数据驱动的电动出租车换电站定容和智能调度

In recent years, along with the introduction of a series of new energy vehicle promotion policies and propulsion of vehicle electrification in many cities throughout our country, the number of electric vehicles has been increasing rapidly, leading to significant boost in demand of charging infrastructures. Under such background, this project studies the capacity configuration and intelligent scheduling in battery swap stations of electric taxis based on real data, which includes the following major aspects:.1. With the basis of intercross and statistical analysis of multi-source urban transportation data, estimate the long term need for battery swap in specific areas and measure the correlation of such need among adjacent areas, and hereby determine the capacities and scales of multiple battery swap stations in these areas;.2. Determine the probabilistic distribution of customer arrival rate and state of charge of batteries, find out their variation regularity over seasons, weathers and different periods of the day; on this basis, compare different assignment and scheduling strategy of batteries such as FIFO, capacity first etc., and establish an optimization model with proposal of corresponding solution algorithms;.3. Establish a multi-objective dynamic optimization model of recharging scheduling of batteries at the station and design solution algorithms, conduct a comparison of a variety of charge scheduling strategies of batteries, under different operation and service modes, and explore the best operation and service mode..This project will put forward theoretical innovation in the areas of multi-objective combinatorial optimization and intelligent algorithm, the research achievements could provide theoretical foundation and decision support for the operations management of battery swap stations.

近年来，随着我国一系列新能源汽车促进政策的出台和各城市车辆电动化的推进，电动车数量迅速增加，对充电基础设施的需求大幅提升。在此背景下，本项目基于真实数据开展电动出租车换电站定容和智能调度研究，重点包括一下内容：.1. 通过对多源城市交通数据的交叉融合与统计分析，测算一定区域内的长期换电需求情况以及相邻区域的需求相关性，据此确定区域内换电站的容量和规模；.2. 确定换电站车辆到达率和电池荷电状态的概率分布函数，随季节、天气、星期及每天不同时段的变化规律；在此基础上，比较先进先出、荷电量优先等不同的电池分配调度策略，建立优化模型并提出求解算法；.3. 建立换电站电池充电调度的多目标动态优化模型并设计求解算法，比较换电站在不同的定价和服务模式时的充电调度策略，探究最佳的运营模式。.本项目将在多目标组合优化和智能算法方面提出理论创新，研究成果可为实践中的换电站运营管理提供理论依据和决策支持。

本项目通过数据分析建立了需求模型，而后以此为输入，采用混合整数规划、约束规划等方法对换电站选址定容问题、电动车路径规划问题以及大规模系统调度问题进行了数学建模，并引入鲁棒优化方法刻画动态环境中的不确定性。项目设计了蚁群算法、基因算法、自适应邻域搜索等智能算法以及列生成等精确求解算法进行问题求解。以上方法，一方面具有创新性且性能优异，推动了调度问题这一运筹学经典领域的理论进展，具有重要的科学和理论意义；另一方面，契合我国新能源战略和经济发展的要求，能高效解决日常经济生活中的优化问题，为交通运输及其他方面的运营管理提供可靠的决策参考和理论支持，并产生经济效益。与此同时，项目培养了一批对科研具有浓厚兴趣和较强创新能力的研究生人才。