Rolling Bearing Fault Diagnosis Method Research Based On EMD And SVM

In the petroleum and petrochemical production process,the failure of a key component of the machinery tends to affect the normal operation of the entire chain of devices,resulting in unplanned downtime or even combustion explosion and other serious accidents.Therefore,the security monitoring and fault diagnosis research on mechanical devices has important practical significance and economic significance.This paper carries out the following research about the fault diagnosis system of rolling bearing that is the petrochemical production equipment universal key components.(1)Based on rolling bearing classification characteristic and its mechanical structure,analyzing the failure modes and bearing vibration mechanism.Bearing vibration signal characteristics are summarized in different categories,and classified bearing vibration signal characteristics in different categories of failure.(2)In fault diagnosis system feature extraction section,using an adaptive amplitude harmonic addition method and improvement mirror extreme continuation method to inhibit the mode mixing and the end effect of Empirical Mode Decomposition.(3)In fault diagnosis system pattern recognition section,using genetic algorithms to optimize parameter of Support Vector Machine,and looking for the optimal kernel parameters and the penalty factor to build the best Support Vector Machine model to raise samples signals identification classification accuracy.(4)Based on laboratory rolling bearing fault signal data,combineing with the improved Empirical Mode Decomposition and optimized Support Vector Machine to build a rolling fault diagnosis system.(5)The development tools for MATLAB GUIDE,developing a rolling bearing fault diagnosis system based on improved Empirical Mode Decomposition and optimized Support Vector Machine,to achieve the improved algorithm software,and enhance the usefulness of the diagnostic system.