Research On Vibration Performance Degradation Assessment Method Of Rolling Bearing Based On Entropy

As one of the key components of rotating machinery system,the vibration performance of rolling bearing significantly affects the operation quality,reliability and life of mechanical equipment.The service of rolling bearings generally goes through a process from normal state to failure,during which a series of degradation of its performance usually occurs.Traditional fault diagnosis methods only focus on whether the rolling bearing is healthy or damaged,and the identification mode is relatively single.When the fault is identified,the rolling bearing has been seriously failed.In order to ensure the high reliability and safe operation of mechanical equipment and timely active maintenance of mechanical equipment,it is necessary to understand the performance degradation process of rolling bearings,and predict its future evolution trend.When studying the evaluation method of vibration performance degradation of rolling bearing,the extraction and prediction of degradation characteristics are two key points restricting the quality of the evaluation model.However,the early fault of rolling bearing vibration signal is very weak.At the same time,the vibration signal of rolling bearing is accompanied by strong noise background.In addition,the vibration signals of rolling bearings are nonlinear and non-stationary.The coupling of the above mentioned conditions makes it a great challenge to accurately extract the rolling bearing degradation characteristics and accurately predict the degradation trend.This paper takes rolling bearing as the research object,aiming at the problems existing in the evaluation of vibration performance degradation of rolling bearing,and researches on the vibration performance degradation feature extraction,vibration performance degradation evaluation and prediction of rolling bearings.The main research contents includes:（1）The rolling bearing fatigue life test is carried out,and the rolling bearing fatigue life test that provides data support for this paper is introduced in det ail,including test rig,test objects,test conditions,test bearing quality inspection and test process.According to the experimental design,the fatigue life test of 7008AC ultra-precision angular contact ball bearings under four working conditions are completed,and the whole life vibration performance data are collected.（2）Conduct chaos feature recognition research on the collected rolling bearing vibration performance data.On the basis of phase space reconstruction,three indexes,namely correlation dimension,Lyapunov index and Kolmogorov entropy,are used to verify the chaos characteristics of rolling bearing vibration performance.At the same time,the predictability analysis process of data series is established.The experimental results show that the vibration performance data sequences of rolling bearings under the two working conditions have chaotic characteristics,which are suitable for short-term prediction（3）In view of the nonlinear characteristics of rolling bearing motion,combined with the strong advantages of entropy method in dealing with nonlinear problems,three entropy-based methods for extracting vibration performance degradation characteristics of rolling bearing are proposed:maximum entropy similarity method,improved fuzzy entropy method and grey entropy method.The maximum entropy similarity method solves the problem that the probability distribution of the vibration performance data of rolling bearings is unknown.The probability density function of the vibration performance data based on the maximum entropy method is constructed,and the vibration performance degradation probability of the rolling bearing is calculated based on the similarity method.The improved fuzzy entropy method solves the problem of low sensitivity of fuzzy entropy in the extraction of vibration performance degradation features of rolling bearings,and the sigmoid function is constructed to measure the distance between the two modes.The grey entropy method solves the problems that the calculation results of the existing nonlinear analysis methods are inconsistent with the real nonlinear dynamic system and the length of the data required for calculation is long.Simulation cases and experimental studies both verify the superiority of the proposed method.（4）Aiming at the localization of traditional rolling bearing performance degradation assessment methods,a noval performance degradation evaluation method based on gray relationship is proposed.This method uses gray relationship theory to evaluate the relationship between the extracted rolling bearing degradation characteristics and reliability,so as to achieve the purpose of evaluating the degradation features of rolling bearing from the overall perspective of rolling bearing reliability evolution.The effectiveness of the proposed method are verified by experiments.（5）Aiming at the problems that the data length of the rolling bearing degradation trend sequence is so short that it is difficult to predict and the assessment method for prediction results is relatively simple,grey bootstrap markov chain prediction model is proposed based on grey prediction theory,markov chain and bootstrap method.At the same time,the evolution trend of rolling bearings vibration performance degradation is described from different perspectives by using estimated truth value X₀,estimated interval[X_L,X_U],predicted reliability H_B and extended uncertainty U.Experimental results show that the proposed method has higher forecast reliability.（6）The degradation trend chaos prediction model of rolling bearing vibration performance is establishe.Firstly,the shape parameters in Weibull distribution is used as characteristic parameters to describe the degradation trend of rolling bearing vibration performance.Then,the degradation trend sequence and vibration performance sequence are qualitatively compared,and the grey system theory is introduced to perform grey correlation analysis on the degradation trend sequence and vibration performance sequence to illustrate the reliability of the qualitative analysis results.Finally,the phase space reconstruction of degradation trend sequence and vibration performance sequence is carried out,and chaos prediction is carried out by first-order local method,weighted first-order local method and improved first-order local method.The experimental study shows that the delay time has little effect on the prediction accuracy,while the embedded dimension has a greater effect on the prediction accuracy.The prediction accuracy of the chaotic prediction method for the vibration performance degradation trend sequence of rolling bearings is significantly higher than that of the rolling bearing vibration performance series.