基于GA-Transmodeler的动态OD矩阵估计方法*

采用系统仿真与遗传优化算法相结合的方法完成离线状态下多时段OD矩阵的估计。系统仿真旨在对多时段的动态OD矩阵实现连续动态交通分配,以得到在各个时段的OD流量对观测路段的分配比例矩阵,遗传算法则通过种群内个体的交叉、变异等遗传算子求解最优条件下的OD矩阵。仿真结果表明,这种仿真优化结合的方法能够充分体现动态交通流的延续性,且遗传算法具有较强的全局收敛性。

GA-Transmodeler based time-dependent origin-destination estimation

This paper used a hybrid simulation-genetic algorithm method to accomplish off-line dynamic OD matrix estimation. The aim of simulation was to accomplish dynamic OD matrix thus OD-to-link flow proportion would be calculated while genetic algorithm was used to solve optimization function by crossover, mutation operator etc.The continuity of dynamic traffic flow and strong global convergence inherent in GA can be verified by the simulation cases.