Principle and application of variable-scale tristable stochastic resonance for fractional power systems
Qiang Ma,Ran Peng,Zhichong Wang,Kai Yang
DOI: https://doi.org/10.1140/epjp/s13360-023-04806-4
2024-01-13
The European Physical Journal Plus
Abstract:To address the limitations of the traditional stochastic resonance in signal-parameter matching, we propose an enhanced tristable potential function model based on the classical tristable system. At the core of this new model lies the system power denoted as α , which can assume both integer and non-integer values. By adjusting the system power α , one can achieve a more precise optimal resonance output. As noise intensity varies, α can be tuned to generate various potential function models with different good depths and shapes. The evaluation index for determining the optimal value of α is the cross-correlation coefficient. Additionally, a scale transformation method is introduced to enhance the capability of stochastic resonance in handling high-frequency signals. We employ the cockroach optimization algorithm, to optimize the system parameters and scale coefficients within the stochastic resonance, resulting in an adaptive stochastic resonance. Following numerous experimental validations, the fractional power stochastic resonance system exhibited superior efficacy in diagnosing bearing faults. Specifically, this system effectively extracts fault signals from rolling bearings in noisy environments, demonstrating its effectiveness in rolling bearing fault diagnosis.
physics, multidisciplinary