Implementation of an integrated real-time process surveillance and diagnostic system for nuclear power plants

Abstract This paper presents two parts, real-time process surveillance unit and fault diagnosis unit, which are separated from each other. However, these units are connected by trigger-rules in a real-time expert system. The design structure that has been adopted is capable of inspecting errors and revising the model. Multilevel Flow Model (MFM), which is a method for functional modeling, is introduced briefly and illustrated with a reactor coolant system. Utilizing functional modeling method to represent system knowledge, this modeling method is especially useful when the domain experts are not available. Considering issues such as loop modeling and mutually exclusive events inevitably exist between the observation points, a novel modeling technology called observation points’ protection was used to build a generic fault model and preserve the statuses of observation points during reasoning within an expert system. This paper also presents minimal candidate and domains of interpretation, which are especially useful for finding the fundamental root cause when multiple faults occur. The process surveillance and diagnostic system is implemented on the platform of G2, which is an environment for developing real-time expert systems. The emulation test was conducted and it has been proven that the fault diagnosis expert system can identify the faults correctly and in a timely manner.