大规模电动汽车的实时V2G功率容量并行预测

Parallel Predictionfor Real-time V2G Power Capacity of Large-scale Electric Vehicles

随着Vehicle-to-grid (V2G)技术的发展,大规模电动汽车所聚合的电池可以作为一个很大的分布式储能系统为配网或微网提供许多辅助支持服务。精确和快速地进行电动汽车V2G功率预测是实现这些服务的关键。用户行为的不确定性和随机性给V2G功率容量预测带来了困难。实时充电数据和用户充电需求的采集,为精确的V2G功率容量预测提供了保障。但是随着电动汽车大规模接入,充电数据呈指数级增长,大规模数据如何快速处理给V2G功率容量预测造成了新的难题。并行预测技术可以解决这个难题:通过搭建基于Spark并行预测平台,在实时充电数据采集和考虑用户需求的基础上,利用分布式并行内存处理技术,最终建立了大规模电动汽车实时V2G功率容量并行预测模型。

With the development of vehicle to grid（V2 G） technology,the battery for the large-scale electric vehicle can be used as a large distributed energy storage system to provide many auxiliary support services to the distribution grid or micro-grid. Quick accurate V2 G power prediction for electric vehicles is the key to the realization of these services. The uncertainty and randomness of user behavior makes it difficult to predict V2 G power capacity. Real-time collection of charging data and user＇s charging demand provides a guarantee for accurate prediction of V2 G power capacity. However,with large-scale access of electric cars,charging data grows in the exponential order so that how to quickly process big data becomes a new challenge for V2 G power capacity prediction. Parallel prediction technology could solve this problem： by establishing a Spark-based parallel prediction platform,on the basis of real-time charging data collection and under consideration of user demand,a parallel prediction model was finally established for real-time V2 G power capacity of large-scale electric vehicles by use of the distributed parallel memory processing technique.