Maintenance work scheduling aid for nuclear power plants

Abstract

A prototype system for computer-aided scheduling of maintenance work for nuclear power plants has been developed using an artificial intelligence technique. The system functions were tested by offline simulations. This system checks for interferences between maintenance tasks or between maintenance tasks and plant operations, finds isolation procedures, and makes the optimum schedule. The hierarchical plant structure data, which consist of physical and functional form knowledge, make it possible to check for interference relations even if some data are lacking. The branch-and-bound method was used for the optimum scheduling. The system configuration is described, along with knowledge-representation and inference issues.<>