The Research On The Application Of Support Vector Machine And Data Fusion In The Prediction Of The Coal And Gas Outburst

Coal and gas outburst is one of the most danger disasters in the disasters of mine gas,it isalso the most common disaster, the prevention and control of the disasters of mine have becomethe priority among priorities, as the safety work in mine.On the basis of the thoroughunderstanding of occurrence mechanism in gas outburst.To realize the accurate prediction for gasoutburst danger has become the main technical mean in the prevention of gas disasters.For thispurpose, Aiming at the need of safety in production, with the Influence of mine gas outburstcontrolling factors as the object of study, with the gas outburst prediction as the purpose,toresearch the application for the prediction for gas outburst with Support Vector Machine andData Fusion Comprehensive.Through to study on the theory and method of Support Vector Machine and Data Fusion inthis paper,to construct technical framework for data fusion of mine gas based on Support VectorMachine, putting forward hierarchical fusion model the forecast of gas outburst, determiningeach layer of the fusion algorithm and the completed main function. In the fusion of feature layer,felecting SVM as algorithm for the fusion of feature layer,to build the prediction model of coaland gas outburst Based on SVM. First, analysising data sources of prediction system,using theanalysis method of gray correlation to Choice the index setrisk of characteristics for mine gasoutburst, According to thedegree of gray correlation, to Determine the main control factorInfluencing mine gas outburst. Taking the main control factor as characteristics that is Input dataof the forecast system,at the time; Using these characteristics index of Support Vector Machineto training Simulatly,and Choicing the right kernel function of Support Vector Machine, Usingthe method of adjusting step grid search and k-Fold cross validation to optimizate the parameterfor Support Vector Machine.The experimental results show that,it can get the good results afterthe fusion of the feature layer, but based on the inherent drawback of SVM, Put forward to useevidence theory of D-S as a method of decision fusion,Using the Predictors of outcome for SVMand several typical index as Body of evidence for fusion of decision layer with evidence theory,Compositing hierarchical fusion architecture of Feature layer and decision layer, IncreasingReliability for policy decision. Last, by choosing some historical highlights data of mining, toverificate prediction model hierarchical fusion in this paper. Theresults show that the predictionresults are more accurate after the fusion of decision layer, Indicating this programme has thegood feasibility and validity.