Valve Fault Diagnosis Of Variable Refrigerant Flow System Based On Neural Network Method

With the rapid development of China’s urbanization,there are more and more architectures using of HVAC system.VRF(Variable refrigerant flow)system is widely used in architectures for its advantages,such as flexible design,high comfort performance and reliability.In the process of actual operation,VRF system performance is easily affected by fault,fault diagnosis is one of the important means to ensure its normal operation.Due to the complexity of VRF system,there is a variety of equipment.If all kinds of faults are diagnosed,the research won’t be analyzed in depth,and the complexity of fault diagnosis will be increased.In thesis,the fault of electronic expansion valve and four-way valve in VRF system are diagnosed and detected.In this thesis,a valve fault diagnosis method of VRF based on the composite model of GA(gene algorithm)and neural network is suggested,by eliminating the transient data with slide window algorithm based on GLASS algorithm as the input data of the composite model.The performance of the model is judged by its correct recognition rate and time consuming of fault diagnosis.The results show that the optimized traditional BP neural network model by using RPROP algorithm can effectively detect the electronic expansion valve leakage fault which the BP model can’t,the fault diagnosis accuracy increases from 77.11% to 92.72%,and the diagnostic time decreases from 248 s to 60 s.In order to further improve the model’s performance,the data quality is in consideration,the slide window algorithm based on GLASS algorithm is used to eliminate the transient data,the data in steady state are selected to input the optimized BP neural network,the fault diagnosis accuracy increases to 96.01%,the diagnostic time decreases to 23 s,it is proved that this algorithm can effectively separates steady state and transient area,meanwhile,it keeps the trend fluctuation of fault information,and improves the reliability of the system.In order to reduce the data dimension for further improving the efficiency of the model,GA algorithm is used to extract the fault feature information intelligently,it shows that the composite model with the optimal feature subset variables not only can improve the fault diagnostic accuracy to 99.91%,but also drop the time consuming to 11 s,the optimal feature subset variables can characterize the abnormal changes when the valve failure happens,and effectively reduce the dimension of the input data,improve the convergence speed and the fault diagnostic accuracy.In this thesis,the valve fault diagnosis of VRF system based on the composite model is proposed to improve the accuracy of the fault diagnosis model and the ability of intelligent extraction of the feature information.It has a certain application value and significance in the intelligent fault detection and diagnosis of VRF system,which is worthy of further study.