Similarity Based System Reconfiguration by Fuzzy Classification and Hierarchical Interpolate Fuzzy Reasoning

In case of redundant systems some kind of faults could be tolerated simply by changing the rulebase of the control system - by a system (strategy) reconfiguration. One solution for this kind of system reconfiguration could be the combination of the fuzzy clustering based symptom evaluation and the hierarchical interpolate fuzzy reasoning - the similarity based system reconfiguration. Its main idea is the following: More similar the actual symptom to one of the system behaviour classes, more similar must be the final conclusion to the decision done by the strategy handling that system behaviour class. A similarity based system reconfiguration method and as an example of its practical application, fault diagnosis and reconfiguration of a simplified three-tank benchmark system is introduced in this paper.