对通用最大熵谱分析算法程序的改进

谱估计学科的发展,极大地提高了人们对振动、冲击及所谓噪声所反映的事物的动态特性和变化过程的认识。在现代谱估计范畴中,最大熵谱分析法是一种比较好的方法,在初期的最大熵谱分析算法中比较常用的方法是伯格算法。但这种算法在许多方面还有很大的局限性,如谱线分裂、谱能量不集中,频率偏移等等。为了克服上述算法的不足之处,对现有的伯格算法进行了较大的改进——在原算法基础上加入滤波器阶数优化,同时结合了MARPLE算法,从而大大地提高了谱估计的性能。

The Program Improving for MEAM

The major problem with morden spectral estimation is how to resolve signal from data samples more effectivly. That is to say,how to improve the resolution of the spectral analysis and perform the spectral analysis more efficiently. In this paper we will dis- cuss the maximum entropy analysis method (MEAM),a method of spectral estimation with extremely high resolution. Through analysising and researching in the MEAM theorem,we have put forward a auto-optimum seeking predictive filter orders algorithm on the basis of BURG algorothm. A lot of experimental results have shown that this algorithm presents ex- tremely high resolution and it can especially be adapted to measure signal from the data sam- ples of one or more sinusoid with GAUSS WHIT E NOISE. In experiment we can completly resolve two sinusoid signals which frequency are 0.165Hz and 0.215Hz from data sample with signal to noise ratios(SNR)being -0.485db. In the following part of this paper we will further analysis another MEAM algorithm with higher resolution named MARPLE algo- rithm,put forward by S. L. MARPLE(an american)