Research On Performance Degradation Evaluation Method Of Rolling Bearing Based On PCA And SVM

Intelligent maintenance of mechanical equipment has become one of the hotspots in the field of fault diagnosis,performance degradation assessment is an important component in the field of intelligent equipment maintenance.Rolling bearings are one of the indispensable parts of rotating machinery and equipment.Due to its long-term operation under high speed and variable load,it will inevitably lead to its performance degradation,which will affect the normal operation of the machine and even endanger the personal safety of the operator.Therefore,to carry out the evaluation of performance degradation of rolling bearings,so as to timely repair,is the key to ensure that rotating machinery safety,stability,precision and efficient operation.The traditional fault diagnosis of rolling bearings focuses on the fault location(outer ring,inner ring,rolling body)to judge,and in the actual production,it is meaningless to use this method to determine the bearing within a component damage,because it is impossible to simply replace the parts and components with damage.In order to determine the replacement time of rolling bearings in machinery and equipment to ensure the safety and stability of the work,a performance evaluation method of rolling bearing performance based on SVM is proposed in this paper.Performance degradation assessment of rolling bearings is the key to maintenance.After analyzing the principle of principal component analysis(PCA)and multidimensional scale analysis(MDS)and support vector machine(SVM)algorithm,a feature extraction method based on PCA and MDS and the performance degradation evaluation algorithm of rolling bearing based on geometric distance of SVM are proposed.Firstly,the loop feature extraction is made by setting the contribution rate of PCA,and then the extracted features are sent into MDS to reduce the dimension of the original data.At the same time,the parameter optimization in SVM algorithm is improved to adapt to the performance evaluation algorithm of rolling bearing.The above state features are extracted from the health status and fault state of the rolling bearing and are classified into training samples and test samples.After training the SVM performance degradation evaluation model with training samples,the test samples were entered into the trained performance degradation evaluation model for performance evaluation.Finally,based on the full life test data and laboratory data of rolling bearing,a reasonable trend of performance degradation of rolling bearing is obtained.