Harmonic and reactive currents estimation with adaptive neural network and their compensation with DSP control in single-phase shunt active power filter

This paper presents an adaptive neural network (ADNN) method which can estimate harmonic current as well as reactive current in single-phase rectifier-fed inductive and capacitive loads in real time. This estimation technique is utilized in single-phase shunt active power filter to improve the power factor in the utility ac power side and make the supply current free from harmonic. The main feature of this method is that it can determine the fundamental and other harmonic current components that exist in the nonlinear distorted load current by sensing of the load current only. The feasibility of the method is substantiated with both simulation result and experimental one applying digital signal processing controlled hardware implementation.