An Immune Detector-Based Method for the Diagnosis of Compound Faults in a Petrochemical Plant

Aiming at the serious overlap of traditional dimensionless indices in the diagnosis of compound faults in petrochemical plants, we use genetic programming to construct optimal indices for that purpose. In order to solve the problem of losing some useful fault feature information due to classification processing, during the generation of the dimensionless index immune detector, such as reduction and clustering, we propose an integrated diagnosis method using each dimensionless index immune detector. Simulation results show that this method has high diagnostic accuracy.