Abstract
At the OECD Halden Reactor Project work has been going on for several years in the field of automatic fault diagnosis for nuclear power plants. Continuing this work, studies are now carried out to combine different diagnostic systems within the same framework. The goal is to establish a general knowledge structure for diagnosis applied to a NPP process. Such a consistent and generic storage of knowledge will lighten the task of combining different diagnosis techniques. An integration like this is expected to increase the robustness and widen the scope of the diagnosis. Further, verification of system reliability and on-line explanations of hypotheses can be helped. Last but not least there is a potential in reuse of both specific and generic knowledge. The general knowledge framework is also a prerequisite for a successful integration of computerized operator support systems within the process supervision and control complex. Consistency, verification and reuse are keywords also in this respect. Systems that should be considered for integration are; automatic control, computerized operator procedures, alarm - and alarm filtering, signal validation, diagnosis and condition based maintenance. This paper presents three prototype diagnosis systems developed at the OECD Halden Reactor Project. A software arrangement for process simulation
More>>
Steinar Brendeford, T
[1]
- Institutt for Energiteknikk, Halden (Norway). OECD Halden Reaktor Projekt
Citation Formats
Steinar Brendeford, T.
General knowledge structure for diagnosis.
IAEA: N. p.,
1996.
Web.
Steinar Brendeford, T.
General knowledge structure for diagnosis.
IAEA.
Steinar Brendeford, T.
1996.
"General knowledge structure for diagnosis."
IAEA.
@misc{etde_454190,
title = {General knowledge structure for diagnosis}
author = {Steinar Brendeford, T}
abstractNote = {At the OECD Halden Reactor Project work has been going on for several years in the field of automatic fault diagnosis for nuclear power plants. Continuing this work, studies are now carried out to combine different diagnostic systems within the same framework. The goal is to establish a general knowledge structure for diagnosis applied to a NPP process. Such a consistent and generic storage of knowledge will lighten the task of combining different diagnosis techniques. An integration like this is expected to increase the robustness and widen the scope of the diagnosis. Further, verification of system reliability and on-line explanations of hypotheses can be helped. Last but not least there is a potential in reuse of both specific and generic knowledge. The general knowledge framework is also a prerequisite for a successful integration of computerized operator support systems within the process supervision and control complex. Consistency, verification and reuse are keywords also in this respect. Systems that should be considered for integration are; automatic control, computerized operator procedures, alarm - and alarm filtering, signal validation, diagnosis and condition based maintenance. This paper presents three prototype diagnosis systems developed at the OECD Halden Reactor Project. A software arrangement for process simulation with these three systems attached in parallel is briefly described. The central part of this setup is a `blackboard` system to be used for representing shared knowledge. Examples of such knowledge representations are included in the paper. The conclusions so far in this line of work are only tentative. The studies of existing methodologies for diagnosis, however, show a potential for several generalizations to be made in knowledge representation and use. (author). 14 refs, 6 figs.}
place = {IAEA}
year = {1996}
month = {Dec}
}
title = {General knowledge structure for diagnosis}
author = {Steinar Brendeford, T}
abstractNote = {At the OECD Halden Reactor Project work has been going on for several years in the field of automatic fault diagnosis for nuclear power plants. Continuing this work, studies are now carried out to combine different diagnostic systems within the same framework. The goal is to establish a general knowledge structure for diagnosis applied to a NPP process. Such a consistent and generic storage of knowledge will lighten the task of combining different diagnosis techniques. An integration like this is expected to increase the robustness and widen the scope of the diagnosis. Further, verification of system reliability and on-line explanations of hypotheses can be helped. Last but not least there is a potential in reuse of both specific and generic knowledge. The general knowledge framework is also a prerequisite for a successful integration of computerized operator support systems within the process supervision and control complex. Consistency, verification and reuse are keywords also in this respect. Systems that should be considered for integration are; automatic control, computerized operator procedures, alarm - and alarm filtering, signal validation, diagnosis and condition based maintenance. This paper presents three prototype diagnosis systems developed at the OECD Halden Reactor Project. A software arrangement for process simulation with these three systems attached in parallel is briefly described. The central part of this setup is a `blackboard` system to be used for representing shared knowledge. Examples of such knowledge representations are included in the paper. The conclusions so far in this line of work are only tentative. The studies of existing methodologies for diagnosis, however, show a potential for several generalizations to be made in knowledge representation and use. (author). 14 refs, 6 figs.}
place = {IAEA}
year = {1996}
month = {Dec}
}