混合有源滤波器的神经网络PI分频控制  

Neural network-PI frequency dividing control of hybrid active power filter

作　　者：张鹏[1] 马建伟[1] 

机构地区：[1]河南科技大学,河南洛阳471000

出　　处：《电源技术》2012年第9期1390-1393,共4页Chinese Journal of Power Sources

摘　　要：针对电源对谐波抑制的要求,提出了基于BP神经网络PI的分频控制方法对谐波进行补偿。该方法通过广义积分PI控制器实现周期量的分频PI控制,不需考虑频率和相位对其的影响,再通过BP神经网络对PI值进行在线调节,从而得到想要的波形。还介绍其详细的实现过程,并与其他的控制方法进行了比较,MATLAB仿真表明,该方法能够实现并联混合有源滤波器的谐波补偿,且在各方有一定优势。For power requirements to harmonic suppression, a control method based on frequency dividing control of BP neural network-PI was proposed to compensate the harmonic. The generalized integral controller was used to achieve divided PI control of the cycle amount with no regard to the effects of frequency and phase, and BP neural network was used to adjust the PI value on-line to get the desired waveform. Also, the detailed implementation process was described, and this method was compared with other control method. The MATLAB simulations show that this method can achieve harmonic compensation of hybrid active power filter, and this method has certain advantages.

关 键 词：谐波抑制 BP神经网络 分频控制 广义积分 

分 类 号：TM531.5[电气工程—电器]