Current Harmonic Compensation by Active Power Filter Using Neural Network-Based Recognition and Controller

In recent decades, the increasing use of nonlinear loads has caused many problems in terms of power quality. These problems include low power factor, and voltage and current harmonics. The distorted voltage can result in increasing temperature of wires and cables, inappropriate performance of protective devices and disturbance in telecommunication lines. Therefore, it would be essential to install filters to omit or damp these distortions. Conventionally, passive filters were used to maintain harmonics under a sensible level. Nevertheless, this kind of filters has many problems such as large size and resonance issues. In recent years, by improvements in power electronics, passive filters have been replaced with active power filters (APF). Controlling APFs using PI, deadbeat, and predictive controllers have been discussed in different works. However, they all need an accurate model of the system or information about the converters. In this paper, we will provide two control strategies: first, an artificial neural network (ANN)-based control method which mimic conventional control methods; second, ANN-based recognition and control method, which does not require any information about the system model. This control method can be well suiting any system because it can control the whole system only based on the effects on the input on the output of the system. In this paper, ANN-based control methods have been discussed. Then, a control method based on ANN recognition and control will be introduced and developed. The simulation results will be brought, discussed, and compared to show the proficiency of the proposed method over the existent methods.