Research On Multi-objective Optimization And Decision-making Method Of Variable Speed-limit Control For Freeways

Freeways have been the key construction objects of the transportation industry in ourcountry. However, with the growing traffic demand, a wide range of problems begin tostand out. Relevant practices show that, As an subsystem of intelligent control, VariableSpeed-Limit has outstanding advantages in resolving those issues. This technology can notonly significantly reduce the dispersion of speed and the accident rate of speed-limit zones,but also make traffic flow smoother, so as to effectively improve the traffic capacity.However, existing researches on Variable Speed-Limit for freeways, which take themaximum traffic flow or the minimum travel time as control target, are still stuck in thelevel of single-objective optimization. They rarely take multi-objective optimization andenvironmental pollution into account. Increasingly serious air pollution is resulting in awide range of persistent fog and haze, so it’s urgent for traffic managers to change theirminds. We can no longer only increase traffic flow or reduce travel time blindly.Multi-objective problem should be considered in Variable Speed-Limit control.Considering of these problems, this paper, which is based on the Youth Foundationproject of Jilin Province—"Research on Variable Speed-Limit algorithm for Freeways",conduct a depth study on the multi-objective problems of Variable Speed-Limit from twoaspects—multi-objective optimization and decision-making. Among them, the former isprerequisite for multi-objective decision-making, it can obtain a Pareto which coordinatesthe.various objective. And the later is the final step to solve the multi-objective problem, itcan select a single solution from Pareto to perform.Therefore, the research contents of this paper generally include the followings:1）Multi-objective optimization of Variable Speed-Limit control for freeways isstudied. Relying on the method of Multi-objective optimization theory, construct themulti-objective optimization model including exhaust emissions, traffic flows and traveltime, combined with the specific characteristics of Variable Speed-Limit. And then, clearthe advantages and feasibility of selecting Genetic Algorithms of the model solution byanalyzing the correlation algorithm. On the basis of specifically matching multi-objective optimization model and Genetic Algorithm, set parameters and clear solving processes.And thus, complete the section of multi-objective optimization for freeways, then obtainthe Pareto, which coordinates the objective functions with each other.2）Multi-objective decision-making of Variable Speed-Limit control for Freeways isstudied. Relying on the summaries and analysis of multi-objective decision, clear theadvantages and feasibility of Reinforcement Learning. And then, based on the principles ofReinforcement Learning, build the model, including basic constituent elements, rewardfunction and criterion function. Finally, by setting the action selecting mechanism, clearthe concrete steps to solve this model. And thus, complete the section of multi-objectivedecision-making.and get a single executable solution from Pareto.3）Simulation of multi-objective optimization and decision-making is conducted. Bycomparing the various simulation software, select Paramics to do the simulationexperiments. Then, build the simulation platform, including of building roads, settingsimulation parameters, laying variable message signs and detectors, determining the resultsof multi-objective optimization and embedding multi-objective decision-making algorithm.Finally, analyzed and compared the output results. Thus completing. the validation offeasibility and effectiveness of multi-objective optimization and decision-making models.4）Full summary and outlook. By induction major results of this paper, present thesurviving shortcomings and later reseach focus.The research on Multi-objective optimization and decision-making method of VariableSpeed-Limit control for freeways in this paper, can not only effectively coordinated thecontrol objectives of emissions, traffic flows and travel time, but also can intelligentlyselect the speed-limit value to performe, according to the real-time traffic conditions. Thispaper is useful for the freeway intelligent control, and the research achievements canprovide some technical supports for Variable Speed-Limit control.