Application of t-Distribution Stochastic Neighbor Embedding (t-SNE) And VMD In Fault Feature Extraction

The running state of the rolling bearing directly affects the overall mechanical performance. Fault detection has become an important research content at the present stage. Based on the need of extracting the eigenvalues of the early weak fault information, this paper proposes a method of constructing high-dimensional feature data based on phase space reconstruction and denoising data by, the signal after the noise of Variational Mode Decomposition (VMD), after noise reduction, the signal is decomposed by VMD, and the method of selecting and reconstructing the modal component with kurtosis and envelope entropy as the comprehensive evaluation index is proposed. The reconstruction component of the optimal modal component is obtained, and then the envelope spectrum analysis of the optimal modal component is carried out to extract the fault characteristic frequency. The effectiveness of this method is verified by analyzing the fault signal of rolling bearing.