Gear fault diagnosis based on support vector machine optimized by artificial bee colony algorithm

Gearbox is an important transmission component in mechanical equipment, which is prone to failure. When diagnosing the faults of gearbox, the recognition ability of support vector machine (SVM) is greatly influenced by the kernel function and its parameters. However, the optimal parameters are difficult to find. To solve this problem, the parameter optimization method of support vector machine based on the artificial colony bee algorithm was presented. The vibration acceleration signals of normal gear, chipped tooth gear, and missing tooth gear were collected and the fault features were extracted based on the EEMD and the kernel function. In the experiments, the optimization of Gauss RBF kernel support vector machine parameters was taken as the object, and the optimize performance of the genetic algorithm, the particle swarm optimization algorithm and the artificial bee colony algorithm were compared and analyzed. The results show that the artificial bee colony takes the least time, while the genetic algorithm takes the longest time, the accuracy of the artificial bee colony algorithm is better than the others. The recognition rates of gearbox faults were improved by using the SVM classification model which was optimized by the bee colony algorithm. (C) 2015 Elsevier Ltd. All rights reserved.