Study On Adaptive Filter And The Application In Gearbox Feature Extraction

The diagnosis based on vibration analysis is an effective method for the gearbox fault diagnosis. The vibration signal was gathered during the operation of the gearbox .And the fault features that may occur was extracted from the vibration using appropriate methods. Thus the location of fault occurring was recognized. However, as excessive background noise when gearbox is running, signal carrying fault is flooded by noise, so that faults are not identified. In order to diagnose fault, collected signal is processed for one thing to enhance the Signal-to-Noise ratio. The paper will adopt adaptive noise canceling technology to process the vibration signal of gearbox, and extract fault features.The core technology of adaptive noise cancellation technology is the adaptive filter. On the basis of studying the basic adaptive filtering principle and the least mean square (LMS) adaptive filter, we analyze a large number of LMS algorithms and put forward a new LMS algorithm based on the correlation hyperbolic tangent function, referred to as CTanh-LMS algorithm. Then the simulation proved that, compared to other adaptive filter algorithms, this method has better noise cancellation capability. And the method is simple and has fast convergence speed. It meets the gearbox vibration signal denoising requirements. Then we analyzed the characteristics of gearbox vibration signals and its fault diagnosis methods.On the basis of simulation, the LMS algorithm based on the correlation hyperbolic tangent function is applied to gearbox fault feature extraction and achieves good result. It can pick-up characteristics of the fault and accurately recognize fault. The proposed algorithm not only solves the problems of gearbox fault feature extraction, but provides new methods and tools for adaptive filtering algorithm. It has an important theoretical and practical value.