Y.R. Guo, X.M. Wang, M. Wang, H. Zhang

Pages: 145-156

Abstract
There are many indicators in the prediction of intelligent traffic congestion, and the process of weight determination is complex, which results in poor accuracy. Based on radial basis function (RBF) network, this paper proposes a method to predict urban congestion. The ability of RBF network to analyze complex parameters is used to collect multi parameter indicators of urban traffic. The multi parameter indicators are used to select the evaluation indicators of urban congestion conditions, and the weights of different evaluation indicators of urban traffic congestion are determined under the analytic hierarchy process (AHP). The grade of urban traffic conditions is divided according to the importance of the indicators, and the urban congestion conditions are predicted according to the weighted average results of each indicator. The experimental results show that under the condition of reasonable distribution of multi parameter weights, the prediction accuracy is about 90%, and the accuracy is improved obviously.
Keywords: radial basis function network; urban congestion; prediction; neural network