Fault Detection of Gear using Periodicity-Induced Overlapping Group Shrinkage.

This paper addresses the weak fault periodic oscillatory feature extraction from vibration signals using Periodicity-induced Overlapping Group Shrinkage (POGS). Incipient signals are considered as periodic group sparse signals submerged in noise. Based on the minimization of a convex cost function involving a group sparsity promoting penalty function and the principle of majorization-minimization, a fast iterative algorithm is engaged for its solution. The method is applied to analyse gearbox vibration signals which was acquired from a machinery fault simulator. The results demonstrate that the method can satisfactorily reveal periodic impulses. [ABSTRACT FROM AUTHOR]