No abstract

No abstract

A double modulation phenomenon is pointed out in the wind turbine gearbox compound fault signals, which consists of a resonance modulation frequency band and an asymmetric modulation frequency band. The modal aliasing and the double modulation phenomena make it difficult for empirical wavelet transform to obtain the spectral components in the meshing modulation regions. The main contribution is to establish an optimal empirical wavelet transform (OEWT) framework for weak feature extraction of compound fault, which incorporates optimal velocity synchronous linear chirplet transform (OVSLCT) and simulated annealing (SA) algorithm to obtain the optimal parameters of the filter bank. The frequency boundary of the fault component is obtained through OVSLCT rather than the extreme value of the spectrum. A filter transition band width optimization scheme is proposed, in which the optimal transition band width of the filter bank in EWT is optimized by SA. Compared with the original EWT and VMD, OEWT can better obtain the compound fault characteristic frequency and solve the mode aliasing problem.

Tool wear status seriously affects the dimensional accuracy and surface quality of the machined parts. Therefore, tool condition monitoring (TCM) is essential in the milling process of aerospace structural parts due to the use of difficult-to-cut materials and complex cutting trajectories. The Higher order spectrum (HOS) was first employed to analyze the vibration signals, and then bispectral features extracted from de-noised signals were used to characterize the tool wear status. The improved Pearson’s correlation coefficient was used for feature selection to reduce the influence of periodic components on feature selection process. Furthermore, a novel objective function was proposed to guide the hyperparameters optimization process of support vector machine based on Bayesian optimization algorithm, in which the effect of imbalanced data on the recognition rates was considered. To demonstrate the effectiveness of the proposed method, a structural part milling experiment was performed on a vertical machining center and vibration signals of spindle were collected. Based on this, an online TCM model was established. The present study suggests that the proposed TCM system is accurate and robust.