Prediction of Mine Pressure Appearance Based on Deep Learning

With the continuous development of the economy, coal mining has entered the stage of deep mining. The mining process of the working face can cause significant deformation and fracture of the overlying strata, which may cause coal mine dynamic disasters such as rock burst. As a new monitoring technology, distributed optical fiber (DOF) sensing system is applied to the deformation monitoring of overlying strata during coal mining, and it also indirectly represents the ground pressure appearance of the working face. In order to accurately predict the occurrence of mining pressure in the working face, this paper focuses on the physical simulation experiment of similar materials for DOF monitoring of overlying rock deformation and working face pressure and uses deep learning algorithms to complete the research work on mining pressure prediction in the working face. This article constructs a CNN-LSTM prediction model for rock pressure manifestation, which combines the advantages of both CNN and LSTM models to extract features from sample data. It can capture the correlation between fiber frequency shift data and rock pressure manifestation and predict the position of rock pressure manifestation in the working face. This model has higher prediction accuracy and stronger prediction ability compared to CNN and LSTM models. The prediction model has certain reference value, providing a new technical approach for predicting the location of mining pressure manifestation and roof management in working faces based on DOF monitoring.