Research On Intelligent Control Algorithm Of Traffic Lights Based On Road Traffic Processing Capability

The increasing number of vehicles leads to frequent traffic congestion,which increases the travel time,fuel consumption and gas emissions significantly.It not only affects comfort of travel but also causes huge pollution to the environment,so the problem of traffic congestion needs to be solved urgently.Traffic intersections are the intersections of all directions of traffic and one of the areas where traffic jams occur frequently.Therefore,if traffic lights can be controlled reasonably and efficiently at intersections,traffic jams can be effectively prevented and alleviated.However,most adaptive traffic light control methods only consider single parameters such as vehicle queue length or traffic flow at the intersection,and do not consider the impact of the traffic capacity of the adjacent intersection on the current intersection;or only consider the control of traffic lights at a single intersection without considering the cooperative control with surrounding traffic lights.For the above problems,this paper proposes a traffic light control algorithm(RTCR)based on the real-time capacity of roads in combination with fog computing,road flow processing capabilities and reinforcement learning theory.The algorithm sets an agent for each intersection on the end-edge-fog-cloud platform.Since the traffic flow information of each intersection will be shared in the fog layer,each intersection can obtain the traffic flow information of its own adjacent intersection to optimize its traffic light decision.The RTCR algorithm uses information such as road flow processing capacity and vehicle queue length to calculate the phase sequence of traffic lights.Combined with the deep Q-learning network(DQN)algorithm in reinforcement learning,the traffic flow information of its own intersection and the adjacent intersection to calculate and optimize the green light duration of each phase of the current intersection,and realize the intelligent control of traffic lights.This article uses Python and VISSIM to build a joint simulation platform on which simulation experiments are performed for single intersection control and multiple intersection control.The simulation results show that compared with traditional traffic light control methods,trunk road control methods(ATL)and Q-learning-based FRTL control methods,the RTCR control method proposed in this paper improves the intersection throughput and reduces the average waiting time of vehicles.The goal of preventing and mitigating traffic congestion was achieved.