基于最小二乘支持向量机的励磁特性曲线拟合  

Curve fitting of excitation characteristics based on the least squares support vector machine

作　　者：尉军军[1] 全力[1] 彭桂雪[1] 胡海斌[1] 

机构地区：[1]江苏大学电气学院,江苏镇江212013

出　　处：《电力系统保护与控制》2010年第11期15-17,24,共4页Power System Protection and Control

摘　　要：针对传统支持向量机在电流互感器铁心励磁特性曲线拟合时样本数目较大出现的训练速度慢、占用内存大的问题,提出了一种新的基于最小二乘支持向量机算法。该算法将实测数据由径向基函数把非线性逼近问题转化为线性逼近问题,依据最小二乘法的思想,利用Matlab7.0求一个线性方程组的解,得到拟合曲线的近似表达式。实验结果表明,新算法训练速度快,误差小、拟合精度高。Aimed at the problems of slow training speed and large memory consumption,which occurs when the traditional support vector machine chooses a larger number of training samples in the excitation characteristics curve-fitting of the iron core of the current transformer,a new algorithm is put forward based on the least squares support vector machine.The algorthm makes the nonlinear approximation problem transform into a linear approximation problem with the radial basis function.Based on the least square principle and by using Matlab7.0 to solve linear equations,approximate expressions can be gotten.The experiment result shows that this new algorithm can improve the speed of training and the accuracy of fitting and reduce the error.

关 键 词：电流互感器 最小二乘支持向量机 非线性 径向基函数 曲线拟合 

分 类 号：TM452[电气工程—电器]