Multi-parameter fusion applied to road traffic condition forecasting

In view of the chaotic characteristic in road traffic flow and the actual traffic condition that cannot be comprehensively reflected by single traffic parameter, a fusion algorithm of multi-parameters for traffic condition forecasting, with the consideration of the relationship between multiple parameters, is proposed. This algorithm is based on the reconstruction of phase space. According to Bayesian estimation theory, the multiple traffic parameters are optimally fused into phase points in the same phase space. Accordingly, the phase space information increases and the phase points are closer to dynamical behavior of the traffic system. On the basis, by using the multi-parameter chaos prediction method, the tendency of dynamic systems from different aspects is described, with reference to the method of predicting single parameter chaotic time series. The experimental results confirm that more features of real traffic condition are reflected by fusing multiple traffic parameters. The multi-parameters forecasting algorithm reduces the prediction error and improves the equalizer coefficients compared with the results generated from single parameter prediction. That is to say, the prediction method used in this paper is effective and accurate for predicting traffic condition based on multiple traffic parameters.