Fault diagnosis approach for bearing based on EMD and slice bi-spectrum

A new approach based on the empirical mode decomposition (EMD) and slice bi-spectrum was presented for fault diagnosis on roller bearings. Original signals were decomposed into a series of intrinsic mode functions (IMFs) of different scales. Envelopes of the IMFs were extracted and a diagonal slice bi-spectrum for the envelopes was computed to extract the non-linear feature deriving from the quadratic phase coupling, as well as the fault characteristic frequencies. An analysis on simulation signals shows that the drawback that traditional envelope spectrum methods based on EMD cannot inhibit the noise can be overcome by this approach. Meanwhile, its computation load is less than traditional high-order spectrum methods. A diagnosis instance of the bearing 6205-2RS JEM SKF was presented to show the feasibility of this approach.