Traffic Signal Light Timing And Optimization In The Time-Varying Universe

At present,continuously increasing travel demand mismatching the construction efficiency of road network infrastructure has become the main contradiction in China’s urban transportation system,and urban congestion becomes normal.Effected by limited road resources,optimizing the traffic control strategy and improving the utilization rate of road network will become one of the important ways to meet people’s traffic demand.Advantages of fuzzy control is obvious,especially in no need for accurate modeling,strong robustness and good control effect for time-varying and nonlinear systems.Therefore,the fuzzy control under time-varying universe is used to deal with the problems of timing and optimizing phase for intersection signal lights.Firstly,in this paper,the universe of cycle is determined for the key traffic flow of every phase.Let the key traffic flow and average queue length be the input,the green light duration of every phase be the output,and the timing for traffic signal lights is pro vised.Secondly,the initial phase structure is optimized to deal with the problem of phase structure mismatch.By people’s travel rules,the traffic flow is generally asymmetry or small,then the phase structure is optimized by seting changeable lanes,adding additional phase and merging phase,and then the corresponding timing numerical procedure is given.Thirdly,Q learning algorithm is used to update fuzzy rules and membership functions for the contraller.Membership functions of input and output fuzzy sets and their corresponding fuzzy rules are adjusted continuously to obtain the best control effect.In this paper,all the combinations of fuzzy subsets for input variables are a state set,that of output variables is a action set,then the Q value table is built.Q table is updated by the interaction of controller and environment to explore the best decision,and then the fuzzy rules and membership function are also updated.Finally,taking the intersection of Mengle Avenue and Mengla Road in Jinhong City as an example,traffic flows of six representative periods were discussed to verify the feasibility and effectiveness of the design scheme by MTLAB and VISSIM.In this paper,a new fuzzy control method based on time-varying universe is proposed.The traffic light timing and intelligent control are synthesized to optimize the phase combination and corresponding time ratio.The fuzzy rules and membership function are optimized by Q-learning algorithm,which effectively improves the traffic capacity of intersection.