基于序列理论的电网用户用电负荷自动预测研究  

Research on Automatic Power Load Prediction of Power Grid Users Based on Sequence Theory

作　　者：冯波 张瑞 孙冲 梁静 吴彬彬 FENG Bo;ZHANG Rui;SUN Chong;LIANG Jing;WU Bin-bin(State Grid Hebei Electric Power Co.,Ltd.,Marketing Service Center,Shijiazhuang 050035 China)

机构地区：[1]国网河北电力有限公司营销服务中心,河北石家庄050035

出　　处：《自动化技术与应用》2023年第2期17-20,共4页Techniques of Automation and Applications

摘　　要：受到用电负荷时间序列的波动性较大的影响,用电负荷预测存在预测误差较大的情况。为此,提出基于序列理论的电网用户用电负荷自动预测平台。以历史电网用户用电负荷数据为基础,对历史电网用户用电负荷数据进行预处理,并利用经验模式分解算法分解负荷序列,选择时间序列理论中的ARMAX模型构建负荷预测模型,以历史负荷作为输入,得出未来某个时间点的用电负荷预测值。实验结果表明:与三种预测平台相比,所研究方法计算得出的MAE和MAPE指标数值更小,说明所研究方法的预测精度更高,误差更小。Affected by the large fluctuation of power load time series, there is a large prediction error in power load forecasting. Therefore, an automatic load forecasting platform for power grid users based on sequence theory is proposed. Based on the power load data of historical power grid users, the power load data of historical power grid users are preprocessed, the empirical mode decomposition algorithm is used to decompose the load sequence, the ARMAX model in time series theory is selected to build the load forecasting model, and the historical load is used as the input to obtain the power load forecasting value at a certain time point in the future. The experimental results show that compared with the three prediction platforms, the Mae and MAPE indexes calculated by the research method are smaller, indicating that the prediction accuracy of the research method is higher and the error is smaller.

关 键 词：序列理论 用电负荷预测 经验模式分解算法 ARMAX模型 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]