An improved underdamped muti-time-delayed feedback tri-stable stochastic resonance and its research in bearing fault detection

Abstract

Stochastic resonance (SR) has proven to be effective in early fault signal detection. This paper proposes a multi-time-delayed feedback tri-stable stochastic resonance (MTFTSR) method to enhance signal detection capabilities. An improved tri-stable model incorporating a multi-time-delay structure is introduced, followed by the derivation of equivalent potential function and signal-to-noise ratio (SNR) expressions. And the MTFTSR method provides a nested approach for multi-parameter optimization to obtain the best output results. A comprehensive investigation of parameter influence on system performance leads to optimization for maximum system efficiency. Simulation and experimental results demonstrate superior SNR and output waveform clarity compared to envelope analysis and conventional SR methods. The findings highlight the method’s effectiveness in extracting bearing fault features and the method also has certain advantages in detecting gearbox fault signal.