In the field of fault diagnosis for rotating machines, the conventional methods or the
neural network based methods are mainly single symptom domain based methods, and the
diagnosis accuracy of which is not always satisfactory. To improve the diagnosis accuracy a method
that combines the multi-class support vector machines (MSVMs) outputs with the degree of
importance of individual MSVMs based on fuzzy integral is presented. This provides a sound basis
for integrating the results from MSVMs to get more accurate classification. The experimental
results with the recognition problem of a blower machine show the performance of fault diagnosis
can be improved.