Sensitive periodic mode decomposition: an effective method for extracting periodic pulse features

The traditional pulse evaluation indexes are sensitive to random pulses and periodic harmonics, and cannot accurately quantify the bearing fault characteristics under the cover of strong background noise. Therefore, the reweighted double domain negative entropy (RDDNE) index is defined, which cannot only keep the sensitivity to fault periodic pulses without prior knowledge, but also has strong robustness to random pulses and periodic harmonics. Based on this, a new method for extracting periodic pulses is proposed, which is called sensitive periodic mode decomposition (SPMD). On the one hand, the SPMD method takes the maximum RDDNE as the optimal goal of decomposition, accurately locks the characteristic information of periodic pulses, and realizes adaptive extraction of weak periodic pulse components. On the other hand, the SPMD method iteratively updates the filter coefficients by means of no prior knowledge, which greatly improves the noise robustness. Compared with the existing decomposition methods, SPMD method can guarantee to focus on the periodic pulse feature information without any prior knowledge. The effectiveness and superiority of the proposed method are verified by bearing simulation and experimental signal analysis.