Verfahren der automatisierten Diagnose technischer Systeme

Klaus Becker

DFKI DFKI Documents (D) 92-18 1992.


This thesis deals with the introduction and judgement of different kinds of methods of diagnosis for technical systems. Theoretical concepts as well as practical realizations are discussed with regard to their suitability for technical diagnosis. Diagnosis systems can be divided into the following categories: heuristic, model-based, case-based, statistic, decision-trees and decicion-tables. Most industrial applications are heuristic diagnosis systems. But especially for technical application areas model-based methods seem to be suitable. No method of diagnosis is absolutely suited to every application area. Therefore, for every new application it is necessary to decide, which method of diagnosis to choose. In many cases different methods of diagnosis are melted together in order to take advantage of the advantages of any single method and to avoid the disadvantages. After the presentation of theoretical concepts and practical realizations the thesis closes with a small practical part: To verify some theoretical results a small system for diagnosing faults of the turning process has been implemented. Additionally, it has been examined, whether a general domain theory serving as a base for the construction, was useful.

D-92-18.pdf (pdf, 8 MB )

German Research Center for Artificial Intelligence
Deutsches Forschungszentrum für Künstliche Intelligenz