Genetic Algorithms Based ANN Approach for Fault Diagnosis

摘要

This paper firstly describes the advantages and the limitations of the Artificial Neural Network (ANN) approach for fault diagnosis. Then combines ANN with Genetic Algorithms (GAs) by utilizing GAs depending on history data or experience data to generate ANNs, which avoids the limitation of the need to design different ANNs when diagnosing different systems, and at the same time the designed ANNs by this approach generally have better performances than the ANNs designed manually.