Application of neural networks: Part 1, To determine the operability of check valves: Part 2, To improve the operation of nuclear power plants; Final report
- Tennessee Univ., Knoxville, TN (United States). Dept. of Nuclear Engineering
Although there are many possible failure mechanisms for check valves, the most common problems are due to system flow oscillations or system piping vibrations that induce check valve component wear, and often component failure. Most failures induce additional vibration or emit sounds that can be detected by monitoring acoustical emissions and vibration from the valve body. The methodology presented in part 1 of this report involves training a neural network to model the internal behavior of the check valve and its internal components at a time when the valve is known to be working properly. Then the neural network model is put into a monitoring mode where the output of the neural network (the predicted values of a parameter) is compared with the actual value of a parameter from the corresponding sensor. If the difference between the actual and predicted values is small, the valve is deemed to be operating properly. If the difference becomes large, something has happened to the internal components of the valve, and it should be disassembled and inspected. In part II of this report, some thirty potential and actual applications of neural networks to improve the operation of nuclear plants, grouped into seven general categories, are discussed. These categories are: Monitoring of Nuclear Power Plant Sensors and Systems; Monitoring of the Thermodynamic Behavior of a Power Plant; Diagnosis of Nuclear Power Plant Transients; Artificial Neural Network Control Systems; Vibrational Analysis using Artificial Neural Networks; Inferential Sensing and Virtual Instrumentation; and Off-Line Applications of Artificial Neural Networks.
- Research Organization:
- Electric Power Research Inst. (EPRI), Palo Alto, CA (United States); Tennessee Univ., Knoxville, TN (United States). Dept. of Nuclear Engineering
- Sponsoring Organization:
- Electric Power Research Inst., Palo Alto, CA (United States)
- OSTI ID:
- 143966
- Report Number(s):
- EPRI-TR-103443-P1-2
- Resource Relation:
- Other Information: PBD: Jan 1994
- Country of Publication:
- United States
- Language:
- English
Similar Records
Neural networks and their potential application to nuclear power plants
Neural networks for the monitoring of rotating machinery
Related Subjects
21 NUCLEAR POWER REACTORS AND ASSOCIATED PLANTS
NUCLEAR POWER PLANTS
REACTOR CONTROL SYSTEMS
REACTOR MONITORING SYSTEMS
NEURAL NETWORKS
VALVES
OPERATION
TECHNOLOGY UTILIZATION
PWR TYPE REACTORS
BWR TYPE REACTORS
DIAGNOSTIC TECHNIQUES
FAILED ELEMENT DETECTION
REACTOR COMPONENTS
PERFORMANCE TESTING