Development of an expert system for signal validation

Diagnosis of malfunctions in power plants has traditionally been in the domain of the process operator, who relies on training, experience, and reasoning ability to diagnose faults. The previous work in the development of operator aids for fault diagnosis has focused on detecting and diagnosing process anomalies. The most significant advances in the application of expert systems to process diagnostics has involved construction of computer models with interconnecting nodes. General principles of the plant process are used to isolate the cause of a possible anomaly to one of the nodes. Another approach to process diagnosis is to identify the cause of a possible anomaly with an event-cause model of the process. We describe a method of signal validation using expert system technology which detects possible anomalies in an instrument channel's output, similar to the procedure used by an operator