基于经验模态分解的短期负荷预测  

Short-Term Load Forecasting in Power System Based on Empirical Mode Decomposition

作　　者：辛鹏[1] 赵阳[1] 王忠义[1] 徐兴峰 

机构地区：[1]吉林供电公司,吉林吉林132012 [2]含山供电公司,安徽含山238100

出　　处：《东北电力大学学报》2008年第4期57-61,共5页Journal of Northeast Electric Power University

摘　　要：提出了一种新的电力系统短期负荷预测混合模型,该模型将经验模态分解(EMD)、支持向量机与BP型神经网络有机结合在一起,充分利用了各方法的特点。利用经验模态分解将负荷序列分解成若干序列,根据各序列的变化特点,在考虑温度影响因素的基础上构建不同的支持向量机模型,然后利用BP网络进行非线性重构得到最终预测结果。仿真结果表明基于该方法的电力系统短期负荷预测具有较高的精度。This paper proposes a new hybrid model for power system short-term load forecasting. In this model, the Empirical Mode Decomposition（ EMD）, Support Vector Machine（SVM） and BP Nature Network are combined organically based on making use of the characteristic of every method. Based on EMD, the load series is decomposed into different lots of calm series, then according to the feature of decomposed components different SVM models are based on considering the influence of climatic factor, and finally using the BP network to reconstruct the forecasted signals of the components, the ultimate forecasting result are obtained. Imitating results show that the proposed forecasting method possesses accuracy.

关 键 词：短期负荷预测 经验模态分解 支持向量机 神经网络 本征模态函数 

分 类 号：C931.2[经济管理—管理学]