组合预测法的改进及其在负荷预测中的应用  

Improved combination forecast method and its application in short-term load forecasting of a power system

作　　者：安德洪[1] 韩文秀[1] 岳毅宏[1] 

机构地区：[1]天津大学管理学院,天津300072

出　　处：《系统工程与电子技术》2004年第6期842-844,共3页Systems Engineering and Electronics

摘　　要：阐述了传统组合预测方法的基本原理,并从理论上论证了组合预测方法相对于单一预测模型的优越性。在此基础上,总结了组合预测方法的具体步骤,对组合预测方法基于最小二乘法的最优权重确定方法进行了改进,提出了一种新的基于BP神经网络模型的最优权重确定方法。最后将改进后的组合预测方法应用于电力系统短期负荷的预测中。通过分析和比较,验证了该方法的有效性。The principle of conventional combination forecast method is explicated, and the predominance of combination forecast method compared with single forecast model is demonstrated from the theoretical aspect. On this basis, the steps of combination forecast method are explained. The method for determining optimal weights in combination forecast based on least square method is improved, and a novel method based on BP artificial neural networks (ANN) is proposed. In the end, the improved combination forecast method is applied to the short-term load forecasting of a power system, and its availability is verified by means of analysis and comparison.

关 键 词：组合预测 电力系统 负荷预测 BP人工神经网络 

分 类 号：TM715[电气工程—电力系统及自动化]