应用启发式最小二乘支持向量机的中长期电力负荷预测  

Medium-and Long-Term Load Forecasting Based on Heuristic Least Square Support Vector Machine

作　　者：李如琦[1] 苏浩益[1] 王宗耀[2] 邓国良[1] 陈铁洲[1] 

机构地区：[1]广西大学电气工程学院,广西壮族自治区南宁市530004 [2]南昌工程学院机械与电气工程学院,江西省南昌市330099

出　　处：《电网技术》2011年第11期195-199,共5页Power System Technology

摘　　要：针对中长期负荷预测,提出了基于年负荷总量、年负荷增长量、年负荷增长率、年负荷增长加速率、年国内生产总值等5个指标的启发式最小二乘支持向量机中长期负荷预测模型。首先,通过核函数将低维输入变量空间映射到高维特征空间,建立核偏最小二乘回归模型,拟合出单位国内生产总值电耗;然后以单位国内生产总值电耗为启发式算子,在历史负荷数据的基础上合理假设待预测年的负荷总量,利用启发式算子反推出该负荷值对应的年国内生产总值,形成支持向量机扩展训练样本,将支持向量机外推预测转化为内插求值。最后,用训练好的支持向量机求出预测结果。实际算例的结果表明,所提出的方法预测精度较高,具有较强的可行性和实用性。Founded on＇ total amount of annual load, annual load growth amount, annual load growth rate, acceleration rate of annual load growth and annual gross domestic product （GDP）, a long-term load forecasting model based on heuristic least square support vector machine （LS-SVM） is built. Firstly, the low dimensional input variable space is mapping to high dimensional characteristic space by kernel function and a kernel partial least squares regression model is built to fit out power consumption per unit of GDP; then taking the power consumption per unit of GDP as heuristic operator, on the basis of historical data the total load amount in the forecasted year is reasonably assumed, and then using the heuristic operator the annual GDP corresponding to the assumed total load amount is obtained by inverse method to form expanded training samples of SVM and the extrapolation forecasting of SVM is changed into interpolation evaluation; finally, the forecasting result is solved by well-trained SVM. Calculation results of actual case show that using the proposed method the forecasting results with higher accuracy can be obtained, and the proposed method is practicable.

关 键 词：负荷预测 支持向量机 核偏最小二乘回归 启发式算子 单位国内生产总值电耗 

分 类 号：TM715[电气工程—电力系统及自动化]