基于智能代理分布式管理系统的电动汽车充放电策略

随着电动汽车的发展与普及，提出了一种用于协调电动汽车充放电的智能多代理分布式管理系统。建立以用户充电费用支出和负荷曲线峰谷差最小化为目标的多目标优化策略，采用蒙特卡洛方法模拟电动汽车用户行为。在峰谷电价的引导下，用户利用谷时段充电、峰时段放电提高自身利益，在初步得到理想的充电时段区间后，引入学习训练机制调整可间断充放电分段数，使负荷曲线最终达到理想的削峰填谷效果，采用宽容分层序列法求解多目标优化。优化结果表明本文的策略平抑负荷波动效果显著，能有效减少用户支出，增加用户收益，实现电网与用户互利双赢。

Electric Vehicles Charging and Discharging Strategy Based on Intelligent-agent Distributed Management System

With the development and popularity of electric vehicles， an intelligent multi-agent distributed management system for coordinating charge and discharge of electric vehicles is proposed. Multi-objective optimization strategy is established with the goal to minimize the charging cost of users and the peak-valley difference of load curve， and Monte Carlo method is used to simulate the users’ behavior. Under the guidance of peak-valley electricity price， users charge in valley time and discharge in peak time to improve their interests. After obtaining the initial ideal charging time interval，a training-learning mechanism is introduced to adjust the number of discontinuous charging and discharging segments until the load curve finally achieves the ideal peak-shifting and valley-filling effect. Tolerant hierarchical sequence algorithm is adopted to solve multi-objective optimization. Finally，the optimizatin results show that the proposed strategy has significant effect on reducing load fluctuation， and can effectively reduce users’ expenditure， increase users’ revenue， realize mutual benefits and double wins between the power grid and users.