奇异谱分析的广义性及其应用特色

GENERALITY AND APPLIED FEATURES FOR SINGULAR SPECTRUM ANALYSIS

本文从理论上证明，奇异谱分析（ＳＳＡ）是一种广义功率谱分析。其广义性及应用特色是：（１）它具有稳定的识别和强化信号功能，在识别优势周期信号及强化显示方面，功能超过最大熵谱分析（ＭＥＳＡ）；（２）其识别和描述信号的方式是时域性的频域特征分析，即时频域相结合，有益于天气气候诊断和预测；（３）它并不受正弦波假定的约束，对于各种非正弦波信号同样有稳定的识别功能。文中实例也验证了理论证明。

It is theoretically demonstrated that singular spectrum analysis(SSA) is a generalized power spectrum analysis in this paper.Its generalities and applied features are as follows:(1)SSA possesses more stable identification ability as well as the intensification showing action for the period signals;so that it is excellenter than maximum entropy spectrum analysis(MESA) to identify predminant period signal and to enhence showing of sigmal;(2)Its manner of identifying and descripting signals is a frequency domain analysis by means of the time domain properties,namely the manner of both combination of time and frequency domains,thus SSA is useful to the diagnosis and forecast for weather and climate;(3)SSA is not only applicable to sine signal detection but also to non sine signal.The calculated cases have virtified above results.