基于自适应神经元的短路电流参数提取

DETECTION PARAMETERS OF SHORT-CIRCUIT CURRENTS BASED ON ADAPTIVE NEURONS

真空断路器实现选相分闸的关键是快速提取短路电流参数。传统快速傅立叶(FFT)和最小二乘参数估计方法由于响应速度慢、计算量大难以满足选相分闸的实时要求;该文基于自适应神经元,给出了一种真空断路器同步开断短路电流时快速提取短路电流参数,预测电弧熄灭时电流零点的方法。介绍了自适应神经元估计短路电流参数的基本原理,采用正交滤波器消除衰减直流分量和误差自相关估计自适应改变学习步长,加快神经元的学习收敛速度和减少稳态误差。MATLAB仿真验证了所提方法的快速性和有效性。

One of the most important issues is to track fault current parameters quickly when vacuum circuit breakers synchronously break short circuit current. The conventional Fast Fourier Transform (FFT) and least square parameter estimation algorithms cannot satisfy real-time requirements because of high computational cost. A quick detecting approach based on linear adaptive linear neurons called Adaline for short-circuit current parameters and the moment of arc extinguishing at current zero point is presented. The fundamental of Adaline applied to estimate short circuit current parameters is given. In order to increase the convergence speed and reduce the maladjustment, an orthogonal filter is used to remove decaying DC components in fault currents and a variable step algorithm for updating the weights is obtained from the error covariance. The result of MATLAB simulation confirms the usefulness of the method and its fast convergence.