论文部分内容阅读
有源电力滤波器(APF,Active Power Filter)是当今重要的谐波治理和无功补偿装置,APF关键的环节是实时准确地检测出谐波电流。针对传统基于瞬时无功理论的ip-iq谐波检测方法实时性差、对系统的补偿能力要求很高等不足,提出了基于BP神经网络与锁相环相结合提取特定次谐波的方案,权值调整采用BFGS拟牛顿算法,该算法可检测出基波、各次谐波分量的幅值和相角,且神经网络的并行运算能力强大,方便应用于三相电路中。改进的特定次谐波检测系统较传统的LC滤波器和并联型APF的组合调节策略响应速度高,系统结构简单,且不易出现高频谐波。通过Matlab仿真实验证明了该算法的可行性和良好的控制效果。
Active Power Filter (APF) is an important harmonic control and reactive power compensation device. The key point of APF is to detect the harmonic current in real time and accurately. According to the traditional ip-iq harmonic detection method based on instantaneous reactive power theory, the real-time performance is bad and the system’s compensation ability is very high. A scheme of extracting specific subharmonic based on BP neural network and PLL is proposed. Adjusted using the BFGS quasi-Newton algorithm, the algorithm can detect the fundamental, the magnitude and phase angle of the various harmonic components, and neural network parallel computing power, easy to apply to the three-phase circuit. The improved specific sub-harmonic detection system has higher response speed than the traditional combination strategy of LC filter and parallel APF. The system structure is simple and the high-frequency harmonics are not easy to appear. The feasibility of the algorithm and the good control effect are proved by Matlab simulation.