Sporathe knowledge network
ES
Universidad Nacional de Colombia

Bogotá - Ingeniería - Maestría en Ingeniería - Automatización Industrial · 2024

A reinforcement learning based load frequency control for power systems considering nonlinearities and other control interactions

Considering the advances of civilizations and the limited existence of resources, concerns about the sustainability of that progress are increasing. There is a rising need for affordable, reliable, and sustainable energy sources. Different international organisms recognize the crucial role of energy in achieving global sustainability, highlighting the need for a transition to modern and green energy alternatives. This transition to renewable energy introduces complexities in control systems, requiring reliable load-frequency controllers. This document discusses the limitations of traditional PI control strategies and studies the use of a reinforcement learning-based algorithm, Proximal Policy Optimization (PPO), in power system control. Results demonstrate the efficacy of PPO in effectively reducing area control error, handling nonlinearities, and adapting to disturbances in the system. The adaptability of PPO to different situations and system changes, with its system-agnostic nature, makes it a promising candidate for improving power system control. Finally, practical considerations such as computational requirements, communication delays, measurement noise, and hardware constraints are acknowledged as challenges that need further exploration for the real-world deployment of PPO in power grid environments. The paper concludes by calling for additional research to address these challenges and optimize the algorithm for real-time applications, emphasizing its advantages in the transition to renewable energies.

Full text 80 pages with text

Read the full thesis Record in the repository

Its library's thesaurus