A Bus Signal Priority Control Method Based on Deep Reinforcement Learning

To investigate the issue of multi-entry bus priority at intersections, an intelligent control method based on deep reinforcement learning was constructed in network environment. Firstly, a dimension reduction for state vector key lane proposed, which contains characteristic parameters such as states, flow and signal timing. Secondly, action that can adjust phase sequence green time same proposed under constraints maximum minimum green. Furthermore, reward function, be uniformly converted into number standard cars, established focusing indexes busload waiting time. Finally, through building experimental environment SUMO simulation, real-time constructed. The results show algorithm effectively reduce buses without affecting overall traffic efficiency. findings provide theoretical basis considering improve operation efficiency public transport.