CapturingDominatingParametersofGasOutburstBasedonBayesianNetworkSUNLianying1,2&PENGSuping2(1InstituteofInformationTechnologyofBeijingUnionUniversity,Beijing100101,China;2TheStateKeylaboratoryofCoalResourcesandSafeMining,ChinaUniversityofMiningandTechnology,Beijing100083,China)Abstract:Gasoutburstisacombinedresultofmanygeologicandminingengineeringfactors.Intheprocessofmining,it’simpossibletocaptureallanalysisparameters.Analyzedthecharacteristicsofforecastinggasdata,amethodforcapturingthedominatingparameterofthegasoutburstbasedonBayesiannetworkispresented,andaparametersensitivitygradednetworkmodelofgasforecastisestablished,bywhichcandirectlyshowstructuralrelationsbetweenparametersandtheireffectinthequantitiesforecastofgasoutburst.Itsefficiencyisprovedbycase.Keywords:gasoutburst;dominatingparameter;Bayesiannetwork;sensitivityanalysis1IntroductionThegasoutburstisoneofthegreatestminesafetyproblemsfacedbycoalmine,whichcannotonlydamagethetunnelfacilitiesanddestroythemine’sventilationsystem,butalsokillagreatdealofpeople.So,effectivelycontrollingandforecastingthegasbecomesadominatingfactorofmine’snormalsafetyproduction.Recentyears,thegasforecastingtechnologyhasbecomeahotresearchpoint.Thetraditionalgasforecastmethodisthatstudiesthenaturallawofgaspresenceandoutburstintheviewofgeologyandbelievesthatthecoalseamsgasisaresultantofgeologicprocessandthattheformation,transportation,presenceandenrichmentofgashaveaclosingrelationwiththegeologiccondition[1,2].Thetechnologyindexsystemthatindicategasoutburstispresented.Withminingdeeply,thegeologicconditionsaremoreandmorecomplexandtherelationbetweengeologicindexesismoreandmorefuzzyanddifficulttodescribe.Inresentyears,bycombiningthebasictheoryofgasgeologywiththepresentartificialintelligenthighandnewtechnologiesprocessingthefuzzyrelationandnon-linearrelationbetweendata,agreatprogresshasbeenmadeinforecastingdisastersthroughusingneuralnetwork,fuzzymathematicsandgreytheory[3,4].Atthesametime,thetechnologyof3Sand3Dseismicexplorationarealsowidelyappliedinestimatinggaszone[5,6].Itcanbeconcludedthattheirsameresearchcharacteristicistocapturethegeologicandminingparametersrelatedtothegasoutburst,andthengetouttheforecastresultbyusingallkindsofanalysismeans(suchasstatisticmethod,artificialintelligence,spaceanalysis)bycomparingthesemethods.Aseveryoneknows,theconservation,transportationandoutburstofgasistheresultofthejointactionofminingtechnologies,geologicalconditions,coalseamsandcharacteristicsengineeringgeologyofrockmass.Duringthegasdisasteranalysis,peopletrytheirbesttogetparametersasmanyaspossible,but,intheactualresearchandproduction,it’simpossibleandinadvisabletocaptureallanalysisparametersowingtothelimitationofallfactors.So,it’snecessarytostudyoncapturingdominatingparameterofthegasoutburst,soastodeterminethedifferentsensitivitytogasoutburstamongallparametersanddirectactualproductionbymonitoringdataaccordingtoeachgrade.Anewmethodispresentedinpaper,westudiedthecharacteristicsofBayesiannetworkandgasforecastparameters,constructedatBayesiangasparametersnetworktocapturethedominatingparametersofgasforecast.Thismodelcanmakefulluseofallkindsofknowledgeandmakethequantitativeanalysisofgasoutburstcometrue.Itsactiveeffectivenessindefendinggasoutburstisprovedbycases.2BayesianNetwork2.1FrameofTheoryTheBayesiannetwork（BN）B＝<N,A,Θ>isapowerfulknowledgerepresen...