现代谱估计方法在仿真可信性研究中的应用

Modern Spectral Estimation and its Application in the Research of Simulation Credibility

仿真可信性研究是仿真研究的重要组成部分，其主要工作包括模型的验证、确认和认定（ＶＶ＆Ａ）。谱分析方法是研究仿真可信性的定量方法，适合于对系统动态性能进行分析，是一种统计检验方法，且可信度较高。特别是现代谱估计方法，既有高分辨率，又可以分析非平稳时间序列，具有广泛的应用。该文分别分析了傅立叶谱估计、最大熵谱估计、交叉谱估计、瞬时谱估计、演变谱估计等的统计特性、应用特点和适用范围。

Simulation credibility is an important constituent part of simulation research, its main role includes model validation, verification and accreditation(VV&A). Spectral analysis is a quantitative method of simulation credibility research, which is suitable for analyzing the dynamic performance of the system, and a statistical test method with high confidence level. Especially, modern spectral estimation is not only of high resolution, but also able to process non stationary time series, so it is widely used. In this paper, it is respectively discussed that the statistical character, application range of Fourier spectral estimation, maximum entropy spectral estimation, cross spectral estimation, Wigner distribution, evolutive spectral estimation.