Neural network controlled three-phase three-wire shunt active power filter

Three-phase shunt active power filters (Shunt APFs) are widely used in industrial applications to compensate for harmonics, generated by nonlinear loads. In order to succeed the compensation it's necessary to assure current control in the AC side of the APF and voltage or current control in the DC side of the APF. In real practice PI controllers are the most common, as a novel and intelligent control technique neural network controllers are under study and investigation in such application. In this paper, a neural network controller is used to control a three-phase three-wire voltage source shunt APF, its performance in terms of harmonic compensation and speed of response are compared to those of classic PI controller.