A Reinforcement Learning-Based Proximal Policy Optimization Approach to Solve the Economic Dispatch Problem
(2025) In Engineering Proceedings 97(1).- Abstract
This paper presents a novel approach to economic dispatch (ED) optimization in power systems through the application of Proximal Policy Optimization (PPO), an advanced reinforcement learning algorithm. The economic dispatch problem, a fundamental challenge in power system operations, involves optimizing the generation output of multiple units to minimize operational costs while satisfying load demands and technical constraints. Traditional methods often struggle with the non-linear, non-convex nature of modern ED problems, especially with increasing penetration of renewable energy sources. Our PPO-based methodology transforms the ED problem into a reinforcement learning framework where an agent learns optimal generator scheduling... (More)
This paper presents a novel approach to economic dispatch (ED) optimization in power systems through the application of Proximal Policy Optimization (PPO), an advanced reinforcement learning algorithm. The economic dispatch problem, a fundamental challenge in power system operations, involves optimizing the generation output of multiple units to minimize operational costs while satisfying load demands and technical constraints. Traditional methods often struggle with the non-linear, non-convex nature of modern ED problems, especially with increasing penetration of renewable energy sources. Our PPO-based methodology transforms the ED problem into a reinforcement learning framework where an agent learns optimal generator scheduling policies through continuous interaction with a simulated power system environment. The proposed approach is validated on a 15-generator test system with varying load demands and operational constraints. Experimental results demonstrate that the PPO algorithm achieves superior performance compared to conventional techniques, with cost reductions of up to 7.3% and enhanced convergence stability. The algorithm successfully handles complex constraints including generator limits, ramp rates, and spinning reserve requirements, while maintaining power balance with negligible error margins. Furthermore, the computational efficiency of the PPO approach allows for real-time adjustments to rapidly changing system conditions, making it particularly suitable for modern power grids with high renewable energy penetration.
(Less)
- author
- Rizki, Adil ; Touil, Achraf ; Echchatbi, Abdelwahed ; Oucheikh, Rachid LU and Ahlaqqach, Mustapha
- organization
- publishing date
- 2025
- type
- Contribution to journal
- publication status
- published
- subject
- keywords
- constraint handling, cost minimization, economic dispatch, generator scheduling, multi-generator systems, power system optimization, proximal policy optimization, real-time optimization, reinforcement learning, renewable energy integration
- in
- Engineering Proceedings
- volume
- 97
- issue
- 1
- article number
- 24
- publisher
- MDPI AG
- external identifiers
-
- scopus:105017598754
- ISSN
- 2673-4591
- DOI
- 10.3390/engproc2025097024
- language
- English
- LU publication?
- yes
- id
- 49087540-cc84-4e4c-acca-50dfce20ffe9
- date added to LUP
- 2025-12-08 11:49:06
- date last changed
- 2025-12-08 11:49:30
@article{49087540-cc84-4e4c-acca-50dfce20ffe9,
abstract = {{<p>This paper presents a novel approach to economic dispatch (ED) optimization in power systems through the application of Proximal Policy Optimization (PPO), an advanced reinforcement learning algorithm. The economic dispatch problem, a fundamental challenge in power system operations, involves optimizing the generation output of multiple units to minimize operational costs while satisfying load demands and technical constraints. Traditional methods often struggle with the non-linear, non-convex nature of modern ED problems, especially with increasing penetration of renewable energy sources. Our PPO-based methodology transforms the ED problem into a reinforcement learning framework where an agent learns optimal generator scheduling policies through continuous interaction with a simulated power system environment. The proposed approach is validated on a 15-generator test system with varying load demands and operational constraints. Experimental results demonstrate that the PPO algorithm achieves superior performance compared to conventional techniques, with cost reductions of up to 7.3% and enhanced convergence stability. The algorithm successfully handles complex constraints including generator limits, ramp rates, and spinning reserve requirements, while maintaining power balance with negligible error margins. Furthermore, the computational efficiency of the PPO approach allows for real-time adjustments to rapidly changing system conditions, making it particularly suitable for modern power grids with high renewable energy penetration.</p>}},
author = {{Rizki, Adil and Touil, Achraf and Echchatbi, Abdelwahed and Oucheikh, Rachid and Ahlaqqach, Mustapha}},
issn = {{2673-4591}},
keywords = {{constraint handling; cost minimization; economic dispatch; generator scheduling; multi-generator systems; power system optimization; proximal policy optimization; real-time optimization; reinforcement learning; renewable energy integration}},
language = {{eng}},
number = {{1}},
publisher = {{MDPI AG}},
series = {{Engineering Proceedings}},
title = {{A Reinforcement Learning-Based Proximal Policy Optimization Approach to Solve the Economic Dispatch Problem}},
url = {{http://dx.doi.org/10.3390/engproc2025097024}},
doi = {{10.3390/engproc2025097024}},
volume = {{97}},
year = {{2025}},
}