Fault Diagnosis Of Aero-engine Rotor Based On The Wavelet Packet Neural Network

Aero-engine is a complex nonlinear systems. Because of its complex structure, it is great impact on the safe operation of the aircraft. Furthermore, because of its high temperature, high pressure, high load work environment, it requires a high degree of safety performance. For rotor system is one of the most important components of aero-engine, whether it can safe operate without fault, directly reflected to the engine system and even the entire aircraft safety. Therefore, it is very important to provide practical and theoretical basis for the aero-engine rotor fault diagnosis system.There are many intelligent fault diagnosis methods. Each method has its advantages and disadvantages. Firstly, wavelet packet is used to reduce the vibration signal. Then wavelet packet analysis is used to decompose denoised signal to four layer and reconstruct them. The ratio of the energy of 16 sub frequency signals of fourth layer and the energy of the original signal is calculated. This ratio is the input vector of the neural network. Next, BP neural network and probabilistic neural network is used to diagnosis fault. And an algorithm is proposed to fuse the results of two kinds of neural networks. The final results are more believable.After the fault diagnosis, reliability analysis is made. The method based on the operating state of the reliability analysis is used to analyze the reliability of aero-engine with small samples. Firstly, the wavelet domain, time domain and frequency domain characteristics of the signal are calculated. Then, kernel principal component analysis is used to extract the main features and the states subspace of different states are generated. Next, the main angle between the current state subspace and the normal state subspace is calculated. The cosine value of the main angle is used as the reliability of the current state. This method does not require a large number of samples as a support, reliability can be analyzed from only a few of the running state of the equipment.Finally, combined with the previous analysis, MATLAB graphic user interface design theory is used to develop rotor signal analysis and fault diagnosis system. The previous research result is showed to user well from the human-computer interaction. So, good human-computer interaction mode is achieved.