摘要

A more effective fuzzy membership function based on samples' affinity was proposed for the measurement of samples uncertainty by synthesizing the advantages of the fuzzy theory and SVM. The method of FSVM (fuzzy support vector machine) was applied to vibration fault analysis of some aeroengines and a better diagnosis mathematical model was established based on FSVM, which was then used in fault diagnosis of aeroengine whole-body vibration. The results reveal that the method has better diagnosing accuracy and better ability to reduce noise.