考虑充电需求的电动公交车运营优化模型与算法

Optimization model and algorithm for electric bus operation considering charging demand

随着新能源汽车技术的快速发展,电动公交车被视为缓解城市交通拥堵和降低环境污染的有效手段,然而电动公交车的里程限制和充电需求等特点使得公交网络设计和运营面临新的挑战。本文在公交分配的基础上,考虑电动公交车充电需求,对新型电动公交的发车频率、运营车辆数、车辆充电计划进行优化设计。构建一个双层规划模型,上层为带有电动公交线路运营充电仿真模块的公交网络优化模型,从运营商的角度来实现乘客出行成本和电动公交网络运营总成本最小;下层基于UE(User Equilibrium)均衡准则来描述乘客出行路径选择行为并预测公交网络流量。提出基于代理模型的算法(Surrogate-model-based Algorithm)来求解所构建的双层规划模型,并嵌入基于超路径的投影算法求解下层电动公交均衡分配问题,利用线路运营充电仿真模块求解上层运营车辆数车辆、充电计划及车次数量。最后采用数值算例验证了该模型和算法的有效性,算例结果显示,所提出的代理模型算法比传统遗传算法求解效率和精度更高,且随着网络规模的增大效果更为明显。

With the rapid development of new energy vehicle technologies,electric buses have been regarded as an effective way to reduce urban traffic congestion and environmental pollution.However,the characteristics of electric buses,such as limited range and charging requirements,make the design and operation of a public transportation network challenging.Based on the transit assignment and considering the charging demand of electric buses,this study optimizes the electric bus line frequency,the number of vehicles,and vehicle charging plan for electric bus systems.A bi-level model is constructed:the upper-level model is the optimization model for the electric bus network with an operation simulation module,which minimizes the travel cost of passengers and the operation cost of the electric bus network;the lower-level model describes the route choice behavior of passengers and predicts the bus network flow based on the UE criterion.In this paper,a surrogate-model-based algorithm is proposed to solve the constructed bi-level model,and a projection algorithm based on a hyperpath is embedded to solve the transit assignment problem in the lower-level model.The number of vehicles,electric bus charging plan,and bus line frequency in the upper-level model can be obtained by using the bus line operation and charging simulation module.Finally,numerical examples are used to verify the effectiveness of the bi-level model and proposed algorithm.The numerical results show that the proposed surrogate-model-based algorithm is much more efficient and accurate than the genetic algorithm,and the improvement is more apparent with increasing network scale.