Freeway Traffic Flow Model Based on Rough Sets and Elman Neural Network

Abstract

Rough sets theory is a new tool for processing fuzzy and uncertain knowledge, and has already been applied to many areas successfully. In this paper, a freeway traffic flow model based on rough sets and Elman neural network is put forward. The main idea of this approach is that some redundant features of sample data are reduced by rough sets firstly, then Elman neural network is used to build traffic flow model. Finally, a freeway with five segments, one on-ramp and one off-ramp is simulated. It is proved that the combined model of rough sets and Elman neural network has higher accuracy and better associational output ability than Elman neural network model by comparing their simulation outputs. The high performance of this combined model provides a novel and practical way to realize on-line modeling of freeway traffic flow.