基于族群进化计算的多项式回归电力负荷预测  

Ethnic Group Evolutionary Computation Based Polynomial Regression Forecast for Electricity Load

作　　者：陈皓[1] 崔杜武[2] 

机构地区：[1]西安邮电学院计算机学院,陕西西安710121 [2]西安理工大学计算机学院,陕西西安710048

出　　处：《软件》2011年第5期34-37,共4页Software

基　　金：国家自然科学基金项目(60873035;61050003)

摘　　要：本文提出利用族群进化算法来有效优化多项式回归分析模型的参数以进行短期电力负荷预测。选择某地区2002年至2009年的用电量为训练数据,将本文提出方法的预测结果与季节指数模型的预测结果进行对比,本文提出方法的拟合值与实际值的平均相对误差较季节指数模型小0.66%。对2010年1月份到10月份的用电量,本文提出方法的预测值与实际值的平均误差仅为1.46%,比季节指数模型小2.3%。此实验结果显示基于族群进化算法优化的多项式回归分析模型不仅是可行的,而且是有效的,它显著提高了对短期负荷预测的准确性和可靠性。In this paper,we propose make use of the ethnic group evolution algorithm （EGEA）,which has a good performance in numerical optimization,to optimize the parameter of a polynomial regression model for forecasting the short-term electricity load.In experiment,the electricity consumption of certain region from 2002 year to 2009 year has been used to be the training data,and the electricity consumption of this region from January to October in 2010 year has been used to be the testing data.Comparing the experimental results with seasonal index model we can find the average relative error of training data decreasing 0.66%,and the average relative error of testing data decreasing 2.3%.The test show making use of EGEA to optimize the parameter of polynomial regression model for forecasting the short-term electricity load is feasible and valid,which improve the accuracy and reliability of result greatly.

关 键 词：多项式回归分析 族群进化算法 参数优化 电力负荷预测 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]