Improved Particle Swarm Algorithm And Its Application In Traffic Signal Control

With the private car ownership in urban areas increasing rapidly, trafficcongestion has been growing and attracting widespread attention. An importantpart of traffic in urban——intersection, is the key point of the whole city traffic.Nowadays, there are many effective methods to improving the operationalefficiency of the intersection. However, improving the intersection signal timingscheme is one of the most direct and effective ways. Therefore, the research onoptimization of intersection signal timing is extremely significant in theory andin practice.Particle Swarm Optimization, PSO, not only has many advantages such as,simple, easy to implement, without gradient information, fast convergence, lessadjustable parameters and flexible application, etc, but shows an excellent effecton continuous optimization and discrete optimization problems. Compared togenetic algorithm, the research on Particle Swarm Optimization is less mature.So, it is deserving of research on the project that improve the deficiencies ofparticle swarm algorithm in combination with other algorithms and apply theimproved algorithm in traffic signal control.In this paper, the research object is the urban road intersection signaltiming, choosing the total delay time, the number of stops and traffic capacity asthe index of signal timing optimization scheme. Then set up multi-object signaltiming model, and use the fuzzy compromise thought to transform it to a singleobjective function, and finally find the solution of single objective function withthe improved particle swarm algorithm, and then get more reasonable signaltiming optimization scheme.The main job of this paper:1. Combine the particle swarm algorithm with Powell to get improvedalgorithm——Powell-PSO algorithm. The algorithm uses the high precision,fast convergence, and local search ability of the Powell to overcome thedrawback of trapped in local optimum, strike a balance between global optimaland local optimal, and significantly improve the searching efficiency of the algorithm. With classic test function testing, numerical results show that theimproved algorithm acquires satisfactory results.2. Combine the simplified particle swarm algorithm with Powell to getimproved algorithm——Powell-SPSO algorithm. The algorithm uses strongsearching ability, high precision and fast convergence of the Powell to improveweak differences between the particles of the simplified particle swarmalgorithm in which each particle uses the same iteration formula for evolution. Italso avoid shortcomings of the prone to premature, slow search speed and so on.With classic test function testing, numerical results show that the improvedalgorithm achieves good results.3. In this paper, according to the actual traffic environment and conditionssuch as, the road width of urban road intersection and traffic flow, theoptimization goal of intersection signal timing is the total delay of the vehicle,traffic capacity and the number of stops. Use the multi-objective programmingthought of fuzzy compromise programming, converting these three objectivefunctions, the total delay time, number of stops and traffic capacity, into a singleobjective function, and then establish the intersection signal timing optimizationmodel. Finally, get a solution to the model of signal timing optimization methodvia designing the solving steps of the algorithm specifically with the Powell-PSO algorithm and Powell-SPSO algorithm respectively.4. Use the Powell-PSO algorithm to solve the model, getting effectivegreen time of single objective function in each phase, and then get the totaldelay time, capacity and the number of stops of the intersection. Comparing thesignal timing optimization scheme of the Powell-PSO with the original timingscheme, the result shows that: the former delay is reduced by47%, the totalnumber of stops is reduced by27%, while traffic capacity is increased by9%;the average delay, the average number of stops and capacity of each phase arewell improved, indicating that the signal timing scheme of the Powell-PSO isbetter and more appropriate to the traffic condition of the intersection, andproving indirectly the effectiveness and practicability of the Powell-PSO. 5. Use the Powell-SPSO algorithm to solve the signal timing optimizationmodel, getting effective green time of single objective function in each phase,and then get the total delay time, capacity and the number of stops of theintersection. Comparing the signal timing optimization scheme of thePowell-SPSO with the original timing scheme, the result shows that: the formerdelay is reduced by42%, the total number of stops is reduced by23%, whiletraffic capacity is increased by9%and the average delay, the average number ofstops and capacity of each phase are well improved. All of these above suggestthat the signal timing scheme of the Powell-SPSO will be able to get a bettercontrol of signal timing on traffic flow, and also manifest the effectiveness andpracticability of the Powell-SPSO.