Weak impact signal detection based on adaptive stochastic resonance with knowledge-based particle swarm optimization

Stochastic resonance is of great importance in the field of signal detection. Suitable system parameters determine the performance of a parameter-induced stochastic resonance detection system. Considering the difficulty of adjusting system parameters and the requirement of real-time detection in the parameter-induced stochastic resonance, knowledge-based particle swarm optimization (KPSO) is proposed to optimize system parameters, which takes the kurtosis index as the fitness function and the property that the impact signal can produce stochastic resonance in a single potential well as the knowledge. Compared with particle swarm optimization (PSO), this algorithm can obtain optimal system parameters more quickly, making energy transfer from the noise to the signal greatly, and produce the best output resonance effect. As a typical large-parameter signal, the impact signal is not satisfied with the stochastic resonance condition apparently. In this paper, we combine the twice sampling with KPSO, realizing weak impact signal detection, and verifying the efficiency and effectiveness of the algorithm.