Application of improved MCKD method based on QGA in planetary gear compound fault diagnosis

An improved maximum correlated kurtosis deconvolution (MCKD) method based on quantum genetic algorithm (QGA) named QGA-MCKD is proposed, which can be used for gear and bearing compound fault diagnosis. Two key parameters, filter length (L) and deconvolution period (T) of MCKD, corresponding to each single fault are adaptively selected by QGA. MCKD is set by the obtained key parameters to process the compound fault signal, and each single fault feature related to the single failed part can be extracted. QGA-MCKD was applied to process the simulated and experimental compound fault signals of planetary gear tooth breakage and bearing rolling element damage, and the gear and bearing fault signals were extracted, respectively. Then power spectrum analysis of gear fault signal and envelop spectrum analysis of bearing fault signal were carried out to diagnose the compound faults. The superiority of QGA-MCKD was verified in comparison with direct spectrum analysis and ensemble empirical mode decomposition (EEMD). The stability of QGA-MCKD was verified in the compound fault diagnosis of gear tooth wear and bearing outer race fault. Results show that QGA-MCKD has a good effectiveness in improving the accuracy of gearbox gear and bearing compound fault diagnosis. (C) 2019 Elsevier Ltd. All rights reserved.