The Research On Harmonics Detection Method Of Active Power Filter

Active power filter (APF) is a new style electronic equipment which can compensate harmonics in power system. Harmonic current detection is one of the key technologies of APF. The improvement of harmonic current detection method is very important for improving the performance of APF. This paper, firstly, introduces the basic structure and principle of APF, and then makes comparison among some conventional harmonics detections methods. Based on the idea mentioned above, two harmonics detection methods concerning neural network algorithm and kalman algorithm are discussed in details.A Hopfield neural network can detect accurately the amplitude and phase angle of the fundamental wave as well as the harmonics. The Hopfield neural network is the single-layer, unity feedback and holo-symmetry network, which uses the energy function to optimize object function and then establishes harmonics detection model. It processes the sample data rapidly and need less training in advance.Harmonic current detection based on kalman algorithm is proposed in this paper. This method is an optimal linear estimate based on the least mean-square error criteria, which can real-time estimate the current process state by using the state equations and the recursive method based on the anterior observation data and the latest observation data. The Kalman algorithm includes only plus, multiplication and inversion calculation of matrix, so it is convenient for real-time treatment. Based on the method, the procedures for harmonics detection is investigated and the harmonics detection model is constructed.The simulation is carried out to prove the effectiveness of these two methods. The results show that these methods detect the amplitudes and phases of fundamental and harmonics rapidly and have better real-time characteristic and accuracy than Fourier method.