基于人工神经网络谐波电流检测算法的电力有源滤波器

Harmonic Current Detection Algorithm for Active Power Filter Based on Artifical Neutral Networks

有源滤波器的有效性在很大程度上依赖于是否能得到真实反映欲补偿的谐波分量的参考信号,因此,有源滤波器设计的关键问题之一就是研究一种好的电流检测算法.在分析并联型有源电力滤波器的工作原理、系统结构的基础上,给出一种新的神经网络的电流检测控制算法,结合此算法设计了三相并联有源电力滤波器(shunt activepower filter,SAF).Matlab仿真结果显示,谐波总畸变率(total harmonic distortion,THD)明显降低,与理论分析一致,证实了该方法的有效性.

Effectiveness of active filter rely on the availability of a reference signal which reflects the harmonic component to be compensated for.As a result,a key issue in the study of active filter is to develop a good current detection algorithm.In view of this,we analyze the working principle and system structure of active power filter,propose a new current detection algorithm using neural networks,and design a parallel three-phase active power filter.In the computer simulation,the total harmonic distortion（THD） is low.The results are in agreement with the theoretical analysis,and therefore confirm validity and feasibility of the new algorithm.