神经网络与锁相环相结合的谐波检测方法

Method of Harmonic Detection Based on Neural Network Combined with Phase-Locked Loop

为提高有源电力滤波器APF(active power filter)对谐波检测的实时性和准确性,本文提出了一种人工神经网络ANN(artificial neural network)与锁相环PLL(phase-locked loop)相结合的谐波检测方法。神经网络计算基波幅值,锁相环输出单位幅值的基波电流,两者相乘为电网基波电流,电网电流减去电网基波电流便可得到纯谐波电流。该方法没有传统ip-iq谐波检测方法中的低通滤波成分,从而在实时性及精确性上有很大提高,延迟小、精度高、易于用软件实现的优点使得该方法更适用于有源电力滤波器的谐波检测。

In order to improve the harmonic detection precision and real-time of active power filter（APF）, a har-monic detection method combined artificial neural network（ANN）with the phase-locked loop （PLL） was propsed in the paper. Neural network computing the fundamental amplitude,and PLL output unit amplitude of the current. Multiplying them obtain grid base wave current, and pure harmonic current can be calculated with grid current minus the grid base wave current. Compared with ip-iq traditional harmonic detection phase, the proposed method does not contain low-pass filter,which is a great improve in accuracy and real-time. The advantages of less delay,high precision and easy to use software to realize make the proposed method more suitable for active power filter harmonic detection. MATLAB simulation verifies its effectiveness and superiority.