A new bearing fault diagnosis method based on MM and EMD

Mathematical morphology (MM) is a very effective signal processing tool which has been widely applied in 2-dimentional signal processing such as image processing. But there is little application in mechanic signals. Empirical mode decomposition (EMD) is a novel self-adaptive method which is based on partial characters of the signal. This paper firstly proves the efficiency of MM in 1-dimentional signal by illustrating an example of a simulation signal. Then in practical application, it combines MM and EMD together to diagnosis bearing fault. First de-noise the practical bearing fault signal by MM filter, and then decompose the de-noised signal into several IMFs by EMD. Finally calculate the power spectral density of each IMF. The result indicates the efficiency of this method.