【摘要】 参数模型谱估计是近二十多年来得到迅速发展的现代谱估计方法。根据极大熵原理，ＡＲ谱估计受到了普遍重视。极大熵谱估计要求信号必须是具有连续谱分布函数的一般线性序列。然而在生物医学工程中遇到的实际信号，其谱函数往往包含很尖锐的谱峰（如脑电信号的α峰），甚至包含离散谱线。实验证明，这种情况下对谱峰频率的估计精度仍然很精确，但对谱幅估计误差较大，国外近些年提出的一种最佳Ｂｕｒｇ修正算法，有一定的改善。我们从工程设计的需要出发，通过大量实验提出一种加修正系数的Ｙ—Ｗ算法，亦获得明显效果，达到了实用的要求。更多还原

【Abstract】 The parametric model spectral estimation is a modern spectrum estimation method that has been developing quickly in recent two decades. According to the Maximum Entropy Principle, AR spectral estimation has been affached importance widely. Maximum entropy spectral estimationrequires the signal must be a general linear sequence that has successive spectral distribution function. But the actual signals in biomedical engineering have characteristics that their spectral function usually contain very sharp spectral peaks, and even discrete spectral lines, Expeximents showed that in such case, the estimation of the frequency of spectral peak is very accurate , but the error of spectral peak amplitude is obvious. Some improvement has been made by optimum tapered Brug algorithm put forward in recent years. Owing to the requirement of engineering design, we put forward a modified Yule Walker algorithm and obvious improvement have obtained.更多还原