Research And Application Of Intelligent Prediction Of Gas Concentration In Driving Face

Xima Coal Mine is a mine with high gas outburst.Gas is the main disaster causing coal mine safety accidents.Driving face is prone to gas accidents because of unclear geological environment,inadequate pressure relief,difficult extraction and lack of management.It is an urgent problem to study advanced dynamic prediction method and develop effective prediction management system to strengthen mine gas prevention and control and improve safety management level.Based on the actual monitoring data of gas concentration in 1216 cut hole and driving face of Xima Coal Mine,this paper studies the law of coal seam gas flow,gas emission and gas concentration distribution in driving face,analyzes the relationship between gas concentration in driving face and geological factors,mine natural factors and driving technology factors.The prediction performance of LSTM memory neural network with different prediction time is studied under different subsamples.Combined with Java language,Sqlite database,MATLAB and Tomcat,the development and deployment of gas intelligent management system in driving face are completed,and the automatic collection and storage,intelligent analysis and prediction of gas concentration monitoring data are realized.The results show that the driving face can be divided into gas folding area,gas unstable area and gas relatively stable area from the heading to the end of the heading.Under the action of multifactor fusion,the distribution of gas concentration in driving face has obvious dynamic nonlinear and spatiotemporal correlation.The prediction time of LSTM neural network is consistent with that of sub-samples.When sub-samples are 1 hour long,the prediction precision of 5 minutes ahead of time is the highest.The gas intelligent management system was applied in 1216 open cut and driving face of Xima Coal Mine.The gas concentration of T1,T2,T3 and T4 was predicted 5,15,20 and 10 minutes ahead,and the applicability and reliability of the model and the system were verified.The research results provide a new prediction technology and management method for the prevention of mine gas disaster.