基于免疫进化支持向量机的年用电量预测
The Long Term Prediction of Annual Electricity Consumption Based on IEA-SVM Model
作者：熊建秋(四川大学 水利水电学院, 四川 成都 610065)；邹长武(四川大学 水利水电学院, 四川 成都 610065)；李祚泳(成都信息工程学院, 四川 成都 610041)
Author：(School of Water Resource and Hydropower,Sichuan Univ., Chengdu 610065,China)；(School of Water Resource and Hydropower,Sichuan Univ., Chengdu 610065,China)；(Chengdu Univ. of Info. Technol., Chengdu 610041, China)
收稿日期：2005-07-14 年卷（期）页码：2006,38(2):6-10
期刊名称：工程科学与技术
Journal Name：Advanced Engineering Sciences
关键字：支持向量机；免疫进化算法；参数优化；年用电量预测
Key words：support vector machine；immune evolutionary；parameter optimization；prediction of annual electricity consumption
基金项目：973国家重点基础研究发展规划资助项目(2002CB412301); 国家自然科学基金资助项目(40271024) ; 成都信息工程学院发展基金资助项目(CSRF200401)
中文摘要
支持向量机(SVM)是在统计学习理论基础之上发展起来的，针对小样本数据且具有优良推广性能的机器学习方法。阐述了SVM的基本原理及特性，并采用一种新的有效随机全局优化技术-免疫进化算法(IEA)对SVM核函数的参数进行了优化。介绍了IEA-SVM算法的设计思想和特点，成功地实现了此模型在年用电量预测中的应用，对四川省电网1978～1998年年用电量状况进行了实例研究，预测值与实际值相差较小，并与基于偏最小二乘回归(PLS)模型的预测成果进行了对比。理论分析和实例结果验证了基于IEA-SVM的年用电量预测方法的正确性和有效性。
英文摘要
Support Vector Machine (SVM) as a machine learning method is based on solid theory foundation of Statistical Learning Theory, and focuses on small samples. It has good generalization and has received good applications. The theory and characteristics of SVM are expatiated, and then the application of SVM to a long term prediction annual electricity consumption from 1978 to 1998 of Sichuan province is proposed. Immune evolutionary algorithm (IEA) that is an efficient random global optimization technique is used to optimize the kernel parameter of SVM. The design idea and characteristics of IEA-SVM are introduced. The results show that the accuracy is higher than those based on partial least square (PLS). It proves that the IEA-SVM method is very effective by the theoretical analysis and practical application.
【关闭】
