基于大数据技术的多变量短期电力需求预测研究  

Multi-Variable Short-Term Power Demand Forecasting Research Based on Big Data Technology

作　　者：袁小凯 李果 黄世平 YUAN Xiaokai;LI Guo;HUANG Shiping(Southern Power Grid Science Research Institute,Guangzhou 510080,Guangdong,China)

机构地区：[1]南方电网科学研究院,广东广州510080

出　　处：《电网与清洁能源》2020年第12期30-34,40,共6页Power System and Clean Energy

基　　金：广东经研院咨询科技项目(SGNXJY30GHKL1902560)。

摘　　要：针对传统多变量短期电力需求预测方法没有归一化处理电力数据,导致预测性能较差、精度较低,提出基于大数据技术的多变量短期电力需求预测方法。在电网大数据框架中,连接MONGOOSE数据库引擎与短期电子服务器,完成大数据技术支持下的短期电力环境搭建。基于大数据技术,通过确定神经预测网络层数的方式,实现电力需求数据的归一化处理,根据多变量短期预测误差的计算结果,实现基于大数据技术多变量短期电力需求预测方法的应用。实验结果表明,研究方法的电力需求预测有效性更好,预测精度更高。In view of the fact that the traditional multivariable short-term power demand forecasting method does not normalize the power data,which results in poor forecasting performance and low accuracy,a multivariable short-term power demand forecasting method based on big data technology is proposed.In the framework of grid big data,MONGOOSE database engine and short-term electronic server are connected to complete the construction of short-term electric environment supported by big data technology.Based on big data technology,the normalized processing of power demand data is realized by determining the number of neural prediction network layers.According to the calculation result of multivariable short-term prediction error,the application of multivariable short-term power demand prediction method based on big data technology is realized.The experimental results show that the proposed method is more effective and accurate in power demand forecasting.

关 键 词：大数据技术 多变量需求 短期电力预测 数据库引擎 电子服务器 预测神经网络 电力需求数据 预测误差 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]