A heated and challenging area of fault diagnostic research has been separating the rolling bearing problem feature information from severe noise disruption. The Symplectic Geometry Mode Decomposition (SGMD) has been effectively applied to rolling bearings with the advantage that it does not involve the subjective definition of parameters and eliminates noise when efficiently reconstructing the modes. Yet, it possesses the following shortcomings the invalid symplectic geometry component (SGC) affects the decomposition accuracy, and the physical meaning of the decomposition results is unclear. Inspired by nonparametric adaptive signal decomposition methods such as SGMD and Matching Pursuit (MP), the paper proposes the Symplectic Sparsest Mode Decomposition (SSMD) method. SSMD first constructs a library of symplectic geometry atoms and constrains the number of atoms by energy threshold, which effectively improves the decomposition speed; then, the quality of atoms and the robustness of the algorithm are improved by initialized adaptive noise reduction of symplectic geometry atoms; finally, symplectic geometry atoms are optimally reconstructed by genetic algorithms with the regularized locally narrowband operator as the optimization target, which obtains the sparsest solution of symplectic geometry mode components while constraining the decomposition result to be locally narrowband signals, so as to make better physical meaning of decomposition results. The comparative analysis results of simulation and experiment show that SSMD has obvious advantages in decomposition accuracy and noise robustness, and is more efficient relative to the decomposition of SGMD.
Extracting the fault characteristic information of rolling bearings from intense noise disturbance has been a heated research issue. Symplectic geometry mode decomposition (SGMD) has already been adopted for bearing fault diagnosis due to its advantages of no subjective customization of parameters and the ability to reconstruct existing modes. However, SGMD suffers from rapidly decreasing calculation efficiency as the amount of data increases, in addition to invalid symplectic geometry components affecting decomposition accuracy. The regularized composite multiscale fuzzy entropy (RCMFE) operator is constructed to evaluate the complexity of each initial single component and minimize the residual energy. Combined with the partial reconstruction threshold indicator to filter out specific significant initial single components, the raw signal can be decomposed into multiple physically meaningful symplectic geometric mode components. Therefore, the decomposition efficiency and accuracy can be enhanced. Thus, a rolling bearing fault diagnosis method is proposed based on partial reconstruction symplectic geometry mode decomposition (PRSGMD). Both simulated and experimental analysis results show that PRSGMD can improve the speed of SGMD analysis while increasing the decomposition accuracy, thereby augmenting the robustness and effectiveness of the algorithm.
It is always a hot and challenging problem to extract the characteristic information of roller bearings from strong noise interference. Conventional Hilbert-Huang Transform (HHT), Local Mean Decomposition (LMD), Local Feature- Scale Decomposition (LCD), and so on have some issues like overenvelope, under-envelope, frequency-chaos, end-point effect, and so on. SGMD is one of the most efficient approaches to reconstruct this model. But SGMD has a drawback that the computation efficiency is reduced quickly with an increase in the quantity of data, and the degradation precision is influenced by the non-valid Symplectic Geometric Component (SGC). On this basis, a RCMFE is proposed, which is used to estimate the complexity of the reconstructed original individual parts and restrict the minimum amount of remaining power. This paper presents a PRSGMD approach. The simulation results indicate that PRSGMD can not only enhance the precision of SGMD but also enhance its robustness and validity. Finally, a Maximal Distance Assessment (DET) is employed in combination with a more interpretable tree-based Light Gradient Boosting Machine (LightGBM) for the intelligence fault diagnosis for rolling bearing.