采用主分量分析方法研究随机共振

噪声能量向信号能量的转换是随机共振具有的独特优点,也是实际应用随机共振进行弱信号检测的基础。由于随机共振系统只在合适量的噪声输入时产生最大的信噪比输出(共振),过多、过少的噪声都会降低系统的性能,因此,如何有效地控制输入到随机共振系统中的噪声十分重要。本文首次提出采用主分量分析方法研究随机共振,其基本思路是分解信号和噪声能量到相互正交的主轴方向上,之后进行重构,通过控制重构主轴的数目(重构维数),获得最优量的输入噪声。实验结果证实了方法的有效性。

Investigating Stochastic Resonance by Principal Component Analysis

The energy transfer of noise to signal is a unique feature of stochastic resonance (SR) through which weak signal can be detected. Because that SR occurs when just amount of noise is added and too little or too much noise result in degraded performance, it is important to control effectively noise injected into a SR system. This paper firstly proposed to investigate SR by principal component analysis (PCA). Basic method is to reconstruct after projecting signal and noise in the directions of orthonormal eigenvectors and to achieve an optimal amount of input noise by choosing a proper reconstruction dimension. The experimental results verified the proposed method's effectiveness.