基于MPSR和IRBM的电力系统中长期负荷预测  

Medium and long-term load forecasting of power system based on MPSR and IRBM

作　　者：姜宇 王致杰(指导)[1] 王鸿[1] JIANG Yu;WANG Zhijie;WANG Hong(School of Electrical Engineering,Shanghai Dianji University,Shanghai 201306,China)

机构地区：[1]上海电机学院电气学院,上海201306

出　　处：《上海电机学院学报》2024年第2期83-88,共6页Journal of Shanghai Dianji University

基　　金：上海市自然科学基金资助项目(15ZR1417300)。

摘　　要：针对电力系统中长期负荷波动大及不确定因素导致负荷预测误差较大的问题,提出了一种基于多变量相空间重构(MPSR)和改进受限波尔兹曼机(IRBM)的电力系统中长期负荷预测方法。首先,利用多元线性回归分析方法分析天气因素与电负荷之间的相关性,并将其与电负荷序列组成多变量时间序列;然后,利用C-C法确定每一时间序列的最优嵌入维数和时间延迟,实现多变量相空间重构;最后,采用多变量相空间重构建立的数据集训练电力系统负荷预测模型,同时利用梯度优化法对参数进行优化,得到预测模型。结果表明:相比长短期记忆神经网络和粒子群优化BP神经网络,所提出的预测方法有较高的精准度。To address the problem of large load forecasting errors caused by large medium and long-term load fluctuations and uncertain factors in power systems,a medium and long-term load forecasting method for power systems based on multivariable phase space reconstruction(MPSR)and improved restricted boltzmann machines(IRBM)is proposed.First,the multiple linear regression analysis method is used to analyze the correlation between weather factors and electrical load,and it is combined with the electrical load sequence to form a multivariate time series.Second,the C-C method is used to determine the optimal embedding dimension and time delay of each time series,to achieve multivariable phase space reconstruction.Finally,the data set established by multivariable phase space reconstruction is used to train the power system load forecasting model.At the same time,the gradient optimization method is used to optimize the parameters to obtain the forecasting model.The results show that compared with long short-term memory neural networks and particle swarm optimization BP neural networks,it is confirmed that the proposed forecasting model can significantly improve forecasting accuracy.

关 键 词：负荷预测 多变量相空间重构(MPSR) 改进受限玻尔兹曼机(IRBM) 长短期记忆神经网络 

分 类 号：TM715[电气工程—电力系统及自动化]