Summary: with the rise of industrial big data and Internet of things, the research on bearing life and health detection will be pushed to a new height. Based on big data, the research on health status evaluation and operation state prediction will help improve the reliability of bearing equipment, avoid deterioration of bearing health status and prolong its working life. This paper summarizes the achievements of bearing health testing and life prediction at home and abroad, and prospects for it. Although some achievements have been made in the existing research, there are few researches on the key technologies such as the operation data collection, degradation fault characteristic information extraction, health status assessment and residual life prediction of bearing equipment. Therefore, it is of great theoretical significance and application value to further improve the service life of bearings.



Key words: rolling bearing; reliability; big data; health monitoring; life prediction



Rolling bearing is one of the key components that determine the health and life of machinery. The research on bearing life and health state has many characteristics, such as many factors, large life dispersion, time-consuming test, difficult data accumulation and difficult theoretical modeling. With the accumulation of data in the process of processing, manufacturing and operation, the development and popularization of Internet of things, cloud computing and intelligent algorithm, the production environment has gradually possessed the basis of big data environment. On this basis, the bearing health management and life prediction can make the bearing realize the mutual comparison with its own state, the accumulation of data and experience model, and the collaborative diagnosis of faults, and then, the bearing can realize the mutual comparison with its own state, the accumulation of data and experience model, and the collaborative diagnosis of faults Become a smart system with self-learning and self-development ability. In such a system, bearing system is no longer an independent individual. Through the information network system, the prognosis analysis of all equipment in the whole manufacturing operation system can be carried out to make the control and decision-making end see the state and operation ability of the bearing. At present, every kind of bearing is managed by its professional sensors and analysis equipment. There is no unified platform to integrate the common features of bearing faults and their corresponding general and reconfigurable algorithms. On the other hand, although sensors and analyzers can complete basic monitoring and prediction tasks, the mining of root failure reasons, high response prediction methods, large-scale equipment clusters and cooperative management of various bearings can not be well adapted. In the aspect of bearing health monitoring and life prediction, it will face severe requirements. Therefore, this paper will introduce the achievements of bearing big data health monitoring and life prediction in detail, and prospect its development direction.



1. Study on the life and reliability of rolling bearing



1.1 development and status of bearing life and reliability testing machine



Bearing life refers to the total number of revolutions or working hours before a fatigue peeling occurs in a rolling body or raceway of a bearing. Bearing life test can not be separated from bearing life testing machine, and the development of bearing life testing machine also witnessed the development of bearing life research. The research on bearing life in China is mainly led by Luoyang Bearing Research Institute and Hangzhou Bearing Test Research Center (United Nations assistance), and other related enterprises' life and reliability test bases are auxiliary, and jointly undertake the bearing life, reliability and performance test research in China's bearing industry. At present, the design, research and development and production of bearing life testing machine in China have been fully autonomous, and even some technical concepts have reached the international leading level, but it started very late compared with many foreign bearing companies such as skalfu, sheffler, Timken and enten. In the early 20th century, the development of bearing industry in China mainly depended on the technical support of the old Soviet Union. The life test of bearings was mainly carried out on the basis of ZS bearing life testing machine, and the evaluation quality of this kind of testing machine was eliminated from the requirements of bearing service performance development. And the Hangzhou bearing test and Research Center (HBRC) introduced from the United States through the United Nations Assistance Project "f&m" The new type of rolling bearing fatigue life tester is not only expensive and monopolized in technology, but also adopts pneumatic high-pressure power source and 60Hz electric frequency, which is not suitable for China's national conditions. Therefore, it is imperative to realize the independent production of bearing life strengthening testing machine. In 1990s, Hangzhou bearing test and Research Center independently developed ablt-1 type automatic control rolling bearing fatigue life strengthening testing machine based on the advanced life testing machine abroad, which opened up a new market and new prospect for the domestic life testing machine, and had an international advanced level at that time. With the birth of the fatigue life strengthening testing machine of ablt-1 rolling bearing, the development of domestic bearing life testing machine is springing up, but it is mostly derived or improved on the basis of ablt-1.





