Abstract:

Battery swapping cabinets have been successively distributed in cities to meet the increasing swapping demand for electric bicycles, which inevitably involves the station location of battery swapping cabinets, the sizing of battery supply, and the prediction of battery swapping demands. In order to improve the utilization rate and reduce the cost of battery swapping, a clustering and forecasting method of battery swapping demand by region is proposed. Firstly, K-means clustering was carried out on the location of the battery swapping cabinets, and the size of cabinet supply was optimized to improve the utilization rate; Then, the Autoregressive Integrated Moving Average model (ARIMA) is used to predict the short-time battery swapping demand. The results indicate that the ARIMA model has a high prediction accuracy in the demand prediction. Compared with other prediction models, better results can be achieved, which indicates that battery swapping demands tend to be linear with time. The optimization method on battery swapping cabinets and the short-term demand prediction results proposed in this paper provide data support for the location of battery swapping stations and the sizing of battery supply

Key words: intelligent transportation, battery swapping demand forecast, ARIMA model, electric bicycle, K- means clustering

摘要：

换电企业在城市内建立换电柜，满足不断增长的电动自行车换电需求，涉及到换电柜的选址，电池的投放和换电需求的预测。本文分析了国内某大型换电企业的换电订单数据，发现换电柜存在使用严重不均衡问题,为提高使用率，降低换电成本，提出按区域对换电需求量进行聚类并预测的方法。首先，对换电柜位置进行K-means聚类，据此优化换电柜的投放量，提高使用率；随 后，采用整合移动平均自回归模型(Autoregressive Integrated Moving Average model, ARIMA)预测短时换电需求。实验发现，ARIMA模型在短时换电订单的需求预测上具有较高的预测精度，与其他基线模型相比，各指标均为最好，说明换电需求在时间上更趋于线性关系。本文提出的换电柜优化方法和短时需求预测结果为换电企业的换电柜选址和电池投放量提供数据支持。

关键词: 智能交通, 换电需求预测, ARIMA模型, 电动自行车, K-means聚类