Skip to main content
U.S. Department of Energy
Office of Scientific and Technical Information

Development of an expert system for signal validation: Topical report

Technical Report ·
OSTI ID:7010984
A signal validation and sensor information expert system is developed and applied to nuclear power plant subsystems. The expert system interactively provides the user with a list of sensors used to monitor a subsystem of a power plant and answers the user's questions about the characteristics of those sensors. The expert system performs an evaluation of the present output characteristics of an instrument channel. The evaluation is fully automated and is designed to mimic an operator's approach to signal validation. The evaluation is designed to detect three types of possible sensor anomalies: (1) incorrect average output, (2) improper noise level,and (3) incorrect response to a system perturbation. There are two tests used to detect each possible anomalous condition. One test is a comparison of the specified sensor characteristic against the same characteristic of its redundant sensors. The second test is a prediction of the sensor's expected behavior and comparison with its observed behavior. The evaluation was tested using operational data from twelve sensors used to monitor level, pressure, primary and secondary coolant flow rates, and hot and cold leg temperatures of a steam generator in a commercial nuclear power plant. The results of the testing demonstrated successfully that the evaluation detected the three possible anomalies, and was able to handle conflicting information using the comparison among redundant sensors and the system specific test used to detect the anomalies.
Research Organization:
Tennessee Univ., Knoxville (USA). Dept. of Nuclear Engineering
DOE Contract Number:
AC02-86NE37959
OSTI ID:
7010984
Report Number(s):
DOE/NE/37959-17; ON: DE88013077
Country of Publication:
United States
Language:
English