Figure 1 fatigue life and reliability strengthening test machine of ablt-1a rolling bearing



With the attention of bearing manufacturers and users on their life test and reliability engineering, the bearing models to be tested are becoming more and more diversified, and ablt-1 testing machine can not meet many different parameters of bearings. Based on the continuous digestion, absorption and improvement of ablt-1 bearing life testing machine, Hangzhou bearing test and Research Center has designed and developed a new bearing test technology and method with complete independent property rights, such as type 2, 3, 4, 5, 6, 7, 8 and 9, etc. The advantages of previous test technology have been absorbed by the ablt series fatigue life and reliability strengthening testing machine, and the automation control level is further strengthened and improved. At present, many enterprises are trying to make progress in the direction of "industry 4, intelligent manufacturing, Internet plus" and so on. Life testing machines are developing to the trend of intelligence and automation. For different industries or enterprises, the demand mode of market is different, the production and processing process of products is different, and the demand for individualization and intelligence is increasing. The life and reliability testing machine of rolling bearing has also been innovated correspondingly. The most important innovation point is the driving system and loading system, because the main human control factors affecting bearing life test are test speed and load. The addition of servo motor and electro-hydraulic servo valve makes the bearing life test more easily meet the requirements of customer personalization and automation. The drive device adopts servo motor, which can convert voltage signal into motor speed to drive control object. In automatic control system, it is used as the executive element, and has the characteristics of small electromechanical time constant, high linearity and starting voltage. The load system uses electro-hydraulic servo valve, after receiving the electrical analog signal, it can output the modulated flow and pressure correspondingly, and then convert the weak electrical input signal of small power into the output of high-power hydraulic energy (flow and pressure), and the load load is more rapid and accurate. Through the personalized design, it can meet the needs of most of the fatigue life strengthening tests of rolling bearings.



The development of bearing life testing machine in China has experienced more than 40 years, and has basically possessed skilled R & D technology, rich test experience and accumulated a large number of test data. Although some research results have been achieved, there are too many factors affecting bearing life, which is too complex, and the data processing of bearing life experiment still needs to be further mproved. At present, the database of the relevant technologies such as fatigue mechanism research, failure factor analysis, material smelting and processing technology and test data analysis should be established, and the recommended values of the correction coefficients for the calculation of the life of the bearings in China should be put forward as soon as possible, so as to be suitable for the service information of various bearings with rapid innovation.



1.2 prediction of bearing life and reliability test



During the whole life cycle operation of bearings, it is likely to be affected by high temperature, poor lubrication, improper assembly, foreign material intrusion and other factors, which will lead to bearing damage and failure. Because of the discrete bearing life, a batch of bearings with the same structure, same material, same heat treatment and processing method have dozens or more difference between the maximum life and the minimum life under the same working conditions. The traditional mathematical statistics method shows that the bearing life test data are almost in accordance with Weibull distribution or logarithmic normal distribution, but it is still difficult to predict in actual conditions. Therefore, it is very important to deal with the bearing life test data effectively. The research institutes at home and abroad actively carry out the relevant research on the bearing life test data. Saxena et al. Uses the power spectral density parameter as the performance degradation index of rolling bearing, and forecasts the remaining service life of the bearing. The density parameter can diagnose the and degree of the fault. Xiao ting and others use kurtosis and multi domain feature sets as trend prediction indicators, which can effectively reflect the running state of bearings, and also predict the performance degradation trend of bearings. Banjevic and other use the proportional risk model to predict the reliability function and residual life of the equipment, and take the covariate of a certain time as the benchmark to predict the remaining life of the equipment. Based on previous research, kacpnynski proposes a prediction model combining monitoring data with material parameters, and uses this model to carry out the life prediction of rolling bearing. Kimotho et al. Proposed a method to optimize kernel function and penalty parameters of SVM by hybrid differential evolution particle swarm optimization algorithm, which improved classification accuracy and residual life prediction accuracy of SVM, and verified by NASA standard bearing failure data. Orsagh et al. Used Yu Harris model to predict the initial time of fatigue spalling failure of rolling bearing, and predicted the failure time of rolling bearing by kotzala Harris model. Panigrahi proposed a DPSO, which is used to solve the problem of maximum likelihood function estimation in the study of bearing performance degradation, and achieved good prediction results.



At present, the statistical life model still plays a leading role in bearing life prediction. However, the test and engineering application show that the life calculated by statistical life model is usually conservative and the bearing life dispersion is large. Therefore, how to complete the bearing life model is a main problem through the research of bearing performance degradation mechanism. With the development of information technology and artificial intelligence, the life prediction method based on state monitoring has become a hot research area of bearing life prediction. With the help of big data and artificial intelligence information technology, dynamic signals reflecting bearing service performance, signal characteristic parameters representing bearing performance decline can be obtained, and mapping relationship between signal characteristic parameters and residual life can be established, so as to realize the prediction of residual life. However, there is no suitable characteristic parameter to measure the evolution rule of the performance decline in the bearing operation. Compared with the traditional life prediction model, artificial intelligence methods such as neural network are not clear enough in physical significance and the influencing factors of parameters are larger. How to carry out in-depth research on the difficulties is very important for the bearing life prediction technology.



