Automatic Generation Control of a two-area power system using Supervised Machine Learning

Automatic Generation Control (AGC) plays a pivotal role in maintaining the delicate balance between electricity consumption and production. Our study delves into the intricacies of this system. To ensure its real-world applicability, we construct a simulation of a regional electricity grid, validated through generator-tripping experiments that closely mirror actual grid scenarios. Employing a well-established AGC control model, we scrutinize a simulation-based approach applied to a two-area thermally linked power system. Here, we propose intelligent techniques like Particle Swarm Optimization (PSO), Genetic Algorithm (GA), and Supervised Machine Learning (SML) to optimize PID controllers, with the primary objective being the swift restoration of the grid's frequency to its desired standard while minimizing undesirable undershoots and overshoots during load fluctuations. Our comparative analysis of these techniques, focusing on settling time and overshoots, highlights the superiority of PSO and GA, showcasing their effectiveness in AGC applications. This research underscores their potential for enhancing grid stability and efficiency, shedding light on areas for further development in the realm of Supervised Machine Learning.