Dynamic analysis of three-dimensional stochastic resonance and its application in signal analysis

Abstract

In response to the difficulty of blind source separation of characteristic signals for mixed signals with severe noise pollution under uncertain observation signals, this paper constructs a multifunctional three-dimensional stochastic resonance (3DSR) system, which can simultaneously perform signal denoising, amplification, and feature restoration processing. The research improves the Ensemble Empirical Mode Decomposition (EEMD) algorithm through the 3DSR system, resulting in obtaining the 3DSR-Ensemble Empirical Mode Decomposition (3DSR-EEMD) algorithm. During the process of signal decomposition, the 3DSR-EEMD algorithm utilizes the 3DSR system to denoise and restore features of the signal, and timely smooth out transient noise interference. This paper proposes a components analysis method of noisy mixed signal based on 3DSR-EEMD. Firstly, the noisy mixed signal is decomposed using the 3DSR-EEMD algorithm. Then, the principal components in the decomposed signals are extracted using the principal component analysis algorithm. Finally, the principal components are processed through the independent component analysis algorithm to obtain the unmixing signals. Apply this method to the component analysis of noisy mixed signals and rolling bearing fault signals. The experimental results show that the components analysis method of noisy mixed signal based on 3DSR-EEMD has good separation performance and noise robustness.