基于大数据技术的电力物资需求分析系统的设计与实现

发布时间：2018-02-05 01:28

本文关键词： 电力物资预测大数据 Hadoop 数据挖掘任务调度管理　出处：《中国科学院大学(中国科学院沈阳计算技术研究所)》2017年硕士论文　论文类型：学位论文

【摘要】：电力物资是电网建设过程中的基本保障。近年来,我国电力需求持续快速增长,为保证电力供应,各网省公司不断加快电力建设,物资管理的重要性越来越显著。物资需求预测是提升电网运作能力的重要手段,准确的预测结果对合理编制物资供应计划,保障物资供应,降低采购成本提供重要的决策依据。现有的电力物资系统使用IOE架构,具有系统成本高昂、可扩展性差等缺点,无法应对国网数据融合背景下带来的分布式海量数据处理的问题。本文在大数据、数据挖掘等相关技术理论的研究基础上,设计并实现了基于大数据技术的电力物资需求分析系统,旨在对物资需求量进行预测,给决策者提供数据支持。系统包含数据中心子系统和应用子系统,本文重点研究了数据中心子系统。数据中心子系统采用分层的思想设计,根据数据处理的不同环节可分为数据采集层、数据存储层、数据预处理和分析层,还特别设计了任务调度管理层。因为数据分析挖掘的流程基本固定,但是随着数据的更新,需要定期重复这一流程,这一过程是繁重而又枯燥的,借助任务调度管理器,实现了流程的自动化。
[Abstract]:Power supply is the basic guarantee in the process of power grid construction. In recent years, the power demand of our country continues to grow rapidly, in order to ensure the power supply, the power construction of the provincial power companies is constantly accelerated. The importance of material management is becoming more and more obvious. Material demand prediction is an important means to improve the operation ability of power grid. Accurate forecasting results can make the material supply plan reasonably and ensure the material supply. Reducing purchasing costs provides an important basis for decision-making. The existing power materials system uses IOE architecture, which has the disadvantages of high system cost, poor scalability and so on. This paper is based on the research of big data, data mining and other related technology theories. This paper designs and implements a power material demand analysis system based on big data technology, which aims at forecasting material demand and providing data support to decision makers. The system includes data center subsystem and application subsystem. This paper focuses on the data center subsystem. The data center subsystem is designed with the idea of stratification. According to the different links of data processing, it can be divided into data acquisition layer, data storage layer, data preprocessing and analysis layer. Task scheduling management is also specially designed because the process of data analysis mining is basically fixed, but with the update of data, this process needs to be repeated regularly. This process is heavy and boring. With the help of the task scheduling manager, the automation of the process is realized.
【学位授予单位】：中国科学院大学(中国科学院沈阳计算技术研究所)
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP311.13;TP311.52

【参考文献】