提出一种基于时空耦合特性和深度学习模型的充电站运行状态预测方法.首先,基于充电站历史运行数据和所在区域的交通通行速度数据集,利用k-means聚类方法将充电站划分为不同类型,分析充电站运行状态在时间上的特性;建立单个充电站的"偏移量-交通-时间"三维矩阵模型,深度挖掘充电站运行状态与周边交通状况在时间和空间上的耦合相关性.其次,将充电站状态与交通状况的时间滞后相关特性进行空间重构,利用卷积神经网络进行特征提取,通过长短期记忆网络进行时间序列预测,构建基于Keras深度学习框架的充电站运行状态多步预测模型.最后,以20个充电站的真实运行数据进行验证,并与多种预测算法进行对比,结果表明,所提方法具有较高的预测精度.

针对含电动汽车(EV)和分布式光伏的主动配电网(ADN),提出了一种基于机会约束规划的能量管理方法.首先,基于EV用户的出行特性和需求,构建了基于分段线性化的EV智能充放电决策模型;其次,通过支路潮流模型与二阶锥松弛,构建了含EV与分布式光伏的ADN一体化数学模型,EV集群作为灵活可控单元主动参与ADN的能量管理;然后,为了充分考虑分布式光伏的不确定性,采用机会约束规划方法描述了含不确定性参数的数学模型,并对模型中的机会约束条件进行确定性转换;最后,考虑不同城市功能区内含EV与分布式光伏的ADN场景,进一步对比分析了不同的机会约束条件置信水平下EV对ADN经济安全运行的影响.仿真结果表明,适度降低机会约束条件的置信水平能在保证EV用户出行需求的同时充分地挖掘EV充放电行为的灵活性,进一步实现ADN的灵活运行.

为了解决极端灾害过后的负荷恢复问题,提出了一种城市道路抢修辅助重要负荷恢复的电动汽车能量时空分层调度策略.在第1阶段,通过建立并集成电动汽车路径规划模型和多时段协同的重要负荷恢复模型,确定电动汽车能量的最优空间配置;在第2阶段,建立道路抢修队路径规划模型并与电动汽车路径规划模型进行统筹优化,确定道路抢修队的抢修顺序与路径,求解电动汽车到达充放电站的时间与剩余能量;在第3阶段,统筹配电网内各供电源和各重要负荷,以最大化加权负荷供电时间为目标,确定能量在时间尺度上的优化配置.最后,考虑配电网-交通网耦合的灾后故障场景,对不同策略的调度结果进行对比分析并探究道路抢修对于负荷恢复的辅助作用.结果表明,所提策略可以充分发挥电动汽车能量的时空灵活性,合理协调电动汽车和道路抢修队之间的行动决策,在一定程度上提高了重要负荷的恢复效果.

针对现有充电站的规划布局不能充分考虑电动汽车动态充电需求分布和用户充电排队问题的不足,提出了一种基于全球定位系统(GPS)轨迹挖掘的电动出租车充电站规划方法.通过对出租车的GPS轨迹数据和城市交通态势数据进行数据处理,挖掘城市居民打车需求的起屹点(OD)分布特征;设计电动出租车的充电仿真算法,模拟实际场景中电动出租车的接单行为、行驶行为及充电行为,建立电动出租车的充电需求时空分布预测模型;在此基础上,综合考虑充电站建设运行成本、电动出租车到站时间成本及充电等待时间成本,建立充电站规划模型.通过实际算例验证了所提规划模型的有效性,并进一步分析了充电站的建设成本系数、电动出租车的时间成本折算系数、权重系数大小等参数对规划结果的灵敏度.结果表明:居民打车需求的OD分布决定了电动出租车的充电需求时空分布情况;在进行充电站规划时,充电机数量受充电站数量的变化影响较大,且电动出租车时间成本的变化对总成本的影响相对较明显.

