Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Efficient Diagnostic Search Via Information Entropy Reduction
Abstract This paper presents a methodology and algorithm that optimize a diagnostic search in terms of the expected time required to reach the conclusion, i.e., to find a failed component. It is assumed that the fault tree describing the relationship among the events is available. The procedure allows one to locate the best candidate among the fault tree events for information acquisition in the following sense. Once the event state is ascertained and this knowledge is propagated through the fault tree, the uncertainty as to which component is in a failed state is minimized. This procedure is based on expected information entropy reduction and, when recurrently applied, should optimize a diagnostic search. The entropy reduction algorithm proved to be time-consuming for large scale systems. In order to make the algorithm suitable for real time applications, the knowledge structure had to be changed. A Transformed Fault Tree (TFT) knowledge structure was introduced. It allows almost instantaneous assessment of fault tree events regarding expected entropy reduction. The above methodology is implemented in a man-machine interface module of Combustion Engineering’s Generic Diagnostic System.
Efficient Diagnostic Search Via Information Entropy Reduction
Abstract This paper presents a methodology and algorithm that optimize a diagnostic search in terms of the expected time required to reach the conclusion, i.e., to find a failed component. It is assumed that the fault tree describing the relationship among the events is available. The procedure allows one to locate the best candidate among the fault tree events for information acquisition in the following sense. Once the event state is ascertained and this knowledge is propagated through the fault tree, the uncertainty as to which component is in a failed state is minimized. This procedure is based on expected information entropy reduction and, when recurrently applied, should optimize a diagnostic search. The entropy reduction algorithm proved to be time-consuming for large scale systems. In order to make the algorithm suitable for real time applications, the knowledge structure had to be changed. A Transformed Fault Tree (TFT) knowledge structure was introduced. It allows almost instantaneous assessment of fault tree events regarding expected entropy reduction. The above methodology is implemented in a man-machine interface module of Combustion Engineering’s Generic Diagnostic System.
Efficient Diagnostic Search Via Information Entropy Reduction
Filshtein, Eugene L. (Autor:in)
01.01.1988
9 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
| Springer Verlag | 2013
Efficient Search Algorithms for Route Information Services of Direct and Connecting Transit Trips
| British Library Online Contents | 1992
Information Search in Databases
| British Library Conference Proceedings | 1994