Multichannel Fault Diagnosis of Planetary Gearbox Via Difference-Average Symbol Transition Entropy and Twin Support Higher-Order Tensor Machine

This article proposes a novel multichannel fault recognition strategy for planetary gearboxes using difference-average symbol transition entropy (DASTE) and twin support higher-order tensor machine (TwSHTM). In the proposed hybrid framework, a new feature representation method, namely DASTE is first proposed to capture rich information inherent in the difference frequency bands of the original data by introducing difference operators and average operators into the symbol transition entropy (STE). The three-order tensor representation is then constructed through stacking DASTEs extracted from each single-channel data. Finally, to further improve the diagnostic accuracy, a new tensor-based classifier, namely TwSHTM is presented, which can directly process the constructed tensor representations without vectorization or fusion. The superiority of the presented methods is confirmed through simulation and experimental cases, and the results suggest that the presented strategy yields higher diagnostic precision and better generalization compared with existing entropy-based multichannel fault recognition schemes.