New Weak Signal Detection Method Based on Adaptive Parameter-induced Stochastic Resonance

Abstract

Considering the requirements of real-time of weak signal detection, the property that bistable system produces the best resonance effect when Kramer rate is two times as the signal modulation frequency is regarded as knowledge. The system parameters are optimized by Knowledge-based genetic algorithm (KGA), which takes respectively the signal-noise-ratio of the output and the mean signal-noise-ratio of the output as the fitness function. Compared with the GA, this algorithm can obtain the optimal structure parameters of adaptive stochastic resonance system more quickly and make the bistable system attain in an optimal state of stochastic resonance. Simulated experiments show that the proposed detection method is highly efficient for parameter optimization and enable to detect the single frequency weak signal and multi-frequency weak signal from heavy noise. The proposed method is further verified the effectiveness by the engineering application, which extracted the vibration source signal of the single crystal furnace.