Neuro-fuzzy control system for active filter with load adaptation

The paper proposes a neuro-fuzzy regulator based control system for parallel active filters, which allows for adaptation of the control mode to changes in the amplitude and spectral composition of the load current. The efficiency of this control system in terms of reactive power compensation quality in the network is demonstrated in a simulation. The main advantage of the proposed control system is in the possibility of its implementation without changes in the compensator circuitry and the switch modulation frequency.