Potential application of neural networks to the operation of nuclear power plants

The application of neural networks, a rapidly evolving technology used extensively in defense applications, to some of the problems of operating nuclear power plants is a logical complement to the expert systems currently being introduced in some of those plants. The potential applications of neural networks include, but are not limited to: (1) Diagnosing specific abnormal conditions. (2) Identifying nonlinear dynamics and transients. (3) Detecting the change of mode of operation. (4) Controlling temperature and pressure during start-up. (5) validating signals. (6) Plant-wide monitoring using autoassociative neural networks. (7) Monitoring of check valves. (8) Modeling the plant thermodynamics to increase efficiency. (9) Emulating core reload calculations. (10) Analyzing temporal sequences in the U.S. Nuclear Regulatory Commission Licensee Event Reports. (11) Monitoring plant parameters. (12) Analyzing vibrations in plants and rotating machinery. The work on such applications indicates that neural networks alone, or in conjunction with other advanced technologies, have the potential to enhance the safety, reliability, and operability of nuclear power plants. 36 refs.