The value of short-time traffic flow prediction in the PSO-RBFNN study

Abstract

Traffic flow data are un-periodical, nonlinear and stochastic, the practicability and accuracy are affected due to its drawbacks of falling into local optimization and low convergence rate. Thus, RBF neural network optimized by particle swarm optimization algorithm (PSO-RBFNN) is proposed to predict traffic flow in the paper. Being easy to realize, simple to operate with profound intelligence background, the parameters and connection weight are optimized by the algorithm and short time traffic flow prediction is simulated by the optimized RBF Neural Network. The prediction results of the instance show that it has better prediction results, higher precision, faster convergence than that of RBF prediction model. The optimized RBF Neural Network is suitable for short time traffic flow prediction. The method has good prediction accuracy and popularization value.