现代谱估计法及应用效果

Modern spectrum estimation method and its application effects.

本文针对Burg谱估计法中存在的问题,讨论了改进Burg谱估计法和改进协方差谱估计法,以理论信号为测试对象,对不同谱估计法的应用效果进行了对比,结果表明:①Burg谱估计法分辨率明显高于Welch谱估计法,但Burg谱估计法存在明显的峰值偏移,改进Burg谱估计法几乎没有峰值偏移;②改进协方差谱估计法和Burg谱估计法都具有较高的分辨率,而前者的波峰较后者更明显、尖锐,对于短数据、信号频率差异较小的信号,前者具有更好的分辩效果,还能抑制谱线分裂和出现假谱峰等问题;③Itakura算法求得的反射系数大于或等于Burg谱估计法求出的反射系数,使得接收到的信号更接近于实际输入信号,因此可用Itakura算法替代Burg谱估计法。实际地震剖面去噪结果表明,以高分辨率谱估计方法为基础的信噪分离方法具有较好的去噪效果。

Regarding the problem in Burg spectrum estimation method the improved Burg spectrum estimation method and improved covariance spectrum estimation method were discussed in this paper, by taking theoretical signal as testing object, the application effects for different spectrum estimation methods were compared, the conclusions are as follow: (1)The resolution for Burg spectrum estimation method is much higher than Welch spectrum estimation method, but the former caused obvious peak migration, while the improved Burg spectrum estimation method hardly generated peak migration, (2)Both improved covariance spectrum estimation method and Burg spectrum estimation method have high resolution, the peak of the former is more obvious and sharper, while for short data and signals with small frequency difference the former has better resolution effects, and it also can restrain spectral line splitting and avoid false spectrum peak. (3)The reflection coefficient estimated from Itakura algorithm is larger than or equal to that from Burg spectrum estimation method, the receiving signal for Itakura algorithm is much similar to real signal, as a result the Itakura algorithm can replace Burg spectrum estimation method. The de-noising results for real seismic sections show that the signal-noise separation technique which are based on high resolution spectrum estimation methods have a excellent de-noise effects.