DETECTING CHAOS TIME SERIES VIA COMPLEX NETWORK FEATURE

In this paper, an effective method from time series to complex network via phase space reconstruction is introduced. We reconstruct the phase space from a time series by the time-delay coordinate method. Each state vector of phase space is regarded as a vertex of network and the connection is based on the distance of the vertices in phase space. The networks corresponding to various time series, the x component of the chaotic Rossler system, noisy periodic time series and random series, display different topology feature. So we can determine whether a time series is chaotic series by the topology feature of corresponding network. Finally, the daily stream-flow series of Yangtze River is investigated to validate the effective of our method.