The Research On Aero-engine Vibration Monitoring And Fault Diagnosis System

Our country is vigorously developing the aero-engine industry at present and the development of aero-engine is at a critical period. We need a lot of running test in order to meet the demands of the aero-engine development. But the whole-body vibration is the key problem that restricts the aero-engine development and greatly increases the cost of running test. Our country has no advanced and mature aero-engine vibration monitoring and fault diagnosis system that can provide effective technical support for the whole-body vibration troubleshooting. So this paper did the research on vibration monitoring and fault diagnosis system for the double rotor aero-engine.At first,this paper deeply analyzed the needs of the system. Then, this paper designed the hardware and software structure of the system. The data acquisition program is designed for the double rotor aero-engine. The rotation speeds of internal rotor and external rotor are collected and two key phase signal are generated to trigger the sample of the corresponding vibration measuring point. At the same time of acquisition, the indexes of vibration signal are extracted and the data acquisition program determines if it needs to alarm or not. A combination alarm method is used to ensure the effectiveness of the alarm and this paper also did the research of the strategy of data storage and management in this system.In order to realize the intelligent diagnosis of aero-engine vibration faults, neural network and expert system are used in this system. Under the circumstances that lack of data samples in the early stage, we add common diagnosis rules into the expert system and use the expert system to realize the fault diagnosis function. After a sufficient number of data samples are accumulated, we use the data samples to train neural networks and improve the knowledge base of expert system. Then the fault diagnosis of aero-engine will be more effective. The neural network module in this system has used a fault diagnosis method that based on neighborhood rough sets and parallel neural networks to solve the problem of selecting the neural network input vector. And this paper had tested the method by using the experimental data and found that appropriate neighborhood radius and parallel neural networks can optimize the structure of neural network and also ensure the accuracy of fault diagnosis. The expert system can be self-learning in the fault diagnosis process and constantly improve the diagnostic rules. Finally the expert system will achieve a better diagnosis of the aero-engine vibration faults.On the basis of the above research, this paper had completed the development of aero-engine vibration monitoring and fault diagnosis system. In this system, a variety of vibration signal analysis methods are used to monitor and analyze the vibration signal of the aero-engine. And this system realizes the visual configuration settings of the measuring point layout. We have tested this system on the aero-engine rotor fault simulation. The test results show that the system is running in good condition and the system can meet the demands of the aero-engine test rig vibration monitoring and fault diagnosis.