Bringing a Nuclear Power Generating Station under the control of an expert system will require an extensive program of coordinated research, development, prototype testing, and regulatory review. Conversion of nuclear energy to electrical energy carries with it a myriad of unique and complicated issues. Some will be purely technical in nature. Other less technical, and more esoteric issues will also impose constraints on the programmer and on the utility. This paper presents and discusses the key issues that relate to the viability of expert systems in the nuclear power arena.
I. STATEMENT OF THE ISSUESAmong the issues that would complicate the programmer's efforts are the sheer volume of information to be processed, the time constraints for operator response, the ties between the operation of any one system and the operation of all others, and the numerous factors that may modify the relative merits of alternative courses of action. Even reactor physics must be taken into account, in that it forms the basis of many of the normal and emergency operating procedures that the expert system must emulate. The utility's operating license adds constraints related to equipment diversity, redundancy, and reliability. It also imposes limitations not only on the time allowed for preventive and corrective maintenance, but also on the plant configuration allowed during such maintenance. Individual Plant Evaluations determine risk factors associated with various failure types. The expert system must be taught to recognize and track those situations that are covered by the operating license, as well as those that affect the risk factors.Other issues relate to the types of equipment installed in the reactor plant. The age of many components will be a factor, as will the lack of standardization of plant design.