提出一种面向提高风电接纳能力的智慧建筑能量管理策略.首先,基于建筑热惯性,构建考虑建筑物内部不同制热区域的能耗预测模型;其次,基于支路潮流模型与二阶锥松弛方法,构建集成智慧建筑的主动配电网(active distribution network,ADN)统一数学模型.随后,基于模型预测控制方法,在保证用户舒适性前提下,对ADN进行能量管理;最后,基于冬季制热场景,通过多种暖通空调(heating,ventilation and air conditioning,HVAC)调控方案对智慧建筑参与ADN优化调度进行分析验证.算例表明,与未考虑ADN与智慧建筑集成的模型及方法相比,所提方法基于集成智慧建筑的ADN统一模型,充分利用了HVAC设备运行模式灵活性和居民舒适温度区间,在保障居民温度舒适性的同时,进一步提高电网的风电接纳能力,并保证ADN的经济安全运行.

大量电动汽车(EV)入网可为电网提供辅助调频服务.针对车网互动(V2G)造成的惯量、阻尼缺失的问题,采用虚拟同步机技术,使得EV具有与同步发电机类似的惯性阻尼特性和频率调节特性.针对EV辅助调频问题,提出考虑用户充电需求的EV智能充放电控制策略.首先,提出调频参与度因子,根据期望荷电状态以及用户的计划充电时间自适应确定一次调频系数,进而得到一次调频功率;然后,充分考虑车主需求,设计T-S型模糊控制器,根据电网频率偏差及调频参与度因子得到二次调频功率,实现EV的智能充放电;最后,在不同电池初始状态以及用户需求情况下进行仿真验证.仿真结果表明,所提控制策略可以根据电网频率波动情况以及用户充电需求程度智能控制EV进行充放电,在满足用户充电需求的情况下参与电网辅助调频服务,减小了电网频率波动,提高了电力系统的稳定性.

针对港口中包含岸电、海上风机与储能的混合能源系统,提出一种系统优化运行方法.该方法建立混合能源系统模型,其中包括岸电负荷预测模型、岸电动态电价模型、风机模型与储能模型.针对岸电负荷不确定性导致的预测精确度较小的问题,提出一种分频段预测方法,该方法利用小波包分解进行信号分频并根据不同频段特点选择不同的预测方法,提高了预测准确率;针对岸电供售电价格机制不明确的问题,提出与岸电用电量线性相关的阶梯服务费模型和随负荷变化动态调整的电价模型,平衡了港口企业、航运企业与电网企业三方利益;针对整数变量引入形成的混合整数非线性规划问题,采用模型预测控制滚动优化方法,利用CPLEX+YALMIP进行求解,改善了开环优化方法在不确定性环境下误差较大的问题,得到了系统总运行成本最小的优化结果.

由于电动汽车用户难以找到充电时间与充电地点之间的平衡点,不能准确把握何时何地进行充电行为.文中首先选取能够准确反映实际道路中用户独特驾驶特性的行驶工况特征参数,采用工况识别法构建电动汽车在实时动态路况下的剩余电量估算模型,判断其出行过程中何时有充电需求;当有充电需求时,通过预测充电站的抵达车辆数,建立电动汽车排队等待时间模型,为用户规划有效充电时段,作为选择充电地点的依据;考虑到充电时间与充电地点的耦合关系,从用户角度出发,在电池剩余电量的约束下,构建用户出行距离、出行时间及充电成本三者权值之和最优为目标的电动汽车充电路径模型,并将其应用于实际交通路网区域中,采用蚁群算法对其进行仿真验证.

该文以含分布式电源的配电网可靠性评估为主要内容,首先研究居民、商业、工业的日负荷特性,建立反映用户用电特性的时序负荷模型.根据分布式电源的出力和负荷需求,建立储能协调运行模型.在此基础上,研究分布式电源的接入以及不同分布式电源出力对各类用户负荷可靠性的影响.最后,基于时序蒙特卡洛模拟算法,对IEEE RBTS BUS6-F1馈线改进系统进行可靠性计算.算例结果表明,分布式电源的接入能有效提高配电网可靠性,且分布式电源对于不同的用户负荷可靠性影响不同.

