Intelligent diagnosis of transparent faults in process control

Various alarm systems with intelligent diagnosis support are being developed for industrial processes. These systems are unable to detect faults in certain situations. There could be some faults which may be generated in a process plant that cannot be detected until they become catastrophic, as the normal function of the process is not disturbed. The authors treat these faults as transparent to the present alarm systems and propose a methodology for detecting them. A fire tube boiler is considered as a process example for investigating the transparent faults. The methodology is based on verifying the relationship between the interdependent process parameters, and has potential to improve the reliability of the system