Abstract: The phenomenon that noise enhancing weak period signals in Non-linear stochastic resonance systems can be utilized detecting weak communication signals in strong noise background. A new method of adaptive stochastic resonance based on artificial-fish swarm optimization is proposed to handle narrowband PSK signals in this article. The promotions in PSK signal processing are introduced in three aspects: Application in narrowband PSK signals parameter estimation, optimizing the performance of the SNR threshold to detect narrow-band PSK signals and improvement inband signal to noise ratio (SNR) of PSK signals. The influence of sampling rate in stochastic resonance of the narrow-band PSK signals is exposed in this article.The result of theoretical analysis and simulation experiments show this method of adaptive stochastic resonance based on artificialfish swarm optimization handling PSK signals can give evident promotion in the performance of parameter estimation and lower the detecting SNR threshold by 3~6 dB on contrast to traditional methods based on spectral lines. Besides, it has good convergence performance. Meanwhile, stochastic resonance is verified useless in detecting narrowband PSK signals directly.