论文部分内容阅读
本文讨论了一种新的扩展自相关匹配技术,简称EAC。众所周知,当我们采用LPC的自相关方法时,被测信号的前P+1个自相关函数与模型相应的自相关函数总是完全相匹配的。而当噪声不断增加时,其高阶自相关函数开始发散。EAC技术在使前P+1个数据自相关函数与模型自相关函数精确匹配的情况下,同时使高阶自相关函数的失配最小以选择噪声自相关函数的估计值,并从总的自相关函数中扣除,从而得到较好的参数估计值。实验表明,EAC技术优于其它方法,如奇异值分解法(SVD)和高阶Yule—Walker方程法(HOYWE)。
This article discusses a new extended autocorrelation matching technique, referred to as EAC. It is well-known that when we adopt the LPC autocorrelation method, the first P + 1 autocorrelation functions of the measured signal are always perfectly matched with the corresponding autocorrelation functions of the model. When the noise increases, the higher-order autocorrelation function begins to diverge. The EAC technique minimizes the mismatch of high-order autocorrelation functions to minimize the mismatch of the high-order autocorrelation functions by accurately matching the autocorrelation function of the first P + 1 data with the autocorrelation function of the model and selects the autocorrelation function of the noise from the total autocorrelation function Correlation function deduction, resulting in better parameter estimates. Experiments show that EAC is superior to other methods such as SVD and HOYWE.