Research On The Application Of Gaussian Process In The Prediction Of Coal/Gas Outbursts

Coal mine production belongs to high-risk industry, coal gas outburst prediction andprevention of risk has been the significant scientific research topics which were concerned by themajor coal-producing countries in the world. Coal and gas outburst research shows managementmeans more than forecast means, gas outburst mechanism is complex, difficultly to standardize,led gas outburst happened though outburst gas detection index was normal. It has been highlyvalued to explore an effective forecast method under the coal mine field by many experts.Underground mine monitoring system has been collected and accumulated the massive data, thedata implicit knowledge of gas outburst risk prediction. This study introduces data mining to coalmine safety research field, according to the existing research results at home and abroad andconsideration of coal and gas outburst risk predictors based on data mining classificationtechnology, developing feature selection of the information of coal and gas outburst featureselection, constructing the coal and gas outburst risk prediction model. Combined coal minedisaster data characteristics and practical application background, it is feasibility andapplicability for Gaussian process method to solve the problem of coal and gas outburst.This paper was based on the comprehensive hypothesis of the coal and gas outburst. UsingLogistic regression method to establish the Logistic regression probability model of coal and gasoutburst and realize the first prominent feature selection. On this basis, the SVM forecastingmodel of coal and gas outburst is set up, comparatively analyze SVM learning precision indifferent attribute condition and finish the second prominent feature selection. On the basis ofanalyzing coal and gas outburst information of mining area of CNY, establishing gas outburstrisk model in Gaussian process a Gaussian process study estimated gas outburst risk model. Inorder to validate the advantage and the effectiveness of the model, compared with other methods,use the same data to study and prediction, and get the result.