由于电动汽车在行驶途中电量耗尽的风险较大,亟须研究一种有利于减轻用户出行焦虑的电动汽车充电路径规划方法.文中通过选取能够准确反映实际道路中用户驾驶特性的行驶工况特征参数,分析各特征参数与耗电量之间的相关性以及特征之间的相关性强弱,采用主成分分析对特征进行降维,基于信息熵模糊聚类方法对行驶工况进行分类,构建电动汽车行驶途中的动态能耗模型.并基于此考虑路径选择及电池剩余电量约束,建立以出行总距离、总时间及充电价格三者权值之和最小为目标的电动汽车充电路径规划模型.以某市实际交通路网规划18 km×18 km区域,分析采用实时能耗对充电路径规划的影响以及不同优化目标对用户充电路径优化结果的影响,验证了所提规划方法的可行性及有效性.

As the key element of active distribution systems (ADSs), energy storage systems (ESSs) play multiple important roles including enhancing reliability and improving economic efficiency. In this study, a novel day-ahead operation strategy is proposed for ESSs, and it is developed as a fuzzy multi-objective model without violating the operating constraints. In this model, three objectives are taken into account adequately and equally, including the enhancement of reliability, the improvement of economic efficiency, and the promotion of operation friendliness. In order to harmonize the relation among these objectives, the approach based on the fuzzy mathematical theory is adopted. To solve this proposed fuzzy multi-objective model, the gravity search algorithm (GSA) is adopted. Simulation results obtained based on the modified IEEE-33 bus distribution system are provided and discussed, where the availability and the feasibility of the proposed model and adopted algorithm are verified.

Electric vehicle users are focused on the convenience of charging,which may cause some problems for power grid at the same time.So how to take both users' convenience and security of power grid into account is an urgent problem to be solved.Aiming at this problem,an ordered charging strategy of electric vehicles based on users' driving behavior is proposed.The principal component analysis and fuzzy clustering algorithm are used to study the electric vehicle users' driving behavior and predict the driving distance,based on which to calculate the charging power of each charging process and to dispatch the electric vehicles according to the load curves of local distribution network.By simulating electric vehicle users' driving behavior,the load curves of distribution network when electric vehicle disordered charging and ordered charging with different user response rates are analyzed and compared,whose results show that the proposed strategy can reduce the peak valley of distribution network load effectively and can increase the electric vehicle users' motivation to ordered charging.

A model based on virtual power plant is proposed to control the charging process in real time,which not only considers the travel behaviors of EV (Electric Vehicle) owners but also relieves the load imbalance of distribution network to realize the interaction between EVs and distribution network.A twolayer algorithm is proposed to solve the model.Its outer layer adopts the static routing algorithm to obtain the load data and the back/forward sweep method to calculate the nodal voltage and power loss of distribution network;its inner layer adopts the aging algorithm to judge the accessing phase of EV and the round-robin scheduling algorithm to balance the three-phase loads of distribution network.The power loss and the load imbalance degree of distribution network can be reduced by adjusting the charging time of EVs.Simulative results show that,the proposed strategy can not only avoid the charging peak and lower the three-phase load imbalance degree of distribution network but also reduce its power loss and ensure its safe and economical operation.

Large scale charging load of electric vehicles will increase the degree of the deviation between node voltage and nominal voltage in a local distribution network, and the deviation which caused by the charging load is random on time and place, traditional reactive compensation method can solve the above problem, but the economy is bad.So a voltage control strategy based on reactive power compensation by electric vehicles is proposed, which aims at implementing reactive power compensation for distribution network and changing the charging active power at the same time through regulating the operating power factor of the chargers effectively.The strategy can ensure the security and stability of the node voltage and the efficiency of distribution network will not decrease.A strategy is proposed to calculate the controllable scope of operating power factors of chargers after analyzing the travel behavior characteristic in a working area of first-tier/second-tier city in China and the operating characteristic of charger.Then, the node voltage regulation model is used which considers all operating power factors of chargers as variables and considers the following conditions as the optimization objective: the deviation between node voltage and nominal voltage is minimum, the variance of the node voltage from current time to all time before is minimum, the state of charge charged in at unit duration is maximum.Finally, immune algorithm is used to get the values of variables.The rationality of the proposed node voltage regulation strategy can be validated through the simulation.

