RBF Neural Network Optimized by Particle Swarm Optimization for Forecasting Urban Traffic Flow
Xiaobin Li · 2009
Accurate traffic flow forecasting is significant to the intelligent traffic guidance and traffic control. RBF neural network (RBFNN) is a feed-forward neural network with one hidden layer and can uniformly approximate any continuous function to a prospected accuracy. Compared with the back propagation feed forward network (BPNN), the RBFNN requires less computation time for learning and higher forecasting accuracy. In order to realize the appropriate choice of the training parameters of RBF neural network, Particle swarm optimization (PSO) is introduced to optimize the parameters of RBF neural network. The experimental results show that the PSO-RBF neural network has higher forecasting accuracy than BP neural network.