A Motor Fault Detection Method Based on Optimized Extreme Learning Machine

摘要

To solve the problems that the motor fault detection algorithm based on mathematical model is hard to be adapted to those categories with complex non-linear fault. The accuracy of existing fault detection algorithms is not satisfied. In this paper, two methods of feature selection and parameter optimization of simulated annealing-based whale optimization algorithm with (SAWOA) optimized extreme learning machine (ELM) are proposed. SAWOA is utilized to optimize the selection of network input variables and to determine parameters of hidden layer nodes of ELM. In order to verify proposed method, a real motor fault detection case is studied in the paper. Several methods are compared under the same conditions to find out the best method with satisfied accuracy and reliability. The experiment results show that proposed method is capable to improve the classification accuracy and reliability, effectiveness and application value are introduced in the article.