奇异谱分析是近年来兴起的一种研究非线性时间序列数据的强大的方法。它根据所观测到的时间序列构造出轨迹矩阵，并对轨迹矩阵进行分解、重构，从而提取出代表原时间序列不同成分的信号。如长期趋势信号、周期信号、噪声信号等，从而对时间序列的结构进行分析，并可进一步预测。(Singular spectrum analysis is a powerful method to study nonlinear time series data in recent years. A trajectory matrix is constructed according to the observed time series, and the trajectory matrix is decomposed and reconstructed, so that the signals representing different components of the original time series are extracted. Such as long-term trend signal, periodic signal, noise signal, and so on, so as to analyze the structure of time series and predict further.)