无人驾驶环境下考虑OD结构的路网容量模型（双语出版）

为了研究未来无人驾驶车辆对路网容量的影响，揭示无人驾驶车辆与普通车辆的相互影响特性，假设无人驾驶车辆遵循系统最优路径，普通车辆遵循用户最优路径，构建了无人驾驶环境下的道路网络储备容量模型。上层模型为满足路段容量约束条件下的最大交通需求，各OD之间的交通需求采用不同的增长乘子；下层模型为两类用户的混合路径选择行为模型，无人驾驶车辆以系统总阻抗最小为目标，而普通车辆以个人出行成本最小为目标。采用多种群遗传算法进行求解，并通过算例验证了模型和算法的有效性和可行性，得到非统一增长乘子下的路网容量，比较了统一增长乘子与非统一增长乘子的异同之处。研究结果表明：①两种计算结果所得到的道路网络容量增长趋势类似，但是非统一增长乘子计算结果大于统一增长乘子计算结果，当无人驾驶车辆市场渗透率达到一定比例时，二者计算结果的差异随着市场渗透率的增加而逐渐减小；②不同OD对的增长乘子不一定相同，无人驾驶车辆的加入可以优化不同地区的OD需求分布，从而提升整个道路网络的容量；③非统一乘子的计算方法可以有效避免不同OD对的干扰作用，提高部分OD对在低市场渗透率下的路径利用率，路段流量分布更加均衡；④当无人驾驶市场渗透率达到较高的比例时，道路网络容量可增加的幅度较小。

A Road Network Reserve Capacity Model in the Autonomous Environment（in English）

In order to study the impact of future autonomous vehicles on road network capacity and reveal the interactive characteristics of autonomous vehicles and traditional vehicles. The autonomous vehicles were supposed to follow the optimal path in the system. Human drivers were assumed to use the shortest route, and a model of road network reserve capacity in the autonomous environment was constructed. The proposed model has two layers. The maximum traffic demand was obtained under the constraint of road capacity in the upper model. The traffic demand between each origin-destination (OD) pairs adopted different growth multipliers; the lower model had mixed equilibrium considering the routing behavior of two types of users. The minimum total system impedance was pursued by the autonomous vehicles, while minimum personal travel cost was achieved by traditional vehicles. The multi-population genetic algorithm was applied to solve this problem. The validity and feasibility of the model and algorithm were verified by an example. Thus, the road network capacity was obtained by considering the non-uniform growth multiplier. The similarities and differences between the uniform and non-uniform growth multiplier were compared. The results show that:① the road network capacity growth trend obtained by the two calculation methods is similar, but the non-uniform growth multiplier calculation results are greater than that of the uniform growth multiplier. When the market penetration rate of autonomous vehicles reaches a certain proportion, the difference between the two results decreases gradually with the increase in market penetration rate. ②The growth multipliers between different OD pairs are not synchronized, and the OD demand distribution in different regions can be optimized after the entry of autonomous vehicles. Therefore, the capacity of the whole road network increases. ③ The non-uniform multiplier calculation method can effectively avoid the interference effect of different OD pairs by improving the path utilization rate of some OD pairs in the low market penetration rate. ④ Traffic flow distribution is more balanced when the autonomous market penetration rate reaches a higher ratio, and the road network capacity can be increased in a smaller range.