Forecast Of Gas Emission In Working Face Based On Second Order Grey-Neural Network Combined Model

One of the worst mine casualties is Coal mine gas explosion accident,Prevent gas accidents in coal production is the most important task of enterprise security.How to accurately、effectively emission of gas and emission analysis and forecast,and make full use of the analysis and the forecast information to realize the safety management decision of coal enterpriser,is one of the important topics to be solved the safety production of mine enterprise.However, many factors affect the amount of gas emission, and the relationship between the factors complex,multi-component random disturbance, which emission law is a dynamic non-linear change process, more difficult to accurately predict the presence of mine gas emission.To simplify the complexity of the problem, many scholars will be considered among the factors affecting the amount and depth of burial of gas emission coal thickness coal seam pressure of methane linear relationship between the use of mathematical statistics theory, analogy, linear regression and other methods gas emission prediction less nonlinear gas emission quantity of the change process to be considered.Main methods of predicting the amount of gas emission, the mining sub-source statistics and forecasting method is static prediction, for now, more effective forecasting method for neural network and gray method, but a method used alone predict the effect is not too good, therefore, this paper attempts gray theory and BP neural network combining methods, Face Gas emission forecast.This paper describes the current status of research used some gas emission prediction methods. Secondly, several main factors affecting the amount of gas emission were introduced. Finally, the second order gray theory and BP neural network theory, a GNNM (2,1) combined forecasting model, and in the field to collect the Face Gas Emission monitoring data as the basis, through programming, using Matlab to achieve the right the face amount of gas emission forecasts showed GNNM (2,1) combined model prediction accuracy above90%, and more stable, can be used to predict the amount of gas emission face.