基于遗传算法的LS-SVM在谐波源建模中的应用  

Application of least squares support vector machine to harmonic sources modeling based on genetic algorithm

作　　者：郑连清[1] 吴萍 刘小龙 

机构地区：[1]重庆大学输配电装备及系统安全与新技术国家重点实验室,重庆400044 [2]四川省超(特)高压运行检修公司,四川成都610000 [3]中海油深圳分公司,广东深圳518000

出　　处：《电力系统保护与控制》2011年第3期52-56,共5页Power System Protection and Control

摘　　要：对现有的电力系统谐波源建模方法进行分析比较,提出基于最小二乘支持向量机的建模方法,并引用遗传算法对参数寻优以提高模型精度。为了验证方法的有效性,在不考虑和考虑供电端电压谐波含量两种情况下采用Matlab对晶闸管控制电抗器进行仿真提取训练数据并建模。结果表明采用基于遗传算法的最小二乘支持向量机建立的谐波源模型精度高,是谐波源建模的有效方法。该方法把建模对象当作黑箱,没有考虑内部机理,也可用于其他非线性负荷建模之中。On the analysis of present methods of modeling harmonic sources in power system,a modeling method for harmonic sources based on least squares support vector machine（LS-SVM） is presented andgenetic algorithm（GA） is recommended for parameter optimization to improve accuracy of the model .To verify the effectiveness of this method,it uses Matlab to simulate a thyristor controlled reactor for getting the training data and models it by LS-SVM in both cases of considering and without considering the supply voltage harmonic content .The results showthe high precision of this harmonic source model,and LS-SVM based on GA is an effective method.This method models the object as a black box,without considering the internal mechanism,and thereforeit can be used for modeling other non-linear loads.

关 键 词：谐波分析 谐波源建模 遗传算法 最小二乘支持向量机 

分 类 号：TM744[电气工程—电力系统及自动化]