面向服务区能源系统经济运行的电动汽车充电负荷调度导向预测策略  

Dispatch-centric forecasting strategy of electric vehicle charging load for economic operation of energy system in service area

作　　者：陆旭东[1] 张晓峰[1] 商淑杰 贺雅楠 束娜[4] LU Xudong;ZHANG Xiaofeng;SHANG Shujie;HE Yanan;SHU Na(China Academy of Transportation Sciences,Beijing 100029,China;Shandong Expressway Infrastructure Construction Co.,Ltd.,Jinan 250000,China;Hainan Traffic Control Energy Co.,Ltd.,Haikou 570000,China;Shandong Electric Power Engineering Consulting Institute Co.,Ltd.,Jinan 250010,China)

机构地区：[1]交通运输部科学研究院,北京100029 [2]山东高速基础设施建设有限公司,山东济南250000 [3]海南交控能源有限公司,海南海口570000 [4]山东电力工程咨询院有限公司,山东济南250010

出　　处：《供用电》2025年第1期96-105,共10页Distribution & Utilization

基　　金：世界银行项目“分布式可再生能源在交通领域的应用政策机制研究”(A1-CS-2021-006);交通运输部清单项目“山东省高速公路零碳服务区建设关键技术研究与工程示范”(2022-ZD3-020);海南省交通科技项目“海南省高速公路零碳服务区建设关键技术研究与示范”(HNJTTKXC-2023-6-28-05)。

摘　　要：公路交通作为碳排放和能源消耗的重要领域,其绿色化转型和能源可持续发展已成为迫切需求。随着电动汽车(electric vehicles, EV)的快速普及,建设高速服务区综合能源系统对于实现公路用能绿色化具有重要意义。高速服务区交通-综合能源系统(transportation-integrated energy system, TIES)的经济运行依赖于准确的日前负荷预测。然而,电动汽车充电负荷的日益增长为TIES的负荷预测及其经济运行优化带来了新的挑战。以最小化统计误差为目标的负荷预测模型难以确保TIES的能源调度结果的最优性。提出了一种以服务区经济运行为导向的含电动汽车充电负荷的预测策略,旨在降低能源调度成本。该策略以深度多层感知器作为基础学习器,结合日期和充电价格信息构建特征工程和含电动汽车充电负荷的预测模型。进一步,通过分析TIES两阶段能源调度问题的特性,提出了一种具有时间不对称和正负偏差不对称的负荷预测微损失函数,将最优调度问题的误差灵敏度融入预测模型中,以优化不同时段的预测误差分布。以江苏某地区电动汽车负荷数据集进行仿真,验证所提方法可以有效降低TIES用能成本,提升用能经济性。Highway traffic is a significant area for carbon emissions and energy consumption.With the rapid proliferation of electric vehicles(EVs),constructing integrated energy systems in highway service areas aids in the green transition of highway energy use and the sustainable development of energy systems.The economic operation of the transportation-integrated energy system(TIES)in highway service areas relies on accurate day-ahead load forecasting.The increasing EV charging load poses challenges to the load forecasting and economic operation of TIES.Existing load forecasting models that minimize statistical errors often fail to ensure optimal energy dispatch outcomes for TIES.This paper proposes a forecasting strategy incorporating EV charging loads,oriented towards the economic operation of service areas,to reduce energy dispatch costs.Utilizing deep multilayer-perceptions as base learners,the forecasting model incorporates feature engineering with date and charging price information,including EV charging loads.Subsequently,by analyzing the characteristics of the two-stage energy dispatch problem in TIES,a load forecasting micro-loss function with time asymmetry and positive-negative bias asymmetry is proposed.This integrates the error sensitivity of the optimal dispatch problem into the forecasting model to improve the distribution of forecasting errors across different time periods.Simulation results based on a charging load dataset from a service area in Jiangsu province demonstrate that the proposed method effectively reduces TIES energy costs and enhances energy efficiency compared to forecasting models using symmetric loss functions in terms of time and bias.

关 键 词：交通能源系统 充电负荷预测 调度导向预测 预测-调度一体化 不对称损失函数 

分 类 号：TM73[电气工程—电力系统及自动化]