Abstract: The principle of local maximum entropy is used to improve the lane-changing rule. Meanwhile, lane-changing with entropy-increasing probability is introduced, and the serial update method is used during the lane-changing. It reflects the electric bicycle flow characteristics very well. Then, the concept of lane-changing wave is introduced, and the formation mechanism and propagation of the lane-changing wave are analyzed, which is used to explain the relationships of lane-changing rate with the number of lanes and between the lane-changing with entropy-increasing probability. Furthermore, the influence on the fundamental diagram is analyzed under different parameter set in deterministic and stochastic cases. If the flow is appropriate, four lanes for electric bicycle lanes can more easily make the flow of metastable state appear in the high branches, and improve the using-efficiency of the road. Finally, field observations were carried out to calibrate the slowdown probabilities. The model is compared with multi-value model, dynamic floor model and field observation data. The results suggest that the proposed model not only overcomes the defect of the existing models, but also has higher accuracy.

Key words: engineering of communications and transportation system, cellular automata, local maximum entropy, lane-changing wave, electric bicycle flow