Study On Adaptive Traffic Signal Control System Based On Cloud Models

Based on research on existing modern traffic signal control systems and practical experiences, this paper proposes a new adaptive traffic signal control method and the corresponding system for single intersection traffic signal control based on cloud models.In the method proposed in this research, the first step consists in collecting traffic information on single intersection and assigning numeric characteristics of cloud models to create a three-dimensional cloud picture via normal cloud arithmetic for transformation between quantitative traffic data and qualitative description. The picture will then create the subordinate degree of every cloud drop to different traffic status and apply the corresponding signal-based rule to control the traffic. Also, by using Q learning method, the parameters used in the models are optimized through iterations. Stop delay is used as the target function to obtain the best parameter configuration.To test its performance, simulations are conducted to compare the method proposed with traditional methods. According to the results of those simulations, combined with Q learning method to optimize its parameter configuration, the proposed control method performs better than the others in self adaptation and efficiency, which proves that the cloud models can be appropriately applied to traffic signal control. The new traffic signal control system thus established is capable of adaptive control due to the two characters of randomicity and fuzzy of cloud models.