Traffic Signal Control Based On Particle Swarm Optimization

Traffic signal control system is the basis of the intelligent transportation system, which can coordinate the timing control project in the area, balance traffic flow in the road network running, and therefore fully exploit benefits of road transport system. However, the method of the traffic signal control lack variability and flexibility, which can not effectively alleviate traffic problems of the complex network of urban roads. Therefore it is necessary to optimize the control algorithm in order to find solutions to the traffic signal control project.Multi-target traditional control algorithm like the linear weighting method, goal programming and constraint method integrate various objective functions into a single objective function and adjust itself by policy makers or the optimization coefficient set carried within itself. While these traditional methods are simple and easy to apply, it is necessary to get a prior and full grasp of knowledge about optimization problems due to the non-linearity, continuity or non-differentiability of the objective function of multi-objective control problems. Thus, these traditional methods are often unable to solve more complex multi-objective control problems. Compared to traditional optimization algorithms of the control system, the evolutionary algorithm is a random search algorithm which mimic the natural selection and biological evolution and is more suitable for practical problems which deal with multi-objective control. Also, because the Particle Swarm Optimization(PSO) algorithm is novel and relatively little applied to research in intelligent traffic control, for complex problems of high-dimension, the PSO algorithm can reduce the amount of computation while guaranteeing an ideal convergence result as much as possible, which not only overcomes the problem that the gradient algorithm can not easily get rid of local optimal solutions, but also to overcome the huge disadvantage of too much computation in the brute-force method. Thus, by appropriately improving the existing algorithms for intelligent traffic control, we can achieve a breakthrough.This thesis mainly adapts the PSO algorithm to find best solutions. First of all, by researching relevant theories of the Analytic Hierarchy Process(AHP) and the PSO algorithm, it focuses on the basic principles, mathematical models and parametric analysis of the PSO algorithm, thus proposes improved methods for parameter optimization of the traffic control system; meanwhile it analyzes the application of the PSO algorithm by studying the comparison between it and existing theories to further deepen the understanding of the algorithm. Secondly, it improves the basic PSO algorithm based on research on the basic theory of the PSO algorithm which is expected to avoid the problem of premature convergence to significantly improve its performance on the basis of the basic PSO algorithm. Finally, by using the VISSIM traffic simulation software, it finishes drawing the transportation network model and setting simulation parameters, and applies the optimized PSO algorithm to simulation experiments of traffic signal control to validate the effect of control applying the PSO algorithm which is optimized with multi-objective optimization expectations.Combining these studies with experimental work, this thesis introduces the thought of multi-objective control, combines the AHP and PSO algorithm to solve the problem of traffic signal control:it weights the control of index layer parameters with the use of multi-objective method, and therefore gets evaluation functions of traffic patency with different objectives; then it uses the PSO algorithms to optimize the parameters of traffic light signals at an acceptable speed and accuracy. Introduction of AHP makes the evaluation function more reasonable and effective, while the use of the PSO algorithm reduces computation as much as possibile while ensuring the reliability of the results.The primary works in this thesis are presented as follows:(1) it studies and researches relevant theories of the Analytic Hierarchy Process(AHP) and the PSO algorithm and focuses on the basic principles, mathematical models, parametric analysis and improvements of the PSO algorithm; meanwhile it analyzes the application of the PSO algorithm by studying the comparison between it and existing theories to further deepen the understanding of the algorithm.(2) it improves the basic PSO algorithm and effectively avoid the problem of premature convergence to significantly improve the performance of the basic PSO algorithm.(3) it uses the VISSIM traffic simulation software and finishes drawing the transportation network model and setting simulation parameters.(4) it applies the optimized PSO algorithm to simulation experiments of traffic signal control, which validates the effect of control applying the PSO algorithm which is optimized with multi-objective optimization expectations.