Abstract | ||
---|---|---|
Although the use of an expert system for the detection and location of faults in the power supply subsystem of the TDFI satellite 1 1 Research activity ART162 carried out by Aérospatiale and Dassault Electronique for the CNES (National Center for Space Research). gives good practical results, there are also a certain number of drawbacks: • - the rules are application-dependent and therefore useless for the diagnosis of other satellites, • - the amount of knowledge required is quite large and contains a lot of redundant information, • - testing and updating of the system are tricky, • - it is far from sure that all possible faults can be taken into account. The diagnosis theory based on models of correct behavior of the components introduced more recently by de Kleer and Williams seems more suited to this type of problem. Using our program CATS/DIANA, we have been able to validate this approach for analog electronic circuits. We have also tested the method on the diagnosis of satellite power supply subsystems; the results were similar to those obtained using an expert system, but without the drawbacks. This paper describes the two methods and compares the results. |
Year | DOI | Venue |
---|---|---|
1992 | 10.1016/0167-739X(92)90050-L | Future Generation Comp. Syst. |
Keywords | Field | DocType |
Expert system,model-based diagnosis,electrical system | Electrical network,Computer science,Expert system,Electric power system,Fault tolerance,Artificial intelligence,Electronic circuit,Reliability engineering,Distributed computing | Journal |
Volume | Issue | ISSN |
7 | 4 | Future Generation Computer Systems |
Citations | PageRank | References |
0 | 0.34 | 3 |
Authors | ||
3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Philippe Devés | 1 | 36 | 7.63 |
Cécile Fischer | 2 | 0 | 0.34 |
Patrick Taillibert | 3 | 77 | 15.76 |