Suitable For Bearing Fault Diagnosis Model With Noise Reduction Method

Aeroengine fault diagnose is a systems engineering which contains many subjects and come down to extensive theoretic knowledge. In our country, this technology's research and application is in the initial stages. Aeroengine is serviced after accident or maintained timely. It will not only waste manpower and material resources, but also bring new fault in the examining and repairing courses. Early aeroengine fault diagnose will solve these problems.Wavelet analysis is a new signal processing method. It is known that wavelet transformation has good performance in local time domain and frequency domain. So it becomes the fastest developed signal analysis method in time-frequency analysis and successfully applied in large-scale mechanical fault diagnose. It provide a effective way for nonstationary signal processing, signal filtering, signal-noise separating and feature extracting.Bearing is a common part in aeroengine which works in high rotate speed, high load and high temperature circumstance. So it becomes the one of easily destroyed part. Aeroengine is one of the large-scaled rotating machines has many faults with bearings. The running state of bearings affects the whole machine's capability.The inspecting signal is interfered severely by the measuring noise which come into being in data collection or the noise which produced by surrounding things when the aeroengine works in outward. The noise brings difficult in aeroengine's healthy state monitoring.Consequently, this article is relied on the background of aeroengine fault diagnose and do the research on the bearing which is the most important part of rotating machine. It analyses where the noise comes form and the models of noise. This article analyses why the bearing fault comes out and designs the bearing fault model by researching on bearing fault reasons and vibrational features. Based on learning wavelet analytic method, this article does the research on wavelet denoising arithmetic of data processing which aims at the stochastic noise and brings out improved denoising arithmetic of mix threshold denoising method and optimizing threshold based on correlation. The denoising effect is improved by the new methods through the experimental simulations and actual fault data denoising. The result indicates that the fault feature remains more integrally after denoising and the new methods fit the bearing for early fault diagnose. The new methods brought out by this article provide technical groundwork for the aeroengine state monitoring and fault diagnose.