Takayasu Kasahara

Automatic Scheduling of Maintenance Work in Nuclear Power Plants

An automatic scheduling method for maintenance work in nuclear power plants has been developed using an AI technique. The purpose of this method is to help plant operators by adjusting the time schedule of various kinds of maintenance work so that incorrect ordering or timing of plant manipulations does not cause undersirable results, such as a plant trip. The functions of the method were tested by off-line simulations. The results showed that the method can produce a satisfactory schedule of plant component manipulations without interference between the tasks and plant conditions.

e-chem page: A Support System for Remote Diagnosis of Water Quality in Boiling Water Reactors

It is important to control and maintain water quality for nuclear power plants. Chemical engineers sample and monitor reactor water from various subsystems and analyze the chemical quality as routine operations. With regard to controlling water quality, new technologies have been developed and introduced to improve the water quality from both operation and material viewpoints. To maintain the quality, it is important to support chemical engineers in evaluating the water quality and realizing effective retrieval of stored data and documents. We have developed a remote support system using the Internet to diagnose BWR water quality, which we call e-chem page . The e-chem page integrates distributed data and information in a Web server, and makes it easy to evaluate the data on BWR water chemistry. This system is composed of four functions: data transmission, water quality evaluation, inquiry and history retrieval system, and reference to documents on BWR water chemistry. The developed system is now being evaluated in trial operations by Hitachi, Ltd. and an electric power company. In addition diagnosis technology applying independent component analysis (ICA) is being developed to improve predictive capability of the system. This paper describes the structure and function of the e-chem page and presents results of obtained with the proposed system for the prediction of chemistry conditions in reactor water.Copyright

Domain model for constructing a knowledge based system

An expert system tool, ASPROGEN (automatic search program generator), is described. It has a domain model editor, by which tool users can define their problems in an object-oriented way, and a built-in function for automatic generation of a domain-specific inference program. This function uses search-based program specifications and an abstract data type of search. The descriptiveness of this domain model is shown by applying ASPROGEN to a scheduling problem.<<ETX>>

Method for detecting and removing interferences among maintenance tasks.

A method, based on AI techniques, has been developed for detecting and removing interferences among maintenance tasks in nuclear power plants. This method helps plant operators by adjusting the time schedule of the various tasks so that incorrect ordering or timing of plant manipulations does not cause undesirable results, such as a plant trip. This method consists of three main parts.(1) Checking for interference among tasks and plant conditions by analyzing effects of manipulating one component on other components.(2) Searching for a method to remove interference by isolating the problem component from the others.(3) Adding new operations and readjusting the time schedule by the branch and bound method.The performance of this method was checked by off-line simulation test. The result showed that the method can examine and remove the interferences among maintenance tasks.