资源描述：
《The Algorithm Study of Sensor Compensation in MWD Instrument Based on Genetic Elman Neural Network》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。
1、ThirdInternationalSymposiumonIntelligentInformationTechnologyandSecurityInformaticsTheAlgorithmStudyofSensorCompensationinMWDInstrumentBasedonGeneticElmanNeuralNetworkJuLi-li,WangXiu-fang,MaSaiWeiChun-mingInstituteofElectricalandInformationEngineeringDirectionalWellCom
2、panyDaqingPetroleumInstituteDagangOilfieldDaqing,ChinaTianjin,ChinaAbstract—InordertoimprovethemeasurementprecisionandstabilityofMWDInstrument,wecreateElmanneuralnetwork2ConstructionofElmanneuralnetworkmodelmodelandutilizeself-adaptivegeneticalgorithmtooptimizeElmannet
3、workstructureisshowninFig.1,whichweightsthresholdvalueoftherightofElmannetwork,whichconsistsoftheInputLayer,theHiddenLayer,theOutputovercomesthedisadvantagesoftraditionalmethod,suchasLayerandtheContext.Thehiddenlayerunitsareusedtotrainingforalongtime,easytofallintoloca
4、loptimalsolution.rememberthehiddenlayerunitsbeforethetimeoftheSimulationresultsshowthattheerroraccuracyincreases3outputvalue,itcanbeconsideredasadelayoperator.ordersofmagnitude,comparedwithElmannetwork,theFeedforwardconnectionscanbeamended,whilethecompensationeffectisv
5、erystable.recursivepartisfixedandcannotbeamended,ThehiddenKeywords-Elmannetwork;Adaptivegeneticalgorithm;layerneuronsaretansig,andtheoutputlayerarepurelincompensation;MWDInstrumentneurons.Setthenetworkinputlayerhavernodes,thehiddenlayerandcontextunitshavennodesrespecti
6、vely,theoutput1Introductionlayerhavemnodes.Mathematicalmodelofthenetworkis[7]MWDinstrumentisanindispensablemeasurementasfollows:equipmentinmoderndirectionaldrilling,itcanprovidereal-time,reliable,accuratemeasurementofazimuthdeviationandsomeotherparameterinformationinth
7、ecaseofcontinuousdrillingprocess.Usuallyworkinginthecomplexenvironmentofhightemperatureandpressure,itisvulnerabletotheimpactofenvironmentalfactorssuchasWhere:uasthenetworkinput,xasthehiddenlayeroutput,temperature,resultedindecreasedmeasurementaccuracy,xasthecontextunit
8、output,yastheoutputforthecpoorstabilityandsomeotherproblems.WiththeI1I2increasinglydemandofthesensors,weneedtocompens
