本文给出将时间序列分解为几个分量的模型,此中各分量用自回归模型表示,并给出估算这些模型的最小 AIC 算法。估算模型的变化方差是其关键问题,采用分段模型和局部模型技术解决。以微震信号的提取为例,对所提出的算法作了说明,利用同样数据组,还对算法的稳定性及可能的简化进行了研究。 This paper presents a model that decomposes time series into several components, where each component is represented by an autoregressive model and gives the minimum AIC algorithm for estimating these models. Estimating the variance of the model is the key issue, which is solved by the segmentation model and the local model. Taking the extraction of microseismic signals as an example, the algorithm proposed is described. Using the same data set, the stability and possible simplification of the algorithm are also studied.