Complexity Measure of Traffic Flow Based on Approximate Entropy and Statistical Complexity

Feng-tao LIU,Guo-guang HE

DOI: https://doi.org/10.3321/j.issn:1001-7372.2007.04.019

2007-01-01

Abstract:In order to quantitatively describe the complexity of traffic flow system,hence two eigen indexes of nonlinear dynamics,ie approximate entropy and statistical complexity were introduced.The approximate entropies and statistical complexities of periodic sequence,white noise sequence,the Henon and Logistic sequences which were situated in the chaotic region,were calculated.The approximate entropies and statistical complexities of five actual time series of traffic flow were calculated.The results show that the two eigen indexes can be applied to effectively measure the randomicity and nonlinearity of a system and resolve the difficult problems of short sequence and comparability in depicting the complexity of a system.They are suitable to measure the complexity of the actual traffic flow.The randomicity of traffic flow system is high and there are some nonlinear structures in the traffic flow system which probably can generate chaos.Manifold methods should be used to measure the complexity of a system.