Research on health monitoring of big data of rolling bearing



Bearing health monitoring is based on the existing known data, and forecasts the change of the bearing operation status trend in a certain time in order to accurately and quickly obtain the fault development information. The state monitoring and health monitoring of bearings can master the degradation process of bearings, prevent the occurrence of larger faults and prevent them from getting ill.



Caesarenedra and others used the correlation vector machine regression algorithm and the logic regression combination method to evaluate the bearing degradation degree and the predicted failure time. Yu et al. Uses local local local projection method to extract the characteristics of bearing operation, and uses Gaussian mixed model and statistical index to evaluate the health status of bearings. The results show that the effect of the feature extraction is obviously better than that of component analysis. Li Xiuwen and others use frequency-domain morphological filtering to reduce noise of acoustic emission signals of low speed rolling bearings. The simulation results are compared with the actual bearing signals. The results show that the method has a good effect. Rojas et al. Proposes a fault diagnosis method of rolling bearing based on SVM, which identifies the state of rolling bearing by SVM. By using discrete wavelet transform and spectral kurtosis analysis, Jeong et al. Obtains the characteristic frequency of faults in each part of rolling bearing, and completes the diagnosis of composite faults of inner ring outer ring, inner ring rolling body, outer ring rolling body and inner ring outer ring rolling body.



For the indicators of bearing operation state monitoring, some can be realized by collecting physical parameters, such as temperature, vibration, noise amplitude, etc., and some need researchers to advance the data by signal processing, such as the degree of temperature change, vibration intensity, sound pressure intensity, etc. Through the characterization of these indexes, the health state of bearings can be evaluated, the prediction and early warning of bearing operation state can be carried out, and the engineering personnel can be guided to take corresponding protection measures to avoid deterioration of bearing health status. But at present, there is no unified platform for bearing health monitoring and analysis to integrate the common characteristics of bearing failure. Moreover, the mining of fault causes, high response prediction methods and collaborative management of various bearings are not perfect.



3. Conclusion



Although some research results have been made by domestic and foreign scholars in bearing health monitoring and life prediction, most of the researches still remain in the conventional fault diagnosis, and few researches on Bearing Based on big data analysis and intelligent algorithm evaluation. Bearing fault prediction and health management based on big data analysis can realize the independent guarantee of bearing equipment. With the development of Internet of things and artificial intelligence, the key technologies such as running data collection, degradation fault feature information extraction and health status assessment will become the development trend of bearing health monitoring and life prediction.



4. Outlook



In 2006, Hangzhou bearing test and Research Center Co., Ltd. established the first post doctoral research workstation in lower urban area of Hangzhou. Up to now, more than 20 postdoctoral students (16 of them are directly recruited by HBRC postdoctoral workstation) have been recruited, and engaged in basic research on the application of bearing related science and technology. He has undertaken several vertical national and provincial level scientific research projects, and cooperated with well-known domestic enterprises, universities and research institutes to complete more than ten horizontal research projects, which has made exemplary contributions to the bearing industry in the fields of production, learning, research and win-win cooperation. It has more than 20 patents and software copyright (including 10 invention patents and all commercialized); participated in more than 20 national standards and industry related standards (including 5 first drafting units (persons); published more than 300 papers; published more than 10 Chinese and English professional works and samples; won more than 10 national and provincial awards, and recently won the science and technology of China Federation of mechanical industry Second prize of technical progress and the title of "national excellent enterprise in research and development and manufacture of special bearing equipment".



After more than 40 years of joint efforts, Hangzhou bearing test and Research Center Co., Ltd. has customers all over the world, mainly for bearing production enterprises, bearing users, bearing testing institutions, colleges and universities and research institutes. Some products have passed the EU certification and successfully entered the European market and the international markets such as the United States, India, Japan, South Korea, etc., which has taken the first key, pioneering and substantive step for China's bearing test and equipment technology to the world. At the beginning of the fifth decade of HBRC ("the 14th five year plan"), adhering to the professional, focused and attentive, based on East China, facing the country, connecting with the international, serving the world!