基于QWDAE和HWMHGRU融合的电力系统短期负荷预测模型  

Short-term Load Forecasting Model for Power System Based on Fusion of QWDAE and HWMHGRU

作　　者：李文升 孙东磊 郑志杰 梁荣 王凇瑶 张智晟 LI Wensheng;SUN Donglei;ZHENG Zhijie;LIANG Rong;WANG Songyao;ZHANG Zhisheng(Economic and Technological Research Institute,State Grid Shandong Electric Power Company,Jinan 250021,China;College of Electrical Engineering,Qingdao University,Qingdao 266071,China)

机构地区：[1]国网山东省电力公司经济技术研究院,济南250021 [2]青岛大学电气工程学院,青岛266071

出　　处：《电力系统及其自动化学报》2023年第9期62-67,共6页Proceedings of the CSU-EPSA

基　　金：国网山东省电力公司科技项目(520625220001)。

摘　　要：为提升电力系统短期负荷预测精度,提出量子加权降噪自编码器和高速通道多层级门控循环单元神经网络融合的短期负荷预测模型。首先利用量子信息处理机制,采用量子加权神经元构建量子加权降噪自编码器,挖掘负荷序列中的有效信息作为输入特征;然后提出具有两级门控结构和高速通道结构的高速通道多层级门控循环单元,构成量子加权降噪自编码器和高速通道多层级门控循环单元融合的短期负荷预测模型。仿真结果表明,所提模型具有较好的预测精度和预测稳定性。To improve the short-term load forecasting accuracy of power system,a short-term load forecasting model based on the fusion of quantum weighted denoising autoencoder(QWDAE)and highway multi-hierarchy gated recur⁃rent unit(HWMHGRU)neural network is put forward.Firstly,by means of the quantum information processing mecha⁃nism,the quantum weighted neuron is used to construct a QWDAE to mine the effective information in the load se⁃quence,which is taken as the input feature.Then,a HWMHGRU with a two-stage gated structure and a highway struc⁃ture is proposed to form the short-term load forecasting model based on the fusion of QWDAE and HWMHGRU.Simula⁃tion results show that the proposed model has a good prediction accuracy and a good prediction stability.

关 键 词：高速通道多层级门控循环单元 量子加权降噪自编码器 短期负荷预测 电力系统 

分 类 号：TM715[电气工程—电力系统及自动化]