Particle Swarm Optimization-Based Variable Scale Asymmetric Stochastic Resonance Bearing Diagnostic Method

Abstract A diagnostic method for bearing faults, centered around the extraction and identification of diagnostic signals, is introduced. This method utilizes a Particle Swarm Optimization (PSO) algorithm to optimize a variable-scale asymmetric stochastic resonance (SR) framework. The PSO algorithm dynamically fine-tunes the parameters of the asymmetric stochastic resonance system to align more effectively with the demands of bearing fault diagnosis. An asymmetric factor-controlled potential function for the stochastic resonance system is established, using the Signal-to-Noise Ratio Improvement (A-SNRI) of the fault signal as the objective function for the optimization algorithm. The PSO algorithm is employed for global optimization to adjust the structural parameters a 0 , b 0 and the asymmetric factor of the asymmetric α bistable stochastic resonance system. Simulations and experimental validations are conducted using the optimized stochastic resonance system parameters, demonstrating the robustness and effectiveness of the algorithm through the extraction of fault characteristic frequencies. Experimental results indicate the proposed bearing fault diagnostic method can stably extract fault characteristic frequencies, effectively filter out noise, and the extracted fault frequencies align with theoretical values.