Soft computing technologies in nuclear engineering applications

Abstract

The application of soft computing technologies, particularly neural networks, fuzzy logic, and genetic algorithms, to the surveillance, diagnostics and operation of nuclear power plants and their components is an area that has great potential for exploitation. Areas of potential application are to the surveillance and diagnostics of complete nuclear power plants and to specific systems such as check valves, instrumentation systems, and rotating machinery. Applications include sensor surveillance and calibration verification, diagnostics of both plant transients and specific faults, efficiency optimization, vibration analysis, loose parts monitoring, and adaptive and/or optimal control. The synergistic benefits of combining the use of neutral networks, fuzzy systems and genetic algorithms are illustrated in several application. Although some of the work cited (e.g. vibration systems) are not necessarily associated with nuclear power plants, the results are directly applicable. Indeed, the methodologies of soft computing technologies have many applications outside the nuclear power field, e.g., fossil-fired power plants, chemical process facilities, high performance aerospace systems, financial market issues, sociological systems, and others.