Fault detection and diagnosis using rule-based support system on fatty acid fractionation column

Abstract

This paper presents a unified approach to process fault detection and diagnosis (FDD) intelligent program for pre-cut fatty acid fractionation column. Process history based methods (rule-based feature extraction) is used to implement the FDD rule-based support system. Plant model was simulated by using an existing commercial process simulator-HYSYS. Plant™ software in order to compute the confidence region. Warning limits for process parameters (temperature, flow rate and pressure) are computed by using statistical techniques. The uncertain information represented on three discrete states ‘high, normal and low’ in production rules form. Process variables are defined as fault if they are deviated outside this region. Identification of causes, consequences and suggested actions for each deviation assisted by Hazard and Operability Study (HAZOP) analysis are generated into rule-based algorithm. Forward chaining strategy is used to interpret the rule-based system. The whole system has been developed using Microsoft Visual C++ programming language. The system was founded to be able to detect faults and promptly diagnose them.