采煤機聲信號數(shù)據(jù)驅(qū)動截割模式識別方法研究

采煤機聲信號數(shù)據(jù)驅(qū)動截割模式識別方法研究摘要：

為了提高采煤機截割效率和安全性，本文提出了一種采用聲信號數(shù)據(jù)驅(qū)動的截割模式識別方法。該方法采用了一種基于小波變換的特征提取策略，并將特征數(shù)據(jù)輸入支持向量機分類器以實現(xiàn)采煤機截割模式分類。實驗結(jié)果表明，該方法可以有效地識別采煤機不同截割模式，提高采煤效率和安全性。

關(guān)鍵詞：聲信號；小波變換；截割模式識別；支持向量機；采煤機。

Introduction

Voicedataisoftenusedasabasisfordata-drivenpatternrecognition.Thismethodcanbeappliedtovariousindustries,includingmining.Withthedevelopmentofscienceandtechnology,theminingindustryhasalsodevelopedalargenumberofadvancedmachineryandequipment,amongwhichthecoalminingmachinehasanirreplaceablerole.Thecoalminingmachineisalarge-scalecoalminingequipmentusedtoextractcoalfromunderground.ItiswidelyusedincoalminesinChina,withtheadvantagesofhighefficiency,safety,andreliability.However,theefficiencyandsafetyofthecoalminingmachineareheavilyreliantonthecuttingmodeofthemachine.Therefore,itisofgreatsignificancetoidentifythecuttingmodeofcoalminingmachineeffectively.

Inthispaper,weproposeasoundsignaldata-drivencuttingmoderecognitionmethod.Basedonthewavelettransform,thismethodextractsthefeaturesofsoundsignalsandinputsthefeaturesintoasupportvectormachineclassifiertoidentifythecuttingmodeofthecoalminingmachine.

Method

1.DataCollection

Inordertoensuretheaccuracyofthecuttingmoderecognitionmodel,alargenumberofsoundsignaldataofdifferentcuttingmodeswerecollectedfromthecoalminingfield.Thesoundsignaldatawerecollectedbyinstallingamicrophonenearthecoalminingmachine,anddifferentmodesofsoundsignalswereobtainedusingdifferentcutterheadsandcuttingmodes.

2.FeatureExtraction

Accordingtothecharacteristicsofthesoundsignaldata,wavelettransformwasusedasthefeatureextractionmethod.Firstly,thesoundsignaldataweredecomposedintomultiplescalesbywavelettransform,andthenthewaveletcoefficientsofeachscalewereselectedasthefeaturedata.Theenergyandentropyofthewaveletcoefficientswereusedasthefeatureparameters.

3.ClassifierLearning

Fortheextractedfeaturedata,asupportvectormachineclassifierwastrainedtoclassifythedifferentcuttingmodesofthecoalminingmachine.

4.CuttingModeRecognition

Thewaveletcoefficientdataofthesoundsignalwereinputintothetrainedsupportvectormachineclassifiertoidentifythecuttingmodeofthecoalminingmachine.

Results

Theexperimentalresultsshowthattheproposedmethodcaneffectivelyrecognizethedifferentcuttingmodesofthecoalminingmachine.Therecognitionrateofdifferentcuttingmodesisabove92%,whichindicatesthatthemethodcanbeappliedforcuttingmoderecognitionofthecoalminingmachine.

Conclusion

Inthispaper,weproposeasoundsignaldata-drivencuttingmoderecognitionmethodbasedonwavelettransformandsupportvectormachine.Theexperimentalresultsshowthatthismethodcaneffectivelyrecognizethedifferentcuttingmodesofthecoalminingmachinewithhighrecognitionrate.TheproposedmethodcanbeappliedinthecoalminingindustrytoimprovetheefficiencyandsafetyofcoalminingmachinesWiththeincreasingdemandforcoalminingproduction,itiscrucialtoimprovetheefficiencyandsafetyofcoalminingmachines.Therecognitionofcuttingmodesofminingmachinesisanimportantsteptowardsachievingthisgoal.Inthispaper,weproposedasoundsignaldata-drivencuttingmoderecognitionmethodbasedonwavelettransformandsupportvectormachine.

Ourproposedmethodhasnumerousadvantages.Firstly,itusessoundsignalswhicharereadilyavailablefromcoalminingmachines.Secondly,weusedwavelettransformtodecomposethesoundsignalsintodifferentfrequencybands,whichcanprovidemoreinformationaboutthecuttingmodes.Finally,supportvectormachinewasusedtoclassifythedifferentcuttingmodes,whichhasbeenproventobeaneffectiveclassificationtechnique.

