摘要
针对短期电力负荷预测问题,提出一种基于串行式遗传算法-反向传播神经网络模型的预测方法。首先,由关联分析法确定负荷主要影响因素。然后,确定反向传播神经网络1的输入量为负荷主要影响因素,输出量为相应负荷值,实现多因素回归预测。最后,将反向传播神经网络2串行式融入,并确定其训练数据集为反向传播神经网络1的预测值集,继而实现时间序列预测。两个反向传播神经网络在训练前均采用遗传算法进行初始权值阈值的优化,该方法实现了多因素回归预测与时间序列预测的融合。仿真结果表明,本文所提方法较其他同类型负荷预测方法具有更高的预测精度,可较好地应用于负荷预测工作。
Aimed at the problem of short-term power load forecasting,a forecasting method based on the serial genetic algorithm-back propagation neural network model was proposed.First,the main influence factors of load were deter⁃mined by the correlation analysis method.Then,the input of BP neural network 1 was determined as the main influenc⁃ing factor of load,and its output was the corresponding load value,thus realizing the multi-factor regression prediction.Finally,BP neural network 2 was combined serially,and its training data set was determined as the prediction value set of BP neural network 1,thereby realizing the time series prediction.Before training,both BP neural networks used the GA to optimize the initial weight threshold.This method realized the fusion of multi-factor regression prediction and time series prediction.Simulation results show that the proposed method has higher prediction accuracy than other simi⁃lar load forecasting methods,and it can better facilitate the load forecasting work.
作者
张海
李士心
石军
刘小钰
王坤
孙夏丽
ZHANG Hai;LI Shixin;SHI Jun;LIU Xiaoyu;WANG Kun;SUN Xiali(College of Electronic Engineering,Tianjin University of Technology and Education,Tianjin 300222,China;Zhumadian Electric Power Company,State Grid Henan Electric Power Company,Zhumadian 463099,China)
出处
《电力系统及其自动化学报》
CSCD
北大核心
2021年第4期97-101,107,共6页
Proceedings of the CSU-EPSA
基金
天津市科技特派员资助项目(19JCTPJC54800)
天津市研究生科研创新资助项目(2019YJSS194)。
关键词
短期负荷预测
反向传播神经网络
遗传算法
串行式融合
关联分析
short-term load forecasting
back propagation(BP)neural network
genetic algorithm(GA)
serial fu⁃sion
correlation analysis
