面向工业用户的混合DWT-DE-RNN电力负荷预测  

Hybrid DWT-DE-RNN Power Load Forecasting for Industrial Users

作　　者：陆心怡 关艳 高曦莹 王馨璐 LU Xin-yi;GUAN Yan;GAO Xi-ying;WANG Xin-lu(State Grid Liaoning Electric Power Marketing Service Center,Liaoning Shenyang 110819,China)

机构地区：[1]国网辽宁省电力有限公司营销服务中心,辽宁沈阳110819

出　　处：《机械设计与制造》2024年第10期73-78,共6页Machinery Design & Manufacture

基　　金：国网辽宁省电力有限公司管理科技项目资助:(SGLNYX00DFJS2250030)。

摘　　要：电力负荷短期预测对于电力行业规划发展具有重要意义。随着电力市场的改革发展,电力负荷的短期预测对于工业制造型企业有效降低用能成本显得极为重要。然而,实际荷载序列数据体现出多重复杂特性,例如非线性、非平稳性和时间变化等因素影响。这里提出一种由离散小波变换(DWT)、差分进化算法(DE)和径向基函数神经网络(RBFNN)组成的三级混合集成短期负荷预测方法。DWT用于分解负荷数据以获得良好的用电特征;DE用于获得RBFNN预测所需的最佳可调参数。使用PJM公用数据集2001年负荷数据和辽宁省某地工业园区2015年整年数据对这里混合集成方法(DWT-DE-RBFNN)进行了评估。将DWT-DE-RBFNN方法与其他三种主流耦合方法(RBFNN、BPNN、SaDE-ELM)进行了比较。统计分析表明,这里所提方法在MAPE、MAD和RMSE的三种标准尺度上表现出更好的预测精度,体现了该方法的先进性。The short-term prediction of power load is of great significance for the planning and development of the power industry.With the reform and development of the electricity market,short-term forecasting of power load is extremely important for industrial manufacturing enterprises to effectively reduce energy costs.However,the actual load sequence data exhibits multiple complex properties,such as nonlinearity,non-stationarity,and time variation.A three-level hybrid ensemble short-term load forecasting method consisting of discrete wavelet transform(DWT),differential evolution algorithm(DE)and radial basis function neural network(RBFNN)is proposed.DWT is used to decompose load data to obtain good power consumption characteristics;DE is used to obtain the best tunable parameters required for RBFNN prediction.The mixed integration method(DWT-DE-RBFNN)was evaluated using the 2001 load data of PJM public data set and the 2015 annual data of an industrial park in Liaoning Province.The DWT-DE-RBFNN method was compared with three other mainstream coupling methods(RBFNN,BPNN,SaDEELM).Statistical analysis shows that the proposed method shows better prediction accuracy on the three standard scales of MAPE,MAD and RMSE,which shows the advanced nature of the method.

关 键 词：混合短期负荷预测 离散小波变换 差分进化 径向基函数神经网络 

分 类 号：TH16[机械工程—机械制造及自动化] TM715[电气工程—电力系统及自动化]