基于时空耦合特性的充电站运行状态预测

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。

Operation State Prediction of Charging Station Based on Spatio-Temporal Coupling Characteristics

A prediction method for the operation status of charging stations based on spatio-temporal characteristics and the deep learning model is proposed.Firstly,based on the historical operation data of charging stations and the traffic speed data set in the area,the k-means clustering method is used to classify the charging stations into different types and the temporal characteristics of the operation status of charging stations are analyzed.The“offset-traffic-time”three-dimensional matrix model of individual charging stations is established to deeply explore the coupling correlation between the operation status of charging stations and the surrounding traffic conditions in time and space.Secondly,the spatial reconstruction of the time-lagged correlation between charging station status and traffic conditions is carried out,the feature extraction is carried out by convolutional neural network,and the time series prediction is carried out by long short-term memory network,so as to build a multi-step prediction model of charging station operation status based on Keras deep learning framework.Finally,the proposed method is validated with the data of 20 charging stations and compared with various prediction algorithms,and the results show that the proposed method has high prediction accuracy.