Research on a Deeply Integrated Model for Structural Optimization in Coal Spontaneous Combustion Temperature...

Area of Science:

• Mining Engineering
• Artificial Intelligence
• Materials Science

Background:

• Coal spontaneous combustion poses significant safety risks in mining operations.
• Conventional temperature prediction methods struggle with generalization and adaptability.
• Need for robust, real-time monitoring systems in coal mines.

Purpose of the Study:

• To develop an advanced protocol for coal spontaneous combustion temperature prediction.
• To overcome limitations of fixed architectures and poor transferability in existing methods.
• To enhance the accuracy and reliability of early-warning systems for coal mines.

Main Methods:

• Utilized a Sparrow Search Algorithm (SSA)-optimized Convolutional Neural Network (CNN)-Long Short-Term Memory (LSTM)-Attention framework.
• Employed CNN for spatial feature extraction and LSTM for temporal dependency capture.
• Integrated an attention mechanism to highlight critical temperature phases and features.
• SSA optimized network depth and hyperparameters for dynamic data adaptation.

Main Results:

• The proposed model demonstrated significantly higher predictive accuracy on homogeneous datasets.
• Achieved robust generalization performance across heterogeneous datasets from different mining conditions.
• The framework proved well-suited for real-time coal mine temperature monitoring.
• Indicated superior performance compared to conventional prediction methods.

Conclusions:

• The SSA-CNN-LSTM-Attention framework offers a superior approach for coal spontaneous combustion temperature prediction.
• The protocol provides a dynamic and adaptable solution for diverse mining environments.
• This advancement is crucial for improving safety and implementing effective early-warning systems in coal mines.