Mengdi Li

Stochastic resonance, recognized as a noise-enhanced signal processing technique, has found extensive applications in fault diagnosis, particularly in weak signal detection. This paper proposes a high-dimensional space coupled piecewise bistable stochastic resonance (HSCPBSR) model. This approach leverages the advantages of high-dimensional systems to address limitations encountered in traditional stochastic resonance models. The HSCPBSR model systematically investigates the relationship between system parameters and the output signal-to-noise ratio (SNR) across varying dimensions. Analysis reveals that outputs from coupling ends with identical parameters exhibit consistent characteristics, and there exists an optimal dimensionality that maximizes the output SNR for a specific input. This finding underlines the ability of the model to adaptively optimize its configuration for enhanced signal processing performance. Extensive simulations validate the HSCPBSR model’s capability to significantly enhance weak signal features while maintaining strong robustness against noise interference. Furthermore, the method is applied to diagnose faults in two bearing types, with the optimal dimension identified for maximum SNR. Results show that the proposed approach surpasses one-dimensional and two-dimensional piecewise bistable stochastic resonance models in detecting weak fault features, demonstrating its practical effectiveness.