A model based fault detection and diagnosis system for rolling mill equipments

In this paper, the implementation of a Model Based Fault Detection and Diagnosis System, that uses fuzzy logic to determinate the nature of the detected faults in rolling mill equipments is presented. The system is built with 4 components which work independently. An Identification module estimates the parameters of a continuous domain second order transfer function model for the process by analyzing the step response. A Predictive model module generates the controlled variable residual which is statistically analyzed in a Detection module. The results of the statistical analysis are fuzzified and processed in a Diagnosis module to determine detected fault's nature. The system is tested using real operation data of a main motor process in order to detect and classify abnormalities into Operation Point Change (OPC) or Process Fault (PF) alarms.