一种基于PCI技术的谱估计方法

A Method of Spectrum Estimation Based on PCI Technique

Burg方法是自回归模型谱估计方法中一个最受欢迎的方法。但Burg方法对含有噪声的正弦信号进行谱估计时存在两个方面的问题,为此该文提出了一种基于主成份求逆技术的谱估计方法。推导了利用主成份求逆技术计算自回归模型系数的解析表达式,利用该方法计算自回归模型系数只涉及到相关矩阵的主成分,不需要进行Burg方法中的Levinson递推,从而避免了出现Burg算法所出现的问题。最后通过计算机仿真验证了对该方法相对于Burg方法的优越性。

The most popular approach of autoregressive model spectrum estimation is Burg method. However when the Burg method is used to estimate the spectrum of sinusoid embedded in noise, it suffers from two problems. A method of spectrum estimation based on Principal Component Inverse technique is proposed in this paper. The analytic expression for computing autoregressive model coefficients by Principal Component Inverse technique is derived. The principal component of autocorrelation matrix is only involved in the method without Levinson recursion in the Burg method, so the problems of Burg method can be avoided. Its superior performance is justified when compared to the Burg method by computer simulation.