基于云计算和改进极限学习机的电网负荷预测  

A Power Grid Load Forecasting Model Combining Cloud Computing and Improved Extreme Learning Machine under Massive Data Acquisition

作　　者：冯桂玲 郑鹭洲 蒋宏烨 李思韬 FENG Gui-ling;ZHENG Lu-zhou;JIANG Hong-ye;LI Si-tao(State Grid Fuzhou Electric Power Supply Company, Fuzhou 35009, China)

机构地区：[1]国网福州供电公司,福州35009

出　　处：《科学技术与工程》2021年第22期9411-9417,共7页Science Technology and Engineering

基　　金：国家电网公司总部科技基金(Sgfjfz00yxjs)。

摘　　要：随着智能电网和通信技术的迅速发展,电网系统采集的用户数据规模呈指数增长,传统电网负荷预测方法难以满足海量负荷数据情形下的高效分析和计算需求。据此,依托电力系统数据采集云平台,提出一种基于云计算和改进极限学习机的电网负荷预测模型,采用Map-Reduce网络架构,部署于Hadoop平台,利用分布式计算方式进行电网负荷的精准建模和预测分析。结果表明,相比已有方法,本研究方法具有负荷预测精度高、运行速度快的优势,可为后续智能电网系统建设及管理运用提供一种新颖的解决思路。With the rapid development of smart grid and communication technology,the scale of user data collected by power grid system has been growing exponentially,and conventional electric power load forecasting algorithms are difficult to deal with the requirement of efficient analysis and calculation under massive data acquisition.Therefore,based on cloud acquisition platform,load forecasting model combining cloud computing and improved extreme learning machine(ELM)were proposed.Map-Reduce network architecture and is deployed in Hadoop platform,and the distributed computing technique was utilized to carry out accurate modeling and forecasting analysis of power grid load.Experimental results demonstrate that the proposed method has the advantages of higher accuracy and faster operation speed compared with existing methods,which can provide a novel solution for the construction and management of smart power grid in the future.

关 键 词：智能电网 电力负荷预测 Map-Reduce网络 改进极限学习机 

分 类 号：TM933[电气工程—电力电子与电力传动]