Research On Gas Area Forecast Model Based On Correlation Analysis Technology

The gas area prediction can guide the work of the gas outburst,reduce the accident rate caused by the gas,and promote production and development of mine.Based on the industry standard,user requirement,existing theories,research results at home and abroad and specific features of the mine,author proposes and establishes the gas area prediction model based on correlation analysis technology by statistics,operations research and neural networks.We get the gas content law in every gas area by the prediction model,then,can forecast the gas content by the law.The predictive values by the model are better and more effective.It can reduce the accident rate caused by the gas.According to experts and professional knowledge in the mine,the particular situation of a coal mine,factors which affect gas content in an area are divided into macro factors and micro factors.The linear relation between micro factors and the gas content of every measuring point is decided by the scatter diagram,the regression line equation fitted by the method of least squares,and the coefficient of determination.In order to reduce the complexity of gas area prediction model,factors with a high correlation is decided by the correlation coefficient,the redundancy factors are removed.The cumulative contribution rate is computed by the method of the principal component analysis.Then the cumulative contribution rate is compared with the threshold and we get the principal component that is most information on behalf of the original factors.Thus the low dimensional feature set is obtained and the complexity of the gas area prediction model is reduced.But it is powerless in the mutation and irregular situation.Thus,author builds a gas area forecast model based on the analytic hierarchy process by macro factors.Predicted results are corrected by the model and macro factors.The accuracy of prediction model is improved.Finally,by the particular situation of a coal mine,the paper evaluates forecast results on real datasets with the proposed method.The result shows the gas forecast model is effective and improves the accuracy of the gas data.It provides a strong favor and basis for the work of gas outburst.