计及评价指标冲突的电动汽车充电负荷区间预测

电动汽车充电负荷具有强随机性，且受电池容量与用户用车行为影响。为有效预测充电负荷时序分布，本文提出一种计及评价指标冲突的充电负荷区间预测方法。首先，该方法分析日间充电负荷间时序相关性，并用强相关历史日充电负荷数据构建充电负荷预测所需的特征集。接着，采用弯曲高斯过程（warped Gaussian process , WGP）方法，并结合多种协方差函数来构建多个充电负荷区间预测模型。为解决多指标评价存在冲突和仅选择最优的一个预测模型会出现极端误差问题，本文应用面积灰关联决策方法，对各模型开展计及评价指标冲突的综合评价，并依据获取的面积灰关联贴近度，构建电动汽车充电负荷组合区间预测模型。实验结果表明，本文提出的方法能够获得更精确、覆盖率更高的充电负荷预测区间。

Interval Forecasting of Electric Vehicle Charging Load Considering Evaluation Index Conflicts

The electric vehicle (EV) charging load has strong randomness. It is affected by battery capacity and user behavior. To effectively forecast the temporal distribution of charging load, this paper proposes a charging load interval forecasting method that takes into account conflicting evaluation indicators. First, this method analyzes the temporal correlation between daytime charging loads and constructs a feature set for charging load forecasting using strongly correlated historical daily charging load data. Next, the Warped Gaussian Process (WGP) method is adopted, and multiple covariance functions are combined to construct multiple charging load interval forecasting models. To solve the problem of conflicts in multiple index and extreme errors in selecting only the optimal forecasting model, the area grey correlation decision-making method is applied. Conduct a comprehensive evaluation of each model taking into account the conflict of evaluation index, and obtain the area grey correlation closeness of each model. Construct an EV charging load combination interval forecasting model based on the proximity degree of area grey correlation. The experimental results show that the method proposed in this article can obtain more accurate and higher coverage charging load forecasting intervals.