Parametric adaptive VMD method and its applications on early fault diagnosis for rolling bearing

Abstract

This paper proposes an improved parametric adaptive variational mode decomposition (VMD) method, in view of the problem with the traditional VMD method in defining parameters the mode number K and the penalty parameter α. First, taking fault impacts and periodicity into consideration, this paper adopts the Gini index and spectrum peak ratio (SPR) index as the basis to construct a weighted spectrum peak ratio (WSPR) index. After that, it takes the maximization of the WSPR index as the optimization objective, using the whale optimization algorithm (WOA) to find the optimal parameter combination for K and α, and avoids the disadvantages of subjective parameters selection. Finally, it follows the maximum WSPR principle to automatically select a sensitive intrinsic mode function (IMF) and extract fault characteristics via envelope demodulation analysis. The proposed method is verified through experimental signal. As shown by the results, this method can effectively extract weak characteristics of early fault signals and offer the accurate judgment of fault types.