基于缓冲修正的样条灰色模型的长期负荷预测

Mid-term Load Forecasting Based on Buffer Operator and Modified Grey Model

负荷数据是电力系统运行和规划的重要依据,准确地预测出在未来一定时期内的变化情况有利于提高电网运行的经济性和可靠性。我们针对传统GM(1,1)模型存在的一些缺陷,根据未来负荷趋势的判断,利用平均弱化缓冲算子(AWBO)对历史数据进行修正,并提出运用三次样条插值方法对灰色预测模型的背景值进行重构,构建改进的灰色预测模型,克服了传统预测模型的不稳定性,最后给出这种预测方法的建模步骤。通过实例验证,选择我国1980-2008年的年负荷数据进行分析,并选择此方法与传统GM(1,1)模型、支持向量回归(SVR)预测模型和人工神经网络预测模型(ANN)进行比较,结果表明此所提方法是可行和有效的。

Electric load is an important basis for operation and plan in power system.An accurate prediction of electric load will improve economy and reliability of power grid.Shortcomings of the traditional GM(1,1) model were aimed at,and the historical data by average weakening buffer operator(AWBO) was corrected according to prediction of electric load trend.The three times spline interpolation was used to rebuild the background value and form the improved GM(1,1) prediction model.The improved model overcome instability of traditional prediction model.Finally,the modeling procedure of the proposed model was proposed.In the experiment research,annual electric load data from 1980 to 2008 was analyzed.The experiment result was compared with traditional GM(1,1) model,support vector regression(SVR) model and artificial neural networks(ANN) model,and demonstrates the feasibility and validity of the proposed model.