中文摘要：

为有效利用径流数据的非线性特征来对径流作分类识别,提出采用递归图和近似熵获取原始时间序列的动力学行为的方法.在计算随机、正弦和Logistic映射3种典型时间序列的递归图的基础上,采用递归图分析获取不同时间尺度的径流时间序列的动力学行为的方法,进行确定性检验,定性的判断径流序列的非线性和原动力系统的确定性.在递归图算法的基础上,定量地将近似熵用于序列的复杂度描述,对4组径流序列的递归图和近似熵进行比较,得出混沌递归分析是一种流型识别的有效辅助诊断工具.

英文摘要：

To make use of the nonlinear property of runoff records in feature extraction to classy the runoff effectively,the method of recurrence plots and approximate entropy are citied to get dynamic characteristics property of original time series.After calculating recurrence plots for Brownian movement,the sinusoid and Logistic map,and the recurrence plots of time series at different time scales are studied quantitatively.Based on the recurrence plots algorithm,approximate entropy is used for characterizing complexity analysis of runoff series qualitatively;the approximate entropy and recurrence plot of four different runoff series are compared.The results indicate that the recurrence analysis is a valid supplementary diagnosis tool for flow pattern identification.