Machine Learning Transit Signal Priority Control of Bus Rapid Transit Based on Connected Vehicles Environment

Nowadays, bus priority plays a critical role for national development with the promotion of urbanization. Nevertheless, the progress of bus system has been seriously hindered by unpunctuality. Based on the existing signal priority control theory, this paper proposes a reinforcement learning signal priority control and vehicle speed guidance method for BRT punctuality rate in connected vehicles environment. The reinforcement learning method has a good self-learning ability in complex multidimensional space because it does not need accurate mathematical model and low requirement for prior knowledge of the environment. Therefore, through the signal timing control and the BRT vehicle speed guidance, the BRT punctuality rate is improved and effect of the priority of signal control to other social vehicles is reduced, and ultimately the purpose of bus priority could be achieved. Through analyzing the result, it can be concluded that the speed guidance and active signal priority system can reduce the average delay time of BRT, hoist the average operating speed and average passenger delay.