混合小生境遗传-模拟退火算法在动态交通分配中的应用

为了解决已有的动态交通分配模型普遍存在的优化计算时间过长,严重影响网络规模扩展的问题,提出了一种新的动态交通分配优化算法。将模拟退火和隔离小生境技术有机地结合起来,融入到遗传训练过程中形成了一种混合小生境遗传-模拟退火算法,不仅可以有效地避免传统遗传算法可能出现的不收敛现象,加快进化速度,具有更强的全局寻优能力,而且计算速度和算法稳定性也得到提高。将其与Papageorgiou M.提出的动态交通分配模型框架相结合,设计了动态交通分配的快速模拟优化算法,并进行了实例研究。仿真结果表明,新的优化算法显著降低了优化计算时间,大大提高了动态交通分配模型的实用价值。

Application of Hybrid Niche Genetic Simulated Annealing Algorithm to Dynamic Traffic Assignment

The existing dynamic traffic assignment models need overlong time to optimize,which seriously influence traffic network extension.In order to solve this problem,simulated annealing and isolation niche which can improve performance of genetic algorithm were well combined,and a hybrid niche genetic simulated annealing algorithm was derived.The new hybrid genetic algorithm can not only efficiently avoid possible non-convergence of genetic algorithm,improve evolution speed and has better global optimization ability,but also improve the computing speed and stability of the algorithm.Combined hybrid niche genetic simulated annealing algorithm with the model of dynamic traffic assignment(DTA)proposed by Papageorgiou M.,a new fast simulated optimization algorithm of DTA was designed and case research was also done.Simulation results show that the new algorithm can evidently reduce the time of optimization calculation and improve the practicality of the model of DTA.