Digital control three-phase shunt active power filter with a new harmonic-current-extraction process

Current harmonics are a major concern in three-phase diode bridge rectifiers with inductor or capacitor filters, and it is necessary to compensate for the harmonics that exist in the nonlinear load, in order to keep the utility supply current sinusoidal and free from undesired harmonics. Harmonic-current extraction is a prerequisite for its compensation; and in this paper, a signal-processing technique with an adaptive neural network (ADNN) is applied to determine the total harmonic current in a three-phase nonlinear load. The main feature of this proposed technique is that it can detect, not only the fundamental components of the load current, but also the different higher-order harmonic-current components. The proposed harmonic-extraction process also has a faster dynamic response as compared to the conventional Fourier transformation. The technique can detect the current harmonics in realtime, and its effectiveness is substantiated with a DSP-based experimental three-phase shunt active power filter. In the shunt-type active power filter, a three-phase voltage-source inverter, controlled by a deadbeat current controller, is used to generate the compensating current and make the supply current free from harmonic components.