基于电动汽车的电力备用服务商策略优化的实验经济学仿真研究

The popularization and application of electric vehicles (EV) is part of the solution to control serious air pollution problems in China's metropolis such as haze. But one of the main obstacles for EV industry is the delay of building a thorough charging service system. Successful potential business models are much required for the charging industry to be stimulated and launch a good start.As an innovative model, charging EVs served as reserves have been mentioned in many studies, however these studies show limitations on the random hypothesis of EV user behaviors, which is because the knowledge on the inside mechanisms that guide users' charging behaviors is limited. Based on the experimental economics method, the project tries to crack the difficulties in modelling subjective behaviors by allowing participants to interact with simulation models through inputing decisions on computer-based user interfaces. The size of effective subjects can be increased and the total time cost can be decreased through the following improvements on the basic method: the experiments on imperfect optimization directions can be filtered; computer agents can be created whose "behaviors" are configured to coincide with the probability distribution of participants' behaviors; participants and computer agents can be mixed in one simulation environment. Based on the above method, the project will improve the design of charging contracts, decouple the entire optimization problem for reserve service providers based on the trajectory-based index system relevant to risk and cost evaluations, and finally realize the coordination in decisions between pricing optimization of charging contracts and bidding optimization in the reserve market.

电动汽车的推广应用有利于控制大城市雾霾等严重空气污染问题，却受制于充电服务体系建设的滞后，需要进一步挖掘潜在的商业模式，激励充电服务产业迅速起步。电动汽车参与电力备用服务的创新模式早已被提出，但相关研究均止步于用户行为的随机性假设，源于缺乏对支配电动汽车用户充电行为的内在规律的认识。本项目基于实验经济学方法，通过允许受试者参与仿真实验并输入决策，破解主观行为的建模难题。通过按最速优化方向筛选实验，产生符合多名受试者决策结果概率分布的计算机代理，采用受试者和计算机代理混合仿真的方法，以扩大实验经济学研究的有效实体规模，缩短研究总耗时。基于上述方法，改善充电合约的设计；利用基于动态轨迹提取的价值、风险评估指标对电力备用服务商策略优化问题进行解耦，实现对充电合约定价和备用市场报价决策优化的协调。

充电设施不足是当前制约电动汽车发展的关键因素之一。而充电设施的健康发展离不开电动汽车与电网的友好互动，结合当下弃风弃光严重、电网对新能源消纳能力不足的问题，开展电动汽车参与备用服务的研究工作具有很强的现实意义。本项目围绕电动汽车参与电网备用服务这一主题，聚焦于其中存在的充电合约设计、需求弹性评估以及运营优化等几类关键问题。. 针对第一类问题，项目首先设计了用户出行和充电意愿的调查问卷，提出了反映用户意愿深层信息的提取方法，以及与受访人群具有相近意愿分布特征的多代理模型构建方法，多代理模型解决了用户意愿难以数学模型化的难题，对掌握多重因素影响下的用户敏感区间具有重要指导意义；在总结上述分析中反映出的用户关切基础上，提出了兼顾用户出行需求和电网调控需求的充电合约设计，这是有序充/放电的核心和前提。. 针对第二类问题，建立了有助于电动汽车参与电网备用服务的能力、价值评估指标体系，填补了有关电动汽车需求弹性分析领域的空白，同时能指导激励机制设计、合约定价和有序充/放电优化；首次对电动汽车短时备用能力进行了详细量化分析。. 针对第三类问题，提出了电动汽车充电合约定价及运营报价决策的一系列协调优化方法，为电动汽车需求弹性的最大化利用、也为本项目成果的工程落地提供了直接的工具支持；此外，本项目研发的基于价值网络的电动汽车个体充/放电策略优化算法在精度不减的前提下，计算效率大大高于主流商用解算器，其计算耗时仅为CPLEX的20%-2%。. 本项目为综合运用问卷、仿真等新型实验经济学方法论对电动汽车“人-车-网”的特殊互动形态进行更全面的研究提供了范例，也同时为计划向电动汽车用户提供有序充/放电、向电网提供需求侧备用服务的售电商提供了一整套先进的分析优化工具包。.