The Study On Prediction Method Of Gas Emission Amount AQPSO-RBF In Fully Mechanized Mining Face And Its Application

With the emergence of high strength low gas exploitation of mines,all kinds of gas disaster problems are becoming more and more serious.Grasping the method of gas emission prediction,and realizing precision prediction of gas emission is the basis of grasping the downhole gas emission rule and gas prevention and control,which is of great significance to the gas disaster governance and the safety of underground work personnel.In terms of the multi-element and non-linear influencing factors of gas emission and the poor accuracy of neural network prediction issues.Exampled by a certain gas test mine of China Coal Group in Shanxi Province as the research object,this paper analyzes of the interaction between the influencing factors and the gas emission based on natural geological and mining factors,and concludes that there are many influence factors of gas emission,and their effect degree is different,the influence factors influence each other.There is a complex and nonlinear relationship among the influencing factors,and with the development of coal mining,the influencing factors show a trend of constant influence and change.In view of the characteristics of gas emission quantity,such as many influencing factors,varying degrees of action and overlapping information,a method for extracting prediction indexes based on factor analysis was proposed after studying and analyzing many prediction methods.In view of the common factor of gas emission quantity,effective information is extracted to the greatest extent by rotating the factor without reducing the original information,so as to achieve the simplified dimensionality reduction of the original variables,reduce information overlap,and improve the validity of the data.In view of the mine gas emission quantity and influencing factors between multivariate nonlinear characteristics such as degeneration in time,building quantum particle swarm optimization using adaptive radial basis function neural network predictive model of gas emission through the research analysis,the particle swarm not only realized the nonlinear prediction indexes for identification and global optimization,but also corrected and optimized the parameters of radial basis function(RBF),so as to produce the optimal results of gas emission prediction indexes from the nonlinear to linear output identification.With the advance of the working surface,the neural network can optimize the local and global nonlinear index information,and finally output the linear results,so as to realize the dynamic prediction of gas emission.Factors are summarized based on the factor analysis,the adaptive quantum particle swarm and radial basis function(RBF)neural network and other related theories,taking MATLAB software as a platform,using the built-in toolbox,taking GUI as a graphical interface,a dynamic gas emission prediction software based on predictive factor analysis and adaptive quantum particle swarm optimization is designed and developed.The software includes the module of factor analysis to select the prediction index and the module of adaptive quantum particle swarm optimization for prediction of gas emission in radial basis function.Through comparative analysis,the software has the characteristics of convenient operation,strong data processing ability and high prediction accuracy,which can meet the real-time prediction requirements of coal mine gas emission.