Workshop diagnosis is an essential element in a vehicle´s lifecycle. Customers expect a quick and reasonably priced remedy in the case of a defect. Service technicians have to be supported facing todays´ highly complex cars. Automatical, computer-based analysis of all available information and facts is needed and automated determination of an efficient check strategy is very desirable. Our approach uses decision methods based on the entropy and cost assessment of tests aiming to determine best tests or a ranked list of tests. The domain, the relations between tests and tested components, is described by naive bayesian networks. This article focuses on the automated construction of the bayesian models and in the determination of a decision function for the test evaluation.
Print ISSN: 0178-2312
Volume: 55, 01/2007
Pages: 28 - 34