随着光伏发电和电动汽车的快速发展,电网的安全和经济运行受到了更多的关注,开展关于能量管理问题的研究有助于保证电网的稳定和高效运行.文中针对并网模式下计及电动汽车和光伏储能的微网能量管理问题,研究了根据电动汽车电池的荷电状态进行能量管理的功率计算模型,提出了兼顾光伏出力、电动汽车充放电功率、电网电价时段划分及储能能量状态的能量管理策略.在此基础上,建立了以电动汽车充放电功率为优化变量,以减少微网运行费用为目标的能量管理模型,并采用基于遗传算法与粒子群算法的混合优化算法(GAPSO)进行求解.最后以实际微网为例进行仿真,结果验证了所述方法的有效性和可行性.

A charging control system model is developed based on the characteristics of local distribution network to meet the demands of both power grid and EV(Electric Vehicle) user,which includes three modules:TOU(Time-Of-Use) charging period division module,coordinated charging module and user response evaluation module. Combined with the results of user response evaluation module,the period division module applies the methods of fuzzy cluster and curve feature identification to divide the periods of TOU charging price,based on which,the coordinated charging module manages EV charging with three control objectives:minimum customer charging fee,minimum load curve variance and earliest charging service. Simulative results indicate that,compared with the uncoordinated charging management,users' charging demand is met with the minimum charging fee and the EV charging load is reasonably shifted from daily load peak period to valley period.

The construction and operation of electric vehicle (EV) charging station is the basic project in the EV industry development. This paper presents a dynamic optimizing method to improve the load characteristics of the distribution system based on the particle swarm optimization (PSO) algorithm. The optimizing strategy is operated by minimizing the load variance for a period as objective function, and the scale of charging station and the demands of customers as constraints. The customers’ travel performance is simulated by the stochastic method of probability distribution. And load characteristics of coordinated and uncoordinated charging are compared in different permeability. The conclusion shows that the dynamic optimizing strategy takes effect in mitigating load variance, optimizing distribution system characteristics and shifting peak load.

Aimed at reducing the peak-valley ratio of power grids,a time-period dividing method is proposed for the time-of-use(TOU) charging price of electric vehicles based on a comparative analysis on time periods of TOU price in power grids and peak-valley time periods of local distribution networks.Based on the TOU charging price proposed,a mathematical optimization model is developed with two control objectives,minimizing the customer charging fee and advancing the charging time of battery as early as possible.With the charging performance of customers simulated by Monte Carlo analysis,uncoordinated and coordinated charging modes are compared according to simulation results.It is indicated that,by responding to the TOU charging price proposed,the peak-valley ratio of power grids can be reduced and the customer satisfaction index can be enhanced.At the same time,the load curves of the local distribution network for different customer response coefficients are simulated to show that the more the customers that are influenced by the charging price,the more favorable it will be to smoothing the load fluctuation.

In order to manage charging load of EV, this paper proposed a charging dispatching control strategy based on TOU charging price and coordinated charging. Fuzzy cluster algorithm and curve identification method have been used for period division of TOU charging price. According to the result of TOU charging price, coordinated charging module manage EV charging with three control objectives, including minimizing the customer charging fee, minimizing load curve variance and advancing the charging time as early as possible. Simulated results indicate that, compared with uncoordinated charging result, the study meet the charging demand of users. And at the same time, the study not only improve the charging satisfaction of EV users, but also charge in the relative lower day and night load period to disperse EV charging power in peak load period.