2006短期负荷预测中实时气象因素的

第30卷 第7期 2006年4月 电 网 技 术 Power System Technology Vol. 30 No. 7 Apr. 2006

文章编号：1000-3673（2006）07-0005-06 中图分类号：TM734；F123.9 文献标识码：A 学科代码：790·625

短期负荷预测中实时气象因素的

影响分析及其处理策略

康重庆1，周安石1，王 鹏2，郑广君2，刘 一2

（1．电力系统及发电设备控制和仿真国家重点实验室(清华大学电机系)，北京市 海淀区100084；

2．北京电力公司 北京市 西城区 100031）

Impact Analysis of Hourly Weather Factors in Short-Term Load Forecasting

and Its Processing Strategy

KANG Chong-qing1，ZHOU An-shi1，WANG Peng2，ZHENG Guang-jun2，LIU Yi2

（1．State Key Lab of Control and Simulation of Power Systems and Generation Equipments(Dept. of Electrical Engineering，

Tsinghua University)；2．Beijing Electric Power Corporation，Xicheng District，Beijing 100031，China）

ABSTRACT: Short-term load forecasting is of great significance to secure and economic operation of power grids, so that novel load forecasting models are always studied to improve forecasting accuracy. Because the portion of the load sensitive to meteorology in the total load becomes more and more great, the key problem in the endeavor of improving the load forecasting accuracy is how to consider the influence of meteorological factors on power loads more reasonably. For a long time past due to that the weather service cannot provide real-time weather forecasting results such as temperature, etc., the overwhelming majority in load forecasting models are established on the basis of daily characteristic meteorological factors, e.g., daily maximum temperature, daily minimum temperature and so on. For short-term load forecasting, after examining the action of meteorological factors and analyzing the strategies to process meteorological factors in different stage, the authors propose a new artificial neural network (ANN) based short-term load forecasting model considering hourly weather factors and strive to find out the correlativity and varying pattern of load curve with real-time meteorological factors such as temperature, relative humidity and other weather variables. Actual applications show that by use of the given forecasting model and processing strategy the obtained forecasting results are more accurate, in addition, the proposed forecasting model is also applicable to ultra-short term load forecasting.

KEY WORDS: short-term load forecasting；hourly weather factors；artificial neural network

基金项目：国家自然科学基金资助项目（50377016）。

Project Supported by National Natural Science Foundation of China (50377016)．

摘要：短期负荷预测对于电力系统安全经济运行有着重要的

作用，因此，人们一直致力于研究新的预测模型，提高预测精度。目前，实现提高预测精度这个目标的关键是如何更加合理地考虑气象因素对负荷的影响，因为气象敏感负荷在总负荷中所所占的比重越来越大。长期以来，鉴于气象部门无法提供实时温度等气象预测结果，电力系统所建立的预测模型绝大多数都是基于日特征气象因素，诸如日最高温度、最低温度等。针对短期负荷预测，作者剖析了气象因素的影响和作用，分析了处理不同阶段气象因素的策略，并提出了考虑实时气象因素的短期负荷预测新模型，该模型基于神经网络，力图寻求温度、湿度等实时气象因素与负荷曲线之间的相关关系和变化规律。实际应用表明，文中的预测模型和处理策略可以得到更加精确的预测结果。此短期负荷预测新模型也适用于超短期负荷预测。

关键词：短期负荷预测；实时气象因素；神经网络

0 引言

气象是指大气中所发生的物理现象和物理过程，它主要研究空气压力的垂直变化和水平变化，空气的水平运动和垂直运动，空气温度的变化，辐射能和热能在大气中的传递、吸收过程以及大气中水的相变过程[1]。从电力气候学的角度来看，在进行电力规划、电力工程设计(如发电厂、变电所选址)、电力生产调度等工作时，必须考虑气候因素。

短期负荷预测是电力系统安全经济运行的重要依据，涉及到负荷、电量、负荷曲线等。气象因素对于短期负荷预测有着至关重要的影响[2-4]。其中，常见的影响因素是温度，其次还有湿度、风力、天气类型等。