基于变步长随机共振的弱信号检测技术

Weak Signal Detection Based on Step-Changed Stochastic Resonance

针对绝热近似小参数随机共振难以满足工程实践中大参数下的弱信号检测,以及单一频率的共振分析在实际应用中的局限性问题,提出了一种变步长随机共振数值算法．该方法通过调整计算步长,使随机共振理论同时适用于大、小参数条件下的弱信号特征提取．计算机仿真结果表明,对变步长随机共振后的信号作幅值谱和小波分析,均能准确得到低信噪比信号中的多个有用成分,充分证明该算法在大参数条件下可对弱信号中的多个特征频率产生共振输出．同时,变步长随机共振也可以有效抑制信号小波分解中由强噪声引起的边频干扰,提高小波分析在低信噪比信号检测中的可靠性．

The stochastic resonance analysis of single-frequency weak signals has limited engineering applications because adiabatic elimination stochastic resonance within small parameters can＇t detect weak signals in large parameters, and engineering signals usually have muhi-frequency features. For this reason, a numerical method called step-changed stochastic resonance was proposed. By adjusting the calculating step, the stochastic resonance method can adapt to weak signal detection in both small and large parameters. Computer simulation results show that the features of multi-frequency weak signals overwhelmed in heavy noise can be detected by step-changed stochastic resonance in both spectrum and wavelet results. Additionally, step-changed stochastic resonance not only decreases the weak signal＇s distortion induced by heavy noise in wavelet analysis, but also improves the reliability of wavelet analysis in weak signal detection under low ratios of signal to noise.