PSO Enhanced and Deep ANN Control for Voltage Regulation and Harmonic Mitigation in Electrical Distribution Networks

Abstract

Modern electrical distribution networks face escalating power quality challenges, including voltage sags/swells and harmonic distortion exceeding IEEE Std 519-2022 limits, driven by renewable integration and non-linear loads. To address these, this study proposed novel particle swarm-enhanced and deep artificial neural network (ANN) controllers for Dynamic Voltage Restorers (DVRs), featuring competitive Particle Swarm Optimisation (PSO) and a 7-layer deep ANN to optimise voltage regulation and harmonic suppression. Validated in MATLAB/Simulink on Nigeria’s Ibadan Distribution Network (IEEE 33-bus system) under multifault scenarios (three-phase sags, sag-induced faults, and combined disturbances), the framework achieved > 99%   voltage stability (restoring voltage to \(\pm\) 1.0 p.u ). It reduced total harmonic distortion (THD) to < 2.5% , outperforming conventional PI controllers (THD >8.5%) and standalone AI methods with 65% faster convergence. The ANN-DVR excelled in complex fault mitigation (THD: 1.78–2.26%), while the PSO-DVR offered computational efficiency (THD: 1.85–2.53%), together providing a robust solution for modern distribution grids requiring stringent power quality compliance.