Improved local mean decomposition for modulation information mining and its application to machinery fault diagnosis

Abstract Local mean decomposition (LMD) has been developed for modulation information mining. Its performance is dependent on boundary condition, envelope estimation, and stopping criterion. This paper proposes a soft sifting stopping criterion that enables LMD to achieve a self-adaptive stop for each sifting process. In the proposed method, we define an objective function that considers two characteristics, namely, the root mean square and the excess kurtosis, of the target signal. To optimize this objective function, a heuristic mechanism is proposed to automatically determine the optimal number of sifting iterations. Experimental results on simulated signals demonstrate the effectiveness of the proposed soft sifting stopping criterion for improving the accuracy of LMD, and finally the proposed method is applied to modulation information mining for gear fault diagnosis.