Harmonic Mitigation in Power Systems Using an ANN-Driven Shunt Active Power Filter: A Simulation-Based Study
Year: 2026
Type: article
Abstract: This paper proposal a design-based artificial neural network (ANN) control approach for a shunt active power filter (SAPF) implemented to a three-phase nonlinear load system. Harmonic variables are extracted from the load current by a Fast Fourier Transform (FFT) and provide to a feed forward ANN that generates three phase compensating current references. The ANN-SAPF is implementation in MATLAB/design under steady-state sudden load fluctuation and asymmetrical load conditions. Using the consistent harmonic result set reported in the results section the source the current total harmonic distortion (THD) is decrease from 7.82% to 2.28%, while the PCC-voltage THD reduce from 3.59% to 1.69%. The dominant 5th and 7th order current harmonics are decreased from 4.85% and 2.35% to 0.92% and 0.66%, respectively. The ANN test the stat regression coefficient is 0.999996 as well as a test MSE of 1.2×10⁻5. The method is presented strictly as simulation validation no experimental processor in the loop hardware in the loop or monitoring real time execution claim is made. IEEE 519 is interpreted according to PCC for voltage and current distortion criteria rather than as a universal 5% current THD limit. The results refer that the integration ANN-driven SAPF provides effective harmonic compensation over the simulated operating conditions.
Institutions: Southern Technical University, Middle Technical University
Language: English
Topic: Power Quality and Harmonics
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7221129147