Fault Detection And Reliability Research Of Compressor Performance Test System

Compressors are the core components of the refrigeration system and their performance is critical to the operation of the entire system.In order to improve the reliability of the system measurement and its operation in the process of testing the performance of the compressor,it needs reasonable system design,reliable sensor,stable control ability and fast and accurate fault monitoring as the guarantee.In this paper,based on the national standard GB/T5773-2004,a fault detection and reliability analysis method based on neural network is proposed,which has a very important significance and application value.First of all,this paper introduces the basic structure and principle of the single-stage positive displacement refrigerant compressors according to GB/T5773-2004.At the same time,the general test bed general construction rules and experimental process methods are explained.Secondly,the experimental methods and principles of the test bench are described.Because this article is mainly for two research objects: Performance Test Test Bench for Automotive Air Conditioning Compressor and Screw compressor performance test bench,the principles,formulas and flow chart of four methods used in the actual project are emphasized,ie the A method,the D method,the G method and the J method.Then,the structure and principle of two test benches of screw compressor performance test bench and automobile air conditioning compressor performance test bench are introduced.Finally,according to the actual project test bench data,we can see that the test bed of the control system has been able to well control the operating parameters to the required conditions.Control accuracy of the test system is well met the requirements of national standards.Secondly,we use the moving window stabilization method and the neural network modeling method to study the reliability of the compressor performance test system.We choose to use a commonly used method of steady-state discrimination,that is,the moving window stabilization method,to screen out the steady-state data in the process of running the system.Through the experimental verification,the window data judgment method has achieved good results.Then,for different compressor performance test systems,we use the most widely used BP neural network,to establish the corresponding neural network model.The established neural network model is used to detect the steady state data of the system after the system has faults.It is found that the neural network model can detect and diagnose the corresponding faults accurately and effectively,so as to detect the specific faults in time and take appropriate measures to quickly eliminate the corresponding faults to ensure the safe operation of the system.Finally,two practical examples of compressor performance test platform are used to verify the fault detection and diagnosis method proposed in this paper,ie the performance test platform of the automotive air conditioning compressor and the performance test platform of the screw compressor.For the performance test platform of automobile air conditioning compressor,three kinds of faults such as compressor oil shortage,condenser water shortage and temperature sensor failure are established according to the three types of faults that occur during the operation of the actual system.The corresponding neural network model is established.For the screw compressor performance test platform,the sensor deviation of the fixed deviation of the fault type is divided into two types.One of the types is the sensor deviation fault which participate in system control.The other one is the sensor deviation fault which do not participate in system control.Then,the primary neural network and secondary neural network are established.After verification of the experiment,the reliability analysis model based on neural network proposed in this paper has a good ability to detect the compressor performance test system.And it can not only detect the corresponding fault,but also can be diagnosed according to the number and type of the parameters which exceed threshold parameter.In conclusion,the experimental results show that the reliability analysis model based on neural network proposed by compressor performance test platform has better detection ability for compressor performance test system.