A simplified prediction model for urban traffic network

Abstract

An accurate model with low on-line computational complexity is often needed for model based control methods such as MPC. The macroscopic S model [1] provides good accuracy in prediction but the nonlinearity makes it with low efficiency when solving optimization in MPC. In this paper, a further simplified prediction model with little loss in accuracy is developed based on the S model. The model gives explicit relationship between green time and queue length in urban traffic network. It can be easily used for analysis and optimization for traffic congestion. Simulation results show that the new model runs faster than S model and it still preserves relatively high accuracy.