Research on coal and gas outburst risk prediction based on improved search algorithm optimized deep learning network.

Area of Science:

• Mining Engineering
• Geological Hazards
• Artificial Intelligence in Geosciences

Background:

• Deep coal mining increases surrounding rock pressure, elevating risks of gas accumulation and coal and gas outbursts.
• Accurate prediction of coal and gas outbursts is crucial for mine safety and operational continuity.

Purpose of the Study:

• To develop a robust and accurate prediction model for coal and gas outbursts.
• To enhance the safety and efficiency of deep coal mining operations through improved hazard prediction.

Main Methods:

• Data preprocessing using boxplots and interpolation, followed by correlation analysis to identify key influencing factors.
• Development of an initial prediction framework using Convolutional Neural Networks (CNN).
• Optimization of CNN hyperparameters with a Chaos Mapping and Levy Flight Improved Crow Search Algorithm (ICSA) to create the ICSA-CNN model.

Main Results:

• The ICSA-CNN model achieved higher accuracy in predicting coal and gas outbursts compared to baseline models.
• Evaluations using indicators and confusion matrices confirmed the model's superior predictive capabilities.
• The developed model exhibited enhanced robustness and generalization ability.

Conclusions:

• The ICSA-CNN model offers a significant advancement in coal and gas outburst prediction technology.
• This AI-driven approach improves safety and security in deep coal mining environments.
• The model's accuracy and reliability contribute to mitigating the risks associated with mining-induced geological hazards.