关键词: 关联交叉口, 短时交通流, 可预测性, 组合预测

Abstract:

In this paper, first , the related intersection is defined according to the relationship between traffic flow time series. Then , several forecastability indexes for the short-term traffic flow of related intersections are proposed for the quantitative analysis , and the computation method of the first Lyapunov index to traffic flow time series is described.Moreover , a nonlinear combination forecast model using RBF neural network is set up based on a set of forecast models after the forecastability analysis. Thus , a combination forecast algorithm for the short-term traffic flow of related intersections comes into being. The results of simulation indicate that the proposed method is more efficient than the single forecast method.

Key words: related intersection, short-term traffic flow, forecastability, combination forecast