JOPSS - Search Results

Since it is desired to enhance availability and safety of nuclear power plants operation and maintenance by removing human factor, there are many researches and developments for intelligent operation or diagnosis using artificial intelligennce (AI) technique. We have been developing an autonomous operation and maintenance system for nuclear power plants by substituting AI's and intelligent robots. It is indispensable to use various and large scale knowledge relative to plant design, operation, and maintenance, that is, whole life cycle data of the plant for the autonomous nuclear power plant. These knowledge must be given to AI system or intelligent robots adequately and opportunely. Moreover, it is necessaly to insure real time operation using the large scale knowledge base for plant control and diagnosis performance. We have been studying on the large scale and real time knowledge base system for autonomous plant. In the report, we would like to present the basic concept and expecting performance of the knowledge base for autonomous plant, especialy, autonomous control and diagnosis system.

JAEA Reports

None

PNC TJ7694 96-001, 171 Pages, 1996/02

PNC-TJ7694-96-001.pdf:6.53MB

no abstracts in English

JAEA Reports

Development of program for arbitrary real time simulation; (1)Design of prototype program

kasahara, Naoto;

PNC TN9410 95-211, 32 Pages, 1995/08

PNC-TN9410-95-211.pdf:1.54MB

In order to optimize elevated temperature structural systems in fast reactor plants, where main loading is thermal stress induced by transient operation of circuit, authors proposed new design frame by applying design by analysis concept to both structural design and system design. A key technology in this design frame is an integrated analysis method for both thermo-mechanical behaviors of structures and plant thermo-hydraulic dynamics, developing of a prototype code of which, named PARTS (Program for Arbitrary Real Time Simulation), was started this year. For the purpose to achieve flexible coupling of several codes, authors designed three categories of calculation parts (objects): (1)thermo-hydraulics of coolant, (2)thermo-mechanical behavior of structures, and (3)material strength. These calculation parts can be handled and connected easily on the PARTS-Workbench. Real time simulation is planed to be accomplished by parallel processing of individual parts calculation, and by prediction of neural-network which learned past calculation results. Object oriented languages, Smalltalk and C++, were adopted for implementation of calculation parts. The PARTS-Workbench was programed by visual Basic and Visual Smalltalk for considering user customization. In the next phase of study, parallel processing function and neural network parts will be incorporated in the PARTS code. Prototype of this code is going to be completed until F.Y.1996. and be applied to study of thermal mitigation structures.

JAEA Reports