Research On Fault Diagnosis Method Of Rolling Bearing Based On Improved Meemd

Rolling bearings are widely used in rotating machinery,and their running state will directly affect the reliability of the entire rotating machinery.Accurate and timely monitoring and diagnosis of its potential faults is of great significance to the safe operation of rotating machinery.Rolling bearing fault diagnosis has become an important branch of rotating machinery fault diagnosis after long-term development,and the fault diagnosis based on vibration signal is one of the most commonly used methods.This paper uses modified ensemble empirical mode decomposition(MEEMD),singular value decomposition(SVD),online sequential extreme learning machine(OSELM),differential evolution algorithm(DE)and other technologies to carry out the feature extraction and fault diagnosis technology research based on vibration signals,and apply it to the fault diagnosis of rolling bearings.The main research contents are as follows:(1)The principle and development of EEMD,CEEMD and MEEMD algorithms are described.EEMD is a method of using noise to assist data analysis,but in practical application,because the number of times of adding white noise is limited,the noise can not be completely eliminated after integrated averaging.CEEMD can reduce the reconstruction error caused by white noise by adding two opposite white noise signals,but the amount of calculation is doubled,and when the parameters are set unreasonably,the IMF component decomposed by CEEMD will also have pseudo components.Aiming at the problems of EEMD and CEEMD in the process of decomposition,MEEMD can effectively eliminate the pseudo components generated by decomposition by introducing permutation entropy,and reduce the reconstruction error and the amount of calculation.The simulation results show that the MEEMD algorithm is effective and superior.(2)To solve the problem that some useful information will be lost when meemd is used to analyze noisy rolling bearing signals,a fault feature extraction method for rolling bearings based on SVD optimized MEEMD is proposed.Firstly,CEEMD is used to decompose the signal layer by layer according to the instantaneous frequency,and the permutation entropy of each IMF component is calculated.Then,the number of effective singular values of SVD is determined by singular value difference spectrum and unilateral maxima principle.The IMF component whose permutation entropy is greater than the threshold is denoised by SVD,and then the denoised signal is reconstructed with the remaining IMF component to retain the useful information of the original signal to the greatest extent.Finally,the reconstructed signal is decomposed by EMD to extract the fault feature information.The simulation results show that the improved MEEMD can obtain fault characteristic signals more accurately.And combined with KNN to achieve fault diagnosis,the actual application of rolling bearings verifies the effectiveness and feasibility of the method.(3)Aiming at the problem of low accuracy of traditional fault diagnosis algorithm,the online sequential extreme learning machine(OSELM)optimized by differential evolution algorithm(DE)is applied to fault diagnosis,and a rolling bearing fault diagnosis method based on improved MEEMD and DE-OSELM is proposed.Firstly,the rolling bearing signal is decomposed by improved MEEMD,and the energy of each IMF component is extracted as the feature,and the feature matrix is obtained by normalization.Secondly,OSELM is optimized by DE to solve the problem of random selection of input weight and hidden layer bias,so as to improve the prediction accuracy and robustness.Finally,the extracted feature set is used as the input of DE-OSELM and trained to realize fault classification and recognition.The superiority and reliability of the proposed method are verified by examples of rolling bearing fault diagnosis.(4)Based on the researched method,using MATLAB GUI platform and adopting modular design ideas,a set of rolling bearing fault diagnosis system was designed and developed.At the same time,the system also integrates some other classic signal analysis methods and pattern recognition methods,which can realize the functions of signal acquisition,display,storage,reading,preprocessing,signal analysis,and fault classification.The practical application has verified the practicability and reliability of the system,which provides an effective and convenient analysis tool for online monitoring and fault diagnosis of rolling bearing operating conditions.Finally,the work of this paper is summarized,and the future research direction is prospected.