MOTOR BEARING FAULT DIAGNOSIS BASED ON RELEVANCE VECTOR MACHINE OPTIMIZE BY IMPROVED FRUIT FLY OPTIMIZATION ALGORITHM

Abstract

Aiming at the fact that the fault diagnosis performance of relevance vector machine（RVM） in motor bearing highly depends on the parameters selection, a motor bearing fault diagnosis method based on RVM optimized by fruit fly optimization algorithm with reverse cognition（RCFOA） was proposed. In order to improve search ability of FOA, reverse cognition strategy was introduced and improved the original FOA algorithm. Use the RCFOA to optimize RVM parameters can effectively improve the classification performance of RVM. Different fault type and different fault degree of motor bearing fault diagnosis experiment results show that the RCFOA can obtain better parameter when compared with some other methods, improved the fault diagnosis accuracy of RVM and can applied to fault diagnosis efficiently.

Keywords

• Fruit fly optimization algorithm;Reverse cognition;Relevance vector machine;Fault diagnosis;Bearing