Research On Fault Diagnosis Method Of Rolling Bearing Based On Mixed Features

Rolling bearings,as one of the basic components in rotating machinery,are often used to support rotating parts.Their role is to reduce the friction coefficient of the equipment during the transmission process.The rolling bearing is also a high-frequency fault component at the same time and the quality of the rolling bearing has a great influence on the reliable operation of the mechanical equipment.Rolling bearings as rotating machinery are widely used in industrial production and are also weak components in mechanical equipment.Therefore,it is of great significance for the diagnosis of rolling bearings.For the problem of fault diagnosis,the rolling bearing is taken as the research object in this thesis and the thesis mainly studies and analyzes from the two aspects of fault feature extraction and fault diagnosis of rolling bearing.The main content is divided into the following sections:Firstly,the basic physical mechanism of rolling bearings is introduced.Based on the vibration mechanism,the main failure modes of rolling bearings are analyzed.And the natural frequency of rolling bearings and the frequency of fault characteristics is derived theoretically.Secondly,considering that the original vibration signals of rolling bearings are non-stationary and non-linear,there are limitations when applying traditional signal processing methods.Therefore,time-frequency analysis method is used for signal processing.A mixed feature extraction method based on Complete Ensemble Empirical Mode Decomposition with Adaptive Noise(CEEMDAN)combined with Fuzzy Entropy(FE)is proposed.This method extracts the time-domain characteristic index and frequency-domain characteristic index of the original vibration signal of the rolling bearing,and the original signal is processed by CEEMDAN.It can effectively suppresses the mode aliasing;then use the Intrinsic Mode Function(IMF)component obtained by the decomposition to calculate the fuzzy entropy value as the time-frequency domain feature;then combine each feature parameter into a mixed feature vector set,which can accurately reflect the features of bearing in different states.Finally,for the fault diagnosis of the rolling bearing,the mixed feature extraction method proposed in this thesis is combined with Generalized Regression Neural Network(GRNN).The smoothing factor parameter of GRNN is optimized by the Grey Wolf Optimizer(GWO)algorithm.Based on the this,a fault diagnosis model for rolling bearing is constructed.The data from the Western Reserve University data center are used for experimental analysis in this thesis and the experimental results show that the diagnosis model of the optimized GRNN has higher accuracy than the original GRNN.The validity and feasibility of the method of this thesis are proved.