In this paper, wavelet analysis is used to detect particular vibration faults. The proposed detection method is based on the stationary wavelet transform. The wavelet coefficients allow analysing the signal changes in different scales. They are fuzzified, leading to partial criteria, relative to the frequency contents in the different frequency bands. Fuzzy aggregation of these partial criteria gives the final decision. The procedure depends on different parameters that must be tuned. Several aggregation possibilities are applied to industrial data from a main drive of a roughing mill. Copyright © 2006 IFAC
