Complete ensemble all time-scale decomposition method and its application in face gear fault diagnosis

As a transmission structure capable of achieving efficient power transmission in multiple directions, face gears have significant application value and prospects. However, there is little research on fault diagnosis technology specific to face gears. Therefore, an excellent signal decomposition method is urgently needed for fault diagnosis of face gear. Traditional methods for signal decomposition, including empirical mode decomposition, struggle to accurately extract fault feature information from face gears due to the issue of mode mixing. To address this problem, we lately proposed a novel signal decomposition method called all time-scale decomposition (ATD). Not only the extreme points construct the baselines of ATD, but also the zero-crossing points are involved, which can effectively mine the feature information at different local scales. While ATD overcomes mode mixing arising from closeness of component center frequencies, its decomposition performance is impacted by anomalous signals. Consequently, combining the concept of improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) for noise assistance and anomalous component detection of complete ensemble local characteristic-scale decomposition (CELCD), this paper further proposes the complete ensemble all time-scale decomposition (CEATD) method based on the ATD method. CEATD can decompose anomalous components through noise ensemble averaging and detect these anomalous components by permutation entropy. The analysis results of simulations and experiments demonstrate that the CEATD method can effectively overcome mode mixing caused by intermittent signals, noisy signals, and closeness of component center frequencies. In face gear fault diagnosis, CEATD can accurately extract the fault mode components.