计及电网和用户行为中多重随机性的电动车群有序充放电策略研究

项目名称： 计及电网和用户行为中多重随机性的电动车群有序充放电策略研究

项目编号： No.51477097

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 电工技术

项目作者： 冯冬涵

作者单位： 上海交通大学

项目金额： 75万元

中文摘要： 规模化电动汽车的有序充放电问题含有多重随机因素，电动汽车在时间和空间上的不确定性、接入电网中随机间歇性电源出力和输变电设备故障停运的不确定性等多重随机因素间存在着复杂的非线性条件相关特征。本项目拟在系统研究主导随机因素间时空相关性的基础上，深入挖掘电动汽车作为灵活调节负荷（充电模式）和分布式移动储能单元（充/放电模式）的潜力，从整体角度优化电动汽车的充放电行为。有序充放电的实现牵涉到各类电动汽车用户、电网运营企业和电动汽车服务公司等多元主体，本项目拟运用合作博弈理论科学分配优化所得的综合效益，在此基础上设计激励相容的协调机制，并构建不完全信息动态多阶段博弈模型研究协调机制下各主体的均衡策略，以实现全局优化结果对各分散决策主体的有效引导。针对问题的复杂程度，本项目拟结合模型特征综合运用主成分分析、分层抽样、模型预测控制和动态系统降维等技术研制有效的求解算法，并进行验证。

中文关键词： 非线性时变相关性；模型预测控制；随机动态优化；不完全信息多阶段博弈；机制设计理论

英文摘要： Grid integration of large scale electric vehicles are confronted with multiple stochastic factors, including uncertainties of electric vehicles at time and space, uncontrollable renewable power such as the output of wind turbine and photovoltaic panels, as well as the random fault of new smart components in the grid, etc. Based on a systematic analysis and comprehensive modeling of the correlation structure between these stochastic factors, this research project aims at cultivating the potential of electric vehicles as adjustable load under charging-only mode and distributed mobile electricity storage units, so that to optimize the charging/discharging behavior from a systematic viewpoint. Analytical model analysis (based on recent research advances of model predictive control technologies) and multi-layer high efficiency sampling technologies will be used to solve the resultant stochastic dynamic programming problem. Charging/discharging coordination problem involves multiple entities such as electric vehicles users, charging/discharging service providers and grid operation organizations, etc. These entities, though with different interests, need to cooperate with each other so that to fully utilize electric vehicles for maximal benefits including energy-saving and CO2 reduction. However, the global optimal solution does not necessary mean the profit/benefit maximization of each and every relevant entity. This project plans to use cooperative game theory to allocate the overall benefits from coordinated charging/discharging optimization. From the allocation, coordination mechanism can be established based on the mechanism design theory. Then an incomplete information dynamic non-cooperative game model is established to analyze the equilibrium strategies of the relevant entities under the coordination mechanism. Based on an in-depth comparison between the centralized optimal coordinated strategy and the decentralized equilibrium strategies under the incentive compatible coordination mechanism, a comprehensive coordination strategy utilizing the advantages of both centralized and decentralized models and its supporting mechanism will be developed. According to the complexity of the problem, Principal Component Analysis, stratified sampling, model predictive control and dimension reduction techniques will be used based on the characteristics of the model, in order to develop efficient solution algorithms. Beside the computer simulation and lab test, Campus level and regional level field test will be carried out, so that to further demonstrate the validity of the proposed models, methodologies and strategies, as well as yield more practical and relevant business models, local policies and other application solutions. The research fruits of this project can lay the scientific foundations for fully utilizing the energy saving and environment protection potentials of electric vehicles to support the safe and economic operation of the power systems.

英文关键词： nonlinear time-dependent correlation;model predictive control;stochastic dynamic programming;incomplete information multi-stage game;mechanism design theory