基于遗传算法优化参数的支持向量机短期负荷预测方法

来源期刊:中南大学学报(自然科学版)2009年第1期

论文作者:吴景龙 杨淑霞 刘承水

文章页码:180 - 184

关键词:遗传算法;支持向量机;参数优化;负荷预测

Key words:genetic algorithm; support vector machine (SVM); parameter selection; power load forecasting

摘    要:通过研究参数选择和支持向量机预测能力的影响,建立利用遗传算法优化参数的支持向量机负荷预测系统。通过遗传算法对支持向量机(SVM)预测模型的各项参数进行寻优预处理,找到最优的参数取值,然后,代入支持向量机SVM预测模型中,得基于遗传算法的支持向量机(GA-SVM)模型,利用此模型对短期电力负荷进行预测研究。通过实例验证,选择河北某地区2005-03-02至2007-05-22每天各个时点的数据进行分析,并且选择SVM模型与BP(Back propagation)神经网络进行对比。研究结果表明:用GA-SVM算法得到的均方根相对误差仅为2.25%,比用SVM模型和BP神经网络所得的均方根相对误差比分别低0.58%和1.93%。所提出的测试方法克服了传统参数选择方法存在的缺点(如研究者往往凭经验和有限的实验给定一组参数,而不讨论参数制定的合理性),提高了支持向量机的预测精度。

Abstract: The impact of different parameters of support vector machine (SVM) model on the power load forecasting was studied, and a short-term power load forecasting system was studied with SVM model whose parameters were optimized by genetic algorithms. The parameters for SVM model were pretreated through genetic algorithms to get the optimum parameter values, and these parameter values were used in the SVM model and genetic algorithm-support vector machine (GA-SVM) model was obtained, which was used to make short power load forecasting. Every hour’s load from 2005-03-02 to 2007-05-22 of an area in Heibei province was taken as the sample data to be analysed. The results show that the root-mean-square relative error of GA-SVM model is only 2.25%, which is less than those of SVM and back propagation (BP) model by 0.58% and 1.93%, respectively. The new model overcomes the shortcomings of parameter selection with traditional methods, for example, researchers often give a set of parameters by virtue of experience or a limited experiments, and don’t discuss the rationality, and the new model improves the forecasting accuracy.

基金信息:国家自然科学基金资助项目

有色金属在线官网  |   会议  |   在线投稿  |   购买纸书  |   科技图书馆

中南大学出版社 技术支持 版权声明   电话:0731-88830515 88830516   传真:0731-88710482   Email:administrator@cnnmol.com

互联网出版许可证:(署)网出证(京)字第342号   京ICP备17050991号-6      京公网安备11010802042557号