Study On Fault Detection And Diagnosis Methods For Variable-air-volume Terminals

Due to the advantages of energy saving,high thermal comfort and easy expansion,variable air volume(VAV)air-conditioning systems have been more and more widely designed and applied in large public buildings.The VAV air-conditioning system is composed of the VAV air handling unit,the VAV terminals and the control system.The VAV terminals maintain the indoor temperature within its set range by controlling the air supply volume of the airconditioning room.In actual buildings,VAV terminals are numerous,scattered and installed in the ceiling,which makes the maintenance and repair of VAV terminals very difficult.Because it is difficult to find the fault of VAV terminals,there are usually a lot of VAV terminals faults in the VAV air-conditioning systems that have been running for many years.The faults of VAV terminals will affect the indoor thermal comfort,increase the system energy consumption and shorten the service life of the equipment.Fault detection and diagnosis of VAV terminals is the core technology to eliminate the negative effects of faults.Therefore,the research on the fault detection and diagnosis method of VAV terminals has important theoretical significance and engineering application value.In this study,a multiple faults detection and diagnosis method for VAV terminals based on parameter self-tuning model and random forest is proposed.The established air temperature parameter self-tuning model of the air-conditioning area is used to detect the faults of the VAV terminals.According to the operation data of VAV air-conditioning system,the chaotic particle swarm optimization algorithm is used to optimize the model parameters of the parameter selftuning model to minimize the deviation between the predicted value of the model and the measured value of the system,so as to improve the prediction accuracy and generalization ability of the self-tuning model of air temperature parameters in air-conditioning area.A multiple faults classifier of VAV terminals based on layered random forest is established.The multiple faults classifier of VAV terminals based on layered random forest is composed of expert rule fault diagnosis layer and random forest fault diagnosis layer.The expert rule fault diagnosis layer is used to diagnose the faults of the VAV terminals with simple fault symptoms,and the random forest fault diagnosis layer is used to diagnose the VAV terminals faults with complex fault symptoms.By integrating the fault diagnosis results of the expert rule fault diagnosis layer and the random forest fault diagnosis layer,the multiple faults of the VAV terminals can be separated and diagnosed.The correctness of the multiple faults diagnosis methods of VAV terminals is verified by using the operation data of VAV terminals in actual buildings.The verification results show that the proposed VAV terminals multiple faults diagnosis method has high fault diagnosis accuracy.The research results can provide theoretical methods and research experience for the development of VAV terminals fault detection and diagnosis systems which suitable for actual buildings.