Fault diagnosis method of planetary gearbox based on acoustic signal IACCUGRAM-AMSESHIRD

To overcome the difficulties in extracting fault features from acoustic signals of planetary gearboxes, this paper introduces a novel fault diagnosis method combining Improved ACCUGRAM (IACCUGRAM) and Adaptive Modified Square Envelope Spectrum Harmonic Interference Ratio Deconvolution (AMSESHIRD). Initially, the feature extraction process in the ACCUGRAM algorithm is refined using an enhanced Gini index. Given the challenge of deconvolution as a non-convex optimization problem prone to local optima, a tailored filter specific to the fault characteristic band identified by IACCUGRAM is developed to initiate the deconvolution process, thus improving efficiency and accuracy in signal interpretation. Additionally, the harmonic interference ratio of the square envelope spectrum is weighted and pre-processed, with the fault period and filter length optimized using a fitness function formulated from a modified smoothness index and a fault characteristic index, optimized through the African Vulture Optimization Algorithm. Deconvolution is then executed using the custom initial filter, pre-optimized criteria, and optimal parameters determined. The final stage involves fault diagnosis of the planetary gearbox through envelope spectrum analysis of the acoustic signal. The effectiveness and superiority of this method are validated through both simulated and experimental signals, with results demonstrating significant improvements over conventional methodologies.