Improved approach for ANN fault diagnosis with rough set theory

摘要

Beginning with analyzing the characteristic of ANN fault diagnosis model, such main problems of ANN model as training sample quality, relative definition, and fault feature extraction algorithm had been solved using rough set theory. An ANN intelligent hybrid system model was presented, its implementation steps were analyzed, and the validity of above method was proved through application example. Simulations results indicate that the method of this paper can solve problems of directly affecting ANN model precision and generalization ability such as ANN architecture, sample size, and sample quality, decrease the computation time, and increase the diagnosis correctness.