大金多联机的故障分析及维修保养摘要摘要：随着全球能源短缺问题的日益严重，如何提高能源效率已经成为一个主要的研究方向。建筑能耗占总能耗的30%以上，空调系统是建筑能耗中的主要耗能设备。当空调系统遇到障碍物时，系统的能耗会大大增加，因此对空调系统进行故障诊断是非常必要的。对于可变流量多连接系统，其有效运行的关键是系统的制冷剂充注水平。然而，由于系统的复杂性，很难做到这一点。因此，正确快速地确定制冷剂充注量对系统的自动控制具有重要意义。提出了一种结合支持向量机(SVM)、最大相关最小冗余度(RMR)和小波去噪的SVM复合模型。小波去噪用于改善采集数据的数据质量，提高SVM模型的泛化能力，m.RMR用于特征提取。通过这种方法，我们得到了最合适的特征序列。之后，进行特征之间的相关性分析，以支持进一步的特征选择。最后，通过组合模型选择特征子集1B作为最优特征子集。与使用完整特征集的分类精度相比，分类精度仅降低2.14%。由于模型性能下降最低，使用的数据大大减少。不再需要某些传感器，从而实现了经济效益和模型性能之间的平衡。该组合模型的诊断结果表明，在特征选择过程中，应认真考虑特征与目标类别的关系以及特征本身的关系。关键词：空调系统故障诊断；多联机；维修保养Absrtact:Withtheglobalenergyshortagebecomingmoreandmoreserious,howtoimproveenergyefficiencyhasbecomeamajorresearchdirection.Buildingenergyconsumptionaccountsformorethan30%ofthetotalenergyconsumption.Airconditioningsystemisthemainenergyconsumingequipmentinbuildingenergyconsumption.Whentheairconditioningsystemencountersobstacles,theenergyconsumptionofthesystemwillgreatlyincrease,soitisverynecessarytodiagnosethefaultsoftheairconditioningsystem.Forvariableflowmulti-connectionsystem,thekeytoitseffectiveoperationistherefrigerantchargelevelofthesystem.However,duetothecomplexityofthesystem,itisdifficulttodothis.Therefore,itisofgreatsignificancefortheautomaticcontrolofthesystemtocorrectlyandquicklydeterminetherefrigerantcharge.ASVMcompositemodelcombiningsupportvectormachine(SVM),maximumcorrelationminimumredundancy(RMR)andwaveletdenoisingisproposed.WaveletdenoisingisusedtoimprovethedataqualityofcollecteddataandthegeneralizationabilityofSVMmodel.mRMRisusedforfeatureextraction.Throughthismethod,wegetthemostsuitablefeaturesequence.Afterthat,correlationanalysisbetweenfeaturesiscarriedouttosupportfurtherfeatureselection.Finally,thefeaturesubset1Bisselectedastheoptimalfeaturesubsetthroughthecombinationmodel.Comparedwiththeclassificationaccuracyusingthecompletefeatureset,theclassificationaccuracyisonlyreducedby2.14%.Sincetheperformanceofthemodelisthelowest,thedatausedisgreatlyreduced.Somesensorsarenolongerneeded,thusachievingabalancebetweeneconomicbenefitsandmodelperformance.Thediagnosisresultsofthecombinedmodelshowthattherelationshipbetweenfeaturesandtargetcategoriesandtherelationshipbetweenfeaturesthemselvesshouldbecarefullyconsideredintheprocessoffeatureselection.Keywords:faultdiagnosisofairconditioningsystem;Multipleconnections;Maintenance目录目录................................................................3一、绪论...........................................................41.1系统故障诊断研究意义.........................................41.2故障诊断历史和研究现状.......................................51.2.1故障诊断发展历史.......................................51.2.2多联机系统的故障诊断发展...............................6二、多联机故障诊断................................................102.1实验描述与数据处理..........................................102.1.1实验描述...........
