基于分布式存储与计算平台的用电量预测研究

发布时间：2019-06-29 14:13

【摘要】：随着SG-ERP和智能电网建设的开展和深入,电网业务数据,以几何级增长的速度在增长,数据来源更加复杂和多样。如何充分利用应用这些巨量的多样化数据,对其进行深入分析以便提供大量的高附加值服务迫在眉睫。因此,本文以《国网信通部关于开展2014年大数据应用试点研究工作的通知》为指导,在湖南省电力公司开展大数据工作,基于公司售电量、全社会用电量、各产业用电量、行业用电量等关键指标数据,结合季节变化、自然增长等外部因素,利用大数据相关技术,建立用电预测分析模型,开展未来用电走势分析,提高统计分析的及时性和准确性,为公司运营管理提供决策支撑。本文首先介绍了课题的研究背景、意义,梳理了国内外关于分布式存储与计算和用电量分析预测的现状,然后结合湖南省实际情况,提出了论文需解决的主要问题和组织架构。其次研究学习了分布式存储与计算平台的相关技术,比如Hadoop、HDFS、HBase、Hive、Ganglia、Sqoop,为课题的进一步研究提供了理论基础。接着对分布式存储与计算平台进行设计与实现,包括平台技术架构设计、物理部署以及管理模块设计与实现。之后分析了当前用电量预测问题,并在分布式存储与计算平台的基础上,提出了基于无模型自适应迭代学习控制的年、月用电量预测模型,基于MapReduce的遗传算法优化神经网络的短期用电量预测模型,并进行实验验证,结果表明本文提出的两种预测模型均能够更快、更精确的对未来用电量走势进行预测。最后在分布式存储与计算平台的基础上,进行了用电量预测系统的设计与实现,包括系统功能设计、架构设计、核心界面实现以及系统在湖南省电力公司应用后的效果。
[Abstract]:With the development and deepening of SG-ERP and smart grid construction, the business data of power grid is growing at the rate of geometric growth, and the data sources are more complex and diverse. It is urgent to make full use of these huge amounts of diversified data and analyze them in order to provide a large number of high value-added services. Therefore, under the guidance of the Circular of the Ministry of Information and Communications of the National Network on the Development of big data's Application pilot Research work in 2014, this paper carries out big data work in Hunan Electric Power Company. Based on the key index data such as electricity sales, electricity consumption in the whole society, electricity consumption in each industry, electricity consumption in the industry, combined with seasonal changes, natural growth and other external factors, this paper uses big data related technology to establish a forecast and analysis model of electricity consumption. Carry out the analysis of future power consumption trend, improve the timeliness and accuracy of statistical analysis, and provide decision-making support for the operation and management of the company. This paper first introduces the research background and significance of the subject, combs the present situation of distributed storage and calculation and electricity consumption analysis and prediction at home and abroad, and then puts forward the main problems and organizational structure to be solved according to the actual situation in Hunan Province. Secondly, the related technologies of distributed storage and computing platform are studied, such as Hadoop,HDFS,HBase,Hive,Ganglia,Sqoop, which provides a theoretical basis for the further research of the subject. Then the distributed storage and computing platform is designed and implemented, including the platform technical architecture design, physical deployment and management module design and implementation. Then the current power consumption prediction problem is analyzed, and on the basis of distributed storage and computing platform, the annual and monthly power consumption prediction model based on model-free adaptive iterative learning control and the genetic algorithm based on MapReduce are proposed to optimize the short-term power consumption prediction model of neural network, and the experimental results show that the two prediction models proposed in this paper can predict the future electricity consumption trend faster and more accurately. Finally, on the basis of distributed storage and computing platform, the design and implementation of power consumption prediction system are carried out, including system function design, architecture design, core interface implementation and the effect of the system in Hunan Electric Power Company.
【学位授予单位】：华北电力大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TM73

【参考文献】