广义最大熵法在估计最小交叉熵谱中的应用

The Application of the Generalized MEM to Minimum Cross-Entropy Spectral Analysis

本文把最大熵谱分析法(MEM)拓广应用于直接从数据记录非迭代地估计有先验谱的最小交叉熵(MCEM)谱.这种新的非迭代快速最小交叉熵谱分析法(MCEM)算法不仅从理论上进一步证明了MCEM是MEM的拓广,而且在实践上真正把MCEM算法实现为增加了预白化和后加权处理的广义MEM算法.分析、比较和计算机模拟表明,所推荐的算法能以明显优于常规的迭代MCEM算法的计算速度和与MEM算法相同的计算效率给出性能较大地比ME谱优的MCE谱.

The MEM is generalized to estimate the Minimum Cross-Entropy(MCM)spectrum of a random signal from its data record when its a priori spec-trum is available.The new, fast and noniterative Minimum Cross-Entropy spectral analysis Method (MCEM) algorithm developed not only proves further in theory that the MCEM is a generalization of the MEM, but also realizes really in practice the MCEM algorithm as a generalization of the MEM's, which is an MEM algorithm with prewhitening and post-weighting.The theoretic analysis and computer simulations demonstrate that the proposed algorithm has the attractive ability to estimate the MCE spectrum whose resolution is much higher than the corresponding ME sp-ectrum's directly from the data record of a random signal and with the same computation efficiency as the fast MEM algorithm.