大金多联机的故障分析及维修保养

大金多联机的故障分析及维修保养Absrtact:With the global energy shortage becoming more and more serious,how to improve energy efficiency has become a major research direction.Buildingenergy consumption accounts for more than 30%of the total energy consumption.Airconditioning system is the main energy consuming equipment in building energyconsumption.When the air conditioning system encounters obstacles,the energyconsumption of the system will greatly increase,so it is very necessary to diagnosethe faults of the air conditioning system.For variable flow multi-connection system,the key to its effective operation is the refrigerant charge level of the system.However,due to the complexity of the system,it is difficult to do this.Therefore,itis of great significance for the automatic control of the system to correctly and quicklydetermine the refrigerant charge.A SVM composite model combining supportvector machine (SVM),maximum correlation minimum redundancy (RMR)andwavelet denoising is proposed.Wavelet denoising is used to improve the dataquality of collected data and the generalization ability of SVM model.m RMR is usedfor feature extraction.Through this method,we get the most suitable featuresequence.After that,correlation analysis between features is carried out to supportfurther feature selection.Finally,the feature subset 1B is selected as the optimalfeature subset through the combination model.Compared with the classificationaccuracy using the complete feature set,the classification accuracy is only reduced by2.14%.Since the performance of the model is the lowest,the data used is greatlyreduced.Some sensors are no longer needed,thus achieving a balance betweeneconomic benefits and model performance.The diagnosis results of the combinedmodel show that the relationship between features and target categories and therelationship between features themselves should be carefully considered in theprocess of feature selection.Key words:fault diagnosis of air conditioning system;Multiple connections;Maintenance