基于支持向量机的电流互感器非线性校正方法研究

Research on nonlinear correction method of current transformer based on support vector machine

针对电流互感器的磁滞特性引起的非线性问题,提出了一种支持向量机的非线性回归校正算法,对电流互感器的输出电压波形进行建模,参数优化和线性回归,从而达到非线性校正的目的。通过支持向量机算法对电流互感器的输出模型拟合仿真,与BP神经网络相比,支持向量机算法具有较好的非线性拟合能力,拟合误差小于3. 987e-05。在构造回归模型的基础上采用最小二乘法建立k值参数与输出电压的线性函数关系式,实现了对电流互感器输入电流的准确计算,使电流互感器非线性输出波形幅度平均均方误差降低到0. 434 9,相位误差小于1. 64%,且对失真畸变的波形进行了修正,提高了电流互感器的测量精度。

For the nonlinear problems caused by the hysteresis characteristics of current transformer,this paper proposes a nonlinear regression correction algorithm based on support vector machine to implement the output voltage waveform modeling of current transformer,parameter optimization and linear regression to achieve nonlinear correction.With support vector machine algorithm for fitting and simulating the output of the current transformer model,compared with BP neural network,support vector machine algorithm has better ability in nonlinear fitting,and the fitting error is less than 3.987e-05.On the basis of constructing the regression model,the least square method was used to establish the linear function relation between the K value parameter and the output voltage,which realizes the accurate calculation of the current transformer input current.This method reduces the average squared error of the current transformer nonlinear output waveform to 0.4349,and the phase error is less than 1.64%,which corrects the distortion waveform,and improves the measurement accuracy of the current transformer.