Gas explosion early warning method in coal mines by intelligent mining system and multivariate data analysis.

Area of Science:

• Mining Engineering
• Data Science
• Disaster Prevention

Background:

• Gas explosion disasters pose a significant threat to coal mine safety.
• Existing early warning systems often lack the accuracy and speed required for effective disaster prevention.
• Intelligent mining systems generate vast amounts of real-time data crucial for improving safety.

Purpose of the Study:

• To develop and validate a highly accurate and rapid early warning model for coal mine gas explosions.
• To leverage multidimensional data from intelligent mining systems for enhanced safety management.
• To compare the performance of a random forest model against a support vector machine (SVM) model.

Main Methods:

• Decomposition of the coal mine disaster system into disaster-causing factors, environments, and vulnerable bodies.
• Establishment of an early warning index system for gas explosions.
• Development and optimization of a random forest classification model using real-time data (mine safety monitoring, personnel positioning, video surveillance).
• Comparative analysis with a support vector machine (SVM) model using Matlab software.

Main Results:

• The optimized random forest model achieved 100% accuracy in predicting gas explosion disasters.
• The support vector machine (SVM) model achieved only 75% accuracy.
• The random forest model demonstrated lower model error and relative error compared to SVM.
• Case studies confirmed the model's applicability and high performance.

Conclusions:

• The optimized random forest model provides a highly effective and accurate method for early warning of coal mine gas explosions.
• Integrating intelligent mining data with advanced algorithms like random forest significantly enhances coal mine safety management.
• This approach offers a novel and powerful tool for proactive disaster prevention in mining operations.