Fault detection and monitoring using parameters identification and principal component analysis. application to rotary machines in skin pass process

A new approach for fault detection and monitoring based on the parameters identification coupled to the Principal Component Analysis (PCA) is proposed in this paper. The proposed Fault Detection and Monitoring consists to apply the PCA method on the dynamic of the identified parameters. Conventional PCA uses the process inputs and outputs as variables which are used in the computing procedure. Using the process parameters behaviour as variables in the PCA computing procedure improve the detect ability by reducing the wrong faults generating by the noise effects. Application on the rotary machines in skin pass machines of cold rolling will be developed in this work.