Degradation Modeling Based On Safty Region For Rolling Element Bearing

With the progress of the society and industry, many industrial types of equipment in major industries are becoming more integrated, smaller in size, faster and more efficient, and the requirements of high reliability for these devices are also increasing.Once the key components of the equipment are failure, the damage could impact the production or cause the destruction of equipment or even endanger the safety of life,resulting in serious social problems. As the most commonly used part of mechanical equipment, bearings are widely distributed in the major rotary machineries. Therefore,the study of the degradation of rolling bearings, and the research of how to recognize the operating status of bearings correctly, has a great significance for equipment health management and maintenance.The degradation process of the component with high reliability and long life, like the rolling bearing, is extremely slow, bringing difficulty to the research on the assessment of rolling element bearing performance degradation. The accelerated degradation test (ADT) is an effective method to cope with such kind of problem. The theory of ADT is based on not changing the principle of degradation and it will lead rolling bearing to degrade much faster by increasing the accelerated stress such as the load or speed. In this way, we can collect the whole life cycle degradation data. On the basis of ADT, this thesis discusses the research about degradation state modeling method. The main contents are as follow.(1) To solve the problem of degradation modeling, this thesis presents a method of degradation modeling for rolling bearing based on safety region. Through the tool of support vector data description and using vibration data which are collected from healthy bearing to train the safety region model, we can obtain the safety region with a closed boundary in kernel space. As the degradation of running status of bearings, the new sample will be gradually deviate from the safety region. Therefore, this thesis uses the distances from the center of safety region to the new sample point to describe the trend of degradation. In this way, we can realize the purpose of degradation modeling for rolling bearing.(2) To solve the problem of feature selection for the degradation modeling method,this thesis proposes a degradation feature selection algorithm based on multi-metric fusion. Feature parameters as the input of the degradation state model based on safety region, will greatly affect the result of degradation modeling. Therefore, we evaluate the whole life cycle degradation feature by using the monotonicity, robustness and relevance metric and come to a comprehensive result to realize the degradation feature selection for safety region modeling.(3) To solve the problem of insufficient data of degradation modeling for validation,we design several groups of accelerated degradation experiments on the basis of the existing bearing ADT platform. In this part of the thesis, considering the stress level, the stopping criterion and the repeatability of ADT, we sum up the methods of the rolling bearing ADT. Finally, the corresponding ADT data are obtained and the proposed methods of degradation modeling based on safety region are verified.