Design and implementation of ANN-based SAPF approach for current harmonics mitigation in industrial power systems

This article describes with Shunt Active Power Filter (SAPF) for compensation of reactive Power (P), current harmonics. An ADALINE based neural network current estimator can control the SAPF to maintain sinusoidal source currents and unity Power Factor (PF). The three-phase (III-Phase) load currents are sensed and measured by Least Mean Square(LMS) algorithm based on ADALINE, the weights obtained with the help of online calculation, and then these weights are multiplied with the help of unit vector, which gives the fundamental frequency of Load Currents (LC). The DC link voltage (V) of the V Source Converter, maintained at a constant value, and it can work as the SAPF, and various simulation results are presented under the steady-state conditions using the PI controller. The switching action of VSC is performed using Hysteresis Current Control based on pulse width modulation, which makes the source currents should follow the reference source currents. The MATLAB based simulation results and implementation results are obtained to determine the efficacy of the SAPF with ADALINE based neural network control for load compensation. SAPF system meets IEEE Standard 519 recommended harmonic standards for different rated nonlinear loads under balanced supply conditions. (C) 2020 Elsevier B.V. All rights reserved.