Aggregation of evidence by fuzzy set operations for vibration monitoring

Abstract

Documents work on the design of a diagnostic system, which performs evidence aggregation from many sensors in order to automate the interpretation of vibration spectra. The decision system proposed is an active system, which does not always fuse the decisions from all the sensors available. The decision module is a fuzzy system, performing the generalized mean operation. The parameters of the generalized mean are chosen by a genetic algorithm. The final decision is obtained from the aggregated decision by means of /spl alpha/-cuts. The method proposed is very general and can be used for any problem involving aggregation of decisions for the purpose of classification.<>