System control of hybrid active power filter for reactive power compensation and harmonic suppression

The paper proposes a system control composed of harmonic detection and current control for hybrid active power filter (HAPF). Method of harmonic detection is presented to get the superior harmonic reference signal for improving compensation performance of HAPF system. Detection method is based on self-adaptive noise countervailing principle. An algorithm of normalized LMS (Least Mean Square) with variable step, which is optimized by establishing the corresponding fuzzy rule, is used for adjusting the weights of adaptive filtering algorithm. Meanwhile, recursive integral method combined with neural network (NN) is adopted to compensate current harmonics of power grid. Neural network is used to optimize and obtain optimal control parameters of recursive integral controller for improving control precision of current control. The proposed harmonic detection and current control methods are all implemented on a DSP-based controller. Simulations and experiments have confirmed that HAPF system based on the proposed control method has better effects than conventional PI control.