Artificial Intelligence-Based Smart Grid Optimization for Sustainable Power Distribution Systems

Abstract

Distribution losses, voltage stability issues and difficulties in integrating renewable energy sources are still significant issues in today's power distribution systems. This study proposes a smart grid optimization framework for sustainable power distribution using the IEEE 33-bus radial distribution system (RDS) and artificial intelligence. Power-flow simulation and scenario generation were performed in MATLAB, and AI prediction and PV placement and sizing were done in Python. The 5120 scenarios were created by changing load multiplier, PV bus, PV size, and solar factor, all of which were generated in MATLAB. A Random Forest Regressor was trained to predict power loss, minimum voltage, voltage deviation and feasibility status. The trained model was then coupled with PSO to determine the best PV site and size. The AI-PSO solution has chosen Bus 7 (2000 kW PV) under a load multiplier of 0.70 and solar factor of 1.00. The active power loss obtained from MATLAB load flow is reduced from 94.9113 kW to 51.6514 kW, which is 45.5793% reduction. The minimum voltage improved from 0.940656 pu to 0.971389 pu. The optimized system also saved 378,956.74 kWh/year energy and 265.27 tons/year estimated CO₂ reduction.