Research On Harmonic Current Detection Method Based On Hopfield Neural Network And High Accuracy Algorithm

Harmonic detecting is one important branch of harmonic issues, which is significant to the management of harmonic. Since the measurement and analysis of harmonic are preconditions to relay protection, identification of fault point and fault type, they are key works to the analysis of power systems. It is the goal of striving for global researchers to detect the transient harmonic fast and accurately. Traditional methods of harmonic detection mainly include: analog filtering method, FFT based method, instantaneous reactive power theory based method, wavelet and neural network based method and so on. However, all methods mentioned above appear disadvantages or limits to some extent which lead to necessary improvements and developments. Especially, the methods could be used in Active Power Filter(APF) receive more attentions paid by academia.Based on necessary discussion on the generation and threat of harmonic, as well as novel methods of harmonic management, this paper proposes a comprehensive method with high accuracy and real time character. Energy function and optimization calculation theories in Hopfield Neural Network(HNN) are used, along with the Fourier Algorithm, achieving the accurate and fast detection of harmonic. Series and parallel processing methods of detection are presented and different features, configurations and corresponding characters are discussed. Accordingly, both integrated HNN harmonic detection theory and system configuration have been proposed based on deduction and analysis. In MATLAB/SIMULINK environment, we can know various characters of this method by series of simulations. Depending on its excellent performances, it could be expected to be utilized in APF systems in future.