Expert systems applied to two problems in nuclear power plants

This dissertation describes two prototype expert systems applied to two problems in nuclear power plants. One problem is spare parts inventory control, and the other one is radionuclide release from containment during severe accident. The expert system for spare parts inventory control can handle spare parts requirements not only in corrective, preventive, or predictive maintenance, but also when failure rates of components or parts are updated by new data. Costs and benefits of spare parts inventory acquisition are evaluated with qualitative attributes such as spare part availability to provide the inventory manager with an improved basis for decision making. The expert system is implemented with Intelligence/Compiler on an IBM-AT. The other expert system for radionuclide release from containment can estimate magnitude, type, location, and time of release of radioactive materials from containment during a severe accident nearly on line, based on the actual measured physical parameters such as temperature and pressure inside the containment. The expert system has a function to check the validation of sensor data. The expert system is implemented with KEE on a Symbolics LISP machine