一种神经元自适应谐波电流检测方法的数字仿真

基于单个神经元的有源电力滤波器单相电路自适应谐波电流检测方法的性能与一个周期内的采样个数、输入量个数、学习率和惯性系数有关。数字仿真发现:一个周期内的采样个数和学习率对该方法的检测精度和动态响应时间影响较大,它们既不能太大也不能太小,应选择适中;输入量的个数对该方法的检测精度影响较大,而对该方法的动态响应时间影响不大,它越小越好,建议取1;惯性系数对该方法的性能影响很小,建议取0。

Digital simulation of an adaptive harmonic current detection method based on a neuron

The performance of the adaptive harmonic current detection method based on a neuron for single-phase's active power filter is relative to the sample number in a cycle,the number of input,learning rate and inertial coefficient. Digital simulation finds that both the sample number in a cycle and learning rate have biggish influence to its detection precision and dynamic responding time,they should be neither too big nor too small,and they should be mezzo. the number of input has biggish influence to its detec...