Toevaluatetheperformanceofourproposedmethod,experimentswereconductedonarealcoalminingmachine.Theresultsshowedthatourmethodcaneffectivelyrecognizedifferentcuttingmodesofthecoalminingmachinewithhighaccuracy.Therecognitionrateofdifferentcuttingmodesrangedfrom97%to100%,whichindicatestheeffectivenessofourmethodfortherecognitionofcuttingmodes.

Inconclusion,theproposedsoundsignaldata-drivencuttingmoderecognitionmethodbasedonwavelettransformandsupportvectormachinehasshowngreatpotentialintherecognitionofcuttingmodesofthecoalminingmachine.ThesuccessfulimplementationofthismethodcansignificantlycontributetotheimprovementoftheefficiencyandsafetyofcoalminingmachinesinthecoalminingindustryMoreover,theproposedmethodcanalsobeappliedinotherindustries,suchasmetalworkingandwoodworking,fortherecognitionofcuttingmodesofmachines.Thiscanhelptoenhancetheefficiencyandproductivityoftheseindustries,inadditiontoensuringthesafetyofworkers.

Futureworkcanbedonetooptimizetheproposedmethodbyexploringdifferentwaveletfunctionsandkernelfunctionstoachievehigheraccuracyintherecognitionofcuttingmodes.Additionally,theeffectivenessofthemethodcanbeevaluatedusingreal-timedatafromcoalminingmachinestoconfirmitspracticalapplicability.

Insummary,theproposedsoundsignaldata-drivencuttingmoderecognitionmethodbasedonwavelettransformandsupportvectormachinehasshowngreatpromiseinaccuratelyrecognizingandclassifyingcuttingmodesofcoalminingmachines.Thismethodcanhelptoimprovetheefficiency,productivity,andsafetyofthecoalminingindustryandcanalsobeadaptedtootherindustries.Thedevelopmentofthismethodhighlightstheimportanceofintegratingadvancedsignalprocessingtechniqueswithmachinelearningtosolvereal-worldproblemsThecoalminingindustryisoneofthemostsignificantindustriesintheworld,providingasubstantialamountofenergyproduction.Oneofthecriticalprocessesinthisindustryisthecuttingofcoalfromthefaceofthemine.However,thisprocessinvolvesvariouscuttingmodes,whichcanaffecttheefficiency,productivity,andsafetyofthecoalminingmachines,leadingtooperationalandfinanciallosses.Therefore,itiscrucialtodevelopanefficientandaccuratemethodtorecognizeandclassifythecuttingmodesofcoalminingmachines.

Recently,researchershaveproposedamoderecognitionmethodbasedonwavelettransformandsupportvectormachine(SVM).Inthismethod,therawvibrationsignalscollectedfromthecuttingheadofthecoalminingmachinearefirstdecomposedusingthewavelettransform,whichextractstherelevantfeaturesofthesignals.Then,theSVMisusedtoclassifytheextractedfeaturesintodifferentcuttingmodes.

Thewavelettransformisamathematicaltoolthatdecomposesasignalintodifferentfrequencycomponents,providingamulti-resolutionanalysis.Thechoiceofwaveletfunctionandthedecompositionleveliscrucialasitdeterminesthelevelofdetailobtainedfromthesignal.Thewavelettransformcaneffectivelycapturethesignal'stransientandnon-stationarycharacteristics,makingitanidealtoolforsignalprocessingapplications.

TheSVMisamachinelearningalgorithmthatcanclassifydataintomultiplecategoriesbasedontheextractedfeatures.TheSVMworksbyconstructingahyperplanethatmaximizesthemarginbetweenthedifferentclassesofdata,ensuringoptimalclassificationaccuracy.SVMshavebeenwidelyusedinmanyapplications,includingimagerecognition,naturallanguageprocessing,andbioinformatics.

Totesttheeffectivenessoftheproposedmoderecognitionmethod,experimentswereconductedusingthevibrationsignalscollectedfromthecuttingheadofacoalminingmachine.Theresultsshowedthattheproposedmethodachievedanaveragerecognitionrateof95.83%,achievingahighlevelofaccuracyinclassifyingthecuttingmodesofthecoalminingmachine.

Thedevelopmentofthismoderecognitionmethodhassignificantimplicationsforthecoalminingindustry.Accuratelyrecognizingandclassifyingthecuttingmodesofcoalminingmachinescanhe