Performance Degradation Evaluation Based On Condition Monitoring For Rolling Bearing

The application of the bearing can be seen everywhere in mechanical equipment, and it is also the most vulnerable. If the bearing breaks down, it usually causes the accident and brings economic losses. The bearing condition monitoring can make a real-time understanding of the bearing running situation to judge if that is going to fail and take relevant measures as early as possible. Then it can avoid accidents in many cases. So in industrial production the bearing condition monitoring, reliability and performance degradation prediction, and fault diagnosis are the concerned issues. The bearing failure time can be known through reliability evaluation.Then maintain the bearing at the right time before to prolong the maintenance cycle and service life. It has very important significance for the normal operation of equipment. This paper has a discussion on the basis of the theoretical method, combining with the instance of rolling bearing. And study performance degradation evaluation methods of the bearing. The concrete content is as follows:(1)Firstly the main condition monitoring technology of rolling bearing, such as sensor detection method of condition monitoring, the state monitoring method of rolling bearing and analysis and diagnosis of vibration signal is discussed. Then the research on feature extraction of bearing vibration signal and signal processing method in theory are done, including time domain analysis, frequency domain analysis and time-frequency analysis method. Each method has its respective characteristics and is suitable for application in different occasions.It is the basis and premise of performance degradation evaluation to master the bearing condition monitoring and feature extraction method.(2)Make performance degradation prediction through building state space model. And improve the solution procedure of the equation of state. Firstly the bearing vibration signal is got through condition monitoring method. The characteristics energy of vibration signal is extracted using wavelet analysis method. Secondly select the frequency band energy whose feature trend accords with the characteristics of degradation as the state evaluation index and establish a state space model. Then calculate the reliability. It focuses on the improved kalman filter, which is the core content of the solving method in the state space model. Finallyit can be seen that the prediction effect of improved kalman filter method is more ideal than the original method, combined with two examples of rolling bearing to verify the validity of this method.(3)The performance degradation evaluation method based on support vector machine(SVM) of univariate and multivariate is proposed. The bearing vibration data are obtained by on-line monitoring technology. Then make feature extraction of signal and select the characteristic index and characteristic variable which can reflect the characteristics of the bearing degradation state. Establish regression prediction equation and forecast the characteristic indexes using the training data of characteristic variables as input. Finally validate on the method of univariate and multivariate support vector machine respectively combining with an instance of rolling bearing. The result shows that the prediction result of multivariate support vector machine(MSVM) is more accurate because it contains more comprehensive information.