基于多维特征分析的月用电量精准预测研究

发布时间：2018-04-13 16:43

  本文选题：配用电大数据 + 用电量预测　； 参考：《电力系统保护与控制》2017年16期

【摘要】：用户用电量的精准预测是智能配用电大数据应用和发展的关键之一。区别于传统的基于行业分类的预测办法,提出基于大数据挖掘技术的用户用电多维度特征识别,以及在此基础上的精准用电量预测方法。基于海量多用户用电特性,建立多维度用电特征评价指标体系。对用户用电特性空间进行聚类和分析,挖掘和识别用电模式。在不同的用电模式下,分别建立用电量时间序列预测模型,避免用电模式差异对预测算法准确性造成的不利影响。该方法适用于大数据平台的分析与处理,算例分析结果表明其相比以往方法能显著提高预测精度和稳定性。
[Abstract]:The accurate prediction of user's electricity consumption is one of the keys to the application and development of intelligent distribution TV university data.Different from the traditional forecasting method based on industry classification, this paper proposes a multi-dimensional feature recognition method based on big data mining technology, and an accurate power consumption forecasting method based on it.Based on the massive and multi-user power consumption characteristics, a multi-dimensional power consumption evaluation index system is established.Cluster and analyze the user's power characteristic space, and mine and identify the power consumption pattern.In order to avoid the adverse influence of the difference of power consumption mode on the accuracy of prediction algorithm, the forecasting model of time series of electricity consumption is established under different power consumption modes.This method is suitable for the analysis and processing of big data platform. The result of example analysis shows that this method can improve the prediction accuracy and stability significantly compared with the previous method.
【作者单位】： 华中科技大学计算机科学与技术学院;广东海洋大学数学与计算机学院;广东省大数据分析与处理重点实验室;国电江苏电力有限公司;远光软件股份有限公司;
【基金】：广东省重大科技专项(2014B010117006) 广东省大数据分析与处理重点实验室开放基金项目(2017005)
【分类号】：TM715;TP311.13
，

本文编号：1745345