摘 要
电力需求预测是电力系统规划和运行的重要环节，是电力企业制定发电、购电计划、
保证电网安全、经济运行的重要依据，因此准确预测未来的电力需求是一项非常重要的课
题。
本文以吉林水电公司电力需求预测软件的实施为例，介绍了基于综合预测模型的数据
挖掘在该公司电力需求预测系统中的应用情况。
首先介绍了电力需求预测的概念、内容及分类以及电力需求预测的研究现状，并对影
响电力需求的各类因素进行了研究分析。通过对吉林水电公司电力需求预测进行现状调查
和需求调研，总结出原系统中的诸多问题和缺陷；提出了新的电力需求预测系统的设计要
求。通过建立了科学有效的电力需求预测综合数据挖掘库，对吉林水电下属各公司的基础
历史数据（如电力营销系统的电量电费数据、负荷系统的历史日负荷数据等）进行采集和
抽取，将采集到的数据进行数据预处理加工，祛除异常数据、冗余数据、噪声数据，进行
了数据集成、转换和规约，使最终的数据满足数据挖掘的要求。最后介绍了新电力需求预
测系统功能设计，并分析对比了传统电力需求预测模型的片面和不足，提出了综合预测的
概念，介绍了基于综合预测模型的设计思路。并利用综合预测模型对该公司的全社会用电
量做实例预测，结果表明该模型的预测精度符合电力需求预测的设计要求，从而验证了该
模型的有效性和可行性，可以辅助公司进行有效电网规划、合理配置电网资源、提高供电
可靠率、提高电力企业投资回报，最终使公司经济效益和社会效益双丰收。
关键词：电力需求，数据预处理，综合预测，全社会用电量
ABSTRACT
Electricity demand forecast is an important part of the power system planning and operation,
the power companies to develop power generation, power purchase plan, an important basis to
ensure grid security, economic operation, therefore accurate prediction of future electricity
demand is a very important issue.
Based on the electricity demand of Jilin hydropower company forecast software
implementation as an example, introduces the application of electricity demand in the company
comprehensive prediction model based on data mining..
First introduced the concept of forecasts of electricity demand and classification, as well as
forecasts of electricity demand Research , research and analysis of the various factors affecting
the demand for electricity, Survey of Jilin Hydro electricity demand forecast and demand
research, summed up the many problems and defects in the original system; the new electricity
demand forecasting system design requirements. Integrated data mining library, on the the Jilin
hydropower subordinate basis of historical data (such as power marketing system of electricity
tariff data, the history of the load system daily load data, etc.) through the establishment of a
scientific and effective electricity demand forecast for the collection and extraction, will be
collected to the data processing data preprocessing, eliminate abnormal data and redundant data,
noise data, the data integration and conversion and the Statute, so that the final data to meet the
requirements of data mining. Finally, the new electricity demand forecasting system functional
design, and analysis and comparison of the one-sided and shortcomings of conventional
electricity demand forecasting model, the concept of comprehensive prediction, the prediction
model based on an integrated design ideas. And use the integrated prediction model society as a
whole of the company's electricity consumption instance prediction, the prediction accuracy of
the model results show that the design requirements, in line with forecasts of electricity demand
in order to verify the effectiveness and feasibility of the model, can help companies effectively
power grid planning, rational allocation of grid resources, improve power supply reliability rate,
to improve power enterprise return on investment, and ultimately the company economic and
social benefits double harvest.
Keywords: demand for electricity, data preprocessing, comprehensive prediction,total electricity
consumption