摘要
电价预测同时受多种因素的影响,是非线性、动态开放的复杂系统,传统方法难以准确地描述这种复杂的特征。文中提出了基于灰自组织特征映射(Grey Self-Organizing Maps,GSOM)和支持向量回归(Support Vector Regression,SVR)的短期电价预测方法。GSOM能综合考虑历史电价、节假日属性、负荷、气象等影响电价的因素,对电价进行聚类。SVR具有全局最优、泛化能力强等显著优点,能对分类后的电价进行比较准确的拟合预测。算例表明基于GSOM和SVR的电价预测方法是有效可行的。
The power price forecasting which is affected by many factors is a nonlinear,dynamic and complicated system and it is difficult to describe such characteristics by traditional methods. The authors present a novel method based on grey self-organizing maps(GSOM)and support vector regression(SVR)for short- term power price forecasting. Considering the influencing factors such as history price data,day type,hour,load,environment weather,etc.,a GSOM network is applied to cluster the input data set into several subsets. Then SVR,the advantages of which include global optima property,more generalized application,high forecasting accuracy and so on,is able to predict the power price exactly. The example shows that the presented method is feasible and effective.
出处
《安徽电力》
2008年第3期69-74,共6页
Anhui Electric Power
