Nonlinear Loads Compensation Using a Shunt Active Power Filter Controlled by Feedforward Neural Networks

The shunt active power filter (SAPF) is a widely used tool for compensation of disturbances in three-phase electric systems. A high number control methods have been successfully developed, including strategies based on artificial neural networks. However, the typical feedforward network, multilayer perceptron, which has provided effective solutions to many nonlinear problems, not yet employed with satisfactory performance implementation SAPF obtaining reference currents. In order prove capabilities this simple topology, work describes suitable strategy use, accurate estimation Fourier coefficients corresponding fundamental harmonic any distorted voltage or current. An training method consisting use patterns. new generation procedure uses random combinations multiple harmonics, possible nominal frequency deviations occurring real design signals through computations presented. objectives were mitigation and factor correction. Practical cases tested simulation also by using an experimental platform, showing feasibility proposal.