A Novel Feature Selection method for Fault Detection and Diagnosis of Control Valves
In this paper, a novel method for feature selection and its application to fault detection and Isolation (FDI) of control valves is presented. The proposed system uses an artificial bee colony (ABC) optimized minimum redundancy maximum relevance (mRMR) based feature selection method to identify the important features from the measured control valve parameters. The selected features are then given to a naïve Bayes classifier to detect nineteen different types of faults. The performance of the proposed feature selection system is compared to that of six other feature selection techniques and the proposed system is found to be superior.
Keywords: Feature Selection, Control Valves, Fault Detection and Diagnosis, Artificial bee colony, Feature selection, naïve Bayes
Download Full-Text
