基于LSTM神经网络的电动汽车充电站需求响应特性封装及配电网优化运行

随着电动汽车的快速增长,大规模电动汽车充电具有随机性、时空耦合性的特点,对配电网运行电压造成越限风险。通过基于价格的需求响应,引导电动汽车在大时空范围有序合理地充电成为重要的技术手段。文章研究基于数据驱动的电动汽车充电站需求响应特性及其参与配电网运行优化调度问题,首先提出单体电动汽车充电模型和计及交通网络拓扑结构的电动汽车行驶特性,建立了区域电动汽车充电站负荷需求响应计算方法;在此基础上,提出了基于LSTM深度神经网络的电动汽车充电站需求响应模型封装方法,得到电动汽车充电站充电成本和充电功率响应之间的映射模型;接着,构建了考虑电动汽车充电站需求响应的区域配电网电压运行优化模型,并采用粒子群算法进行求解;最后,通过对包含3个充电站的33 节点系统的算例对比分析,验证了所述电动汽车充电站需求响应及其参与配电网优化运行方法的有效性,为数据驱动方法解决电动汽车充电和需求响应问题提供借鉴。

Demand Response Package Model of Electric Vehicle Charging Station Based on LSTM Neural Network and Optimal Operation of Distribution Network

With the rapid growth of electric vehicles (EVs), the charging characteristics of large-scale EVs are randomness and spatiotemporal coupling, which poses a risk of exceeding the limit on the operating voltage of distribution network. Through the demand response (DR) based on price, it has become an important technical means to guide the orderly and reasonable charging of EVs in a large space-time range. In this paper, the DR characteristics of EV charging station based on data-driven and its participation in the operation optimization of distribution network are studied. Firstly, the charging model of single EV and the driving characteristics of EV considering the topological structure of traffic network are proposed, and the load simulation calculation method of regional EV charging station is established. On this basis, the electric power system based on LSTM deep neural network is proposed. The mapping model between charging cost and power response of EV charging station is obtained by encapsulating the DR model of EV charging station. Furthermore, a voltage operation optimization model of regional distribution network considering the DR of EV charging station is constructed, and the model is solved with particle swarm optimization algorithm. Finally, the comparison and analysis of the 33-node system with 3 charging stations verify the effectiveness of the proposed method of EV charging station DR and its participation in distribution network operation optimization. It provides reference for data-driven method to solve the problem of EV charging and demand response.