Application of CMRDE in bearing fault diagnosis

When a rolling bearing fails, its vibration signal often has nonlinear and nonstationary characteristics. Reverse dispersion entropy ( RDE ) can effectively measure complexity change and nonlinear dynamic catastrophe behavior of vibration signals，but single scale RDE value can ' t fully reflect complexity and nonlinear characteristics of vibration signals. Here, inspired by multi-scale entropy and aiming at shortcomings of traditional multi-scale coarsening mode, a composite multi-scale reverse dispersion entropy (CMRDE) was proposed. Through simulation signal analysis, CMRDE was compared with multi-scale reverse dispersion entropy ( MRDE ) and RDE. The results showed that CMRDE can not only reflect difference of signal complexity at different scales, but also change more smoothly and fluctuate less. Then, CMRDE was applied in fault feature extraction of rolling bearing, a rolling bearing fault diagnosis method based on CMRDE, ensemble empirical mode decomposition ( EEMD) and support vector machine optimized with Cuckoo search (CS-SVM) algorithm was proposed. Finally, the proposed method was applied to analyze rolling bearing experimental data, and the results were compared with those obtained using the existing methods. It was shown that compared with other methods，the proposed method can effectively identify bearing fault types，the extracted fault feature errors are smaller, and its fault recognition rate is higher. © 2022 Chinese Vibration Engineering Society. All rights reserved.