Enhanced fault diagnosis of segmented asymmetric stochastic resonance in rotating machinery under strong noise environment

In industrial applications, strong noise hampers the extraction of reliable features from mechanical equipment, crucial for detecting faults. Stochastic resonance, unlike other methods, enhances weak signals effectively in noisy environments. However, it often suffers from oversaturation, a common issue when used to improve signal clarity. Therefore, this study introduces a method to prevent saturation with piecewise asymmetric stochastic resonance. A novel potential function is used. This allows the derivation of the output signal-to-noise ratio (SNR) in a bistable system under harmonic excitation. The method effectively manages the conversion of energy states and mitigates the influence of noise through dynamic adjustments to the barrier depth, width, and slope. Furthermore, system parameters are refined using an optimization algorithm to enhance performance and efficiency by optimizing the system response under noise conditions, thereby improving signal detection and reliability. Applied to the bearing fault datasets from Shandong University of Science and Technology, the results indicate that this enhanced method achieves a higher output SNR and a more pronounced peak at the fault characteristic frequency compared to traditional stochastic resonance methods. This study significantly enhances signal processing efficiency and noise tolerance in stochastic resonance, providing more reliable technical support for fault diagnosis in industrial machinery with severe noise interference, thereby improving maintenance efficiency and operational safety.