Fault Diagnosis of Generator Using CMAC Neural Network with Gauss Basis Function
Wan Shuting1; He Peng1; Zhao Songjie2
1.Department of Mechanical Engineering, North China Electric Power University, Baoding, 0710032. Department of Mechatronics Engineering, HandanPolytechnic College, Handan,056000
Abstract:Based on CMAC neural network with Gauss basis function, a novel method was proposed for fault diagnosis of turbo-generator. In order to achieve higher precision and better generalization ability, this method used Gauss basis function in CMAC neural network. Because electrical and mechanical coupling characteristics of the generator, made integrated mechanical and electrical characteristics as a neural network training input samples. Through MATLAB simulation, we got completely correct diagnosis results with the high convergence speed and accuracy that met the requirements of on-line monitoring. By comparing CMAC network training results when used the different values of the learning rate and generalization constant value at the same time, the study analyzed the influence of the learning rate and generalization constant to the neural network.