基于图WaveNet的电动汽车充电负荷预测

Graph WaveNet Based Charging Load Forecasting of Electric Vehicle

为了更好地挖掘电网-交通网强耦合态势下电动汽车充电负荷的时空动态特征,提高充电负荷预测精度,提出了一种基于图WaveNet的电动汽车充电负荷预测框架。首先,将耦合的电网-交通网中的充电站看作充电负荷节点;然后,把充电站的充电负荷数据作为节点的特征信息,将各个节点构造成一张图,并把蕴含充电负荷空间维信息的图和充电负荷的时间维信息输入自适应图WaveNet框架中进行预测;最后,以中国某市城区内的充电站负荷数据为例,将基于自适应图WaveNet框架的预测结果与现有方法的预测结果进行对比,验证了所提方法的正确性和有效性。

In order to better mine the spatial-temporal dynamic characteristics of electric vehicle charging load under the situation of strong grid-transportation network coupling and improve the accuracy of charging load forecasting,a framework of graph Wave Net based charging load forecasting for electric vehicles is proposed.First,the charging stations in the coupled grid-transportation network are regarded as charging load nodes.Then,by regarding the charging load data of the charging stations as the characteristic information of the nodes,all the nodes are constructed into a graph,and the graph containing the spatial-dimension information of charging loads and the time-dimension information of charging loads are input into the adaptive graph Wave Net framework for forecasting.Finally,taking the charging station load data in an urban area of a city in China as an example,the forecasting results based on the adaptive graph Wave Net framework are compared with the forecasting results of the existing methods,and the correctness and effectiveness of the proposed method are verified.