AG
Ali Ghrayeb
info
Please Note
<p>This page displays the records of the person named above and is not linked to a unique person identifier. This record may need to be merged to a profile.</p>
2 records found
1
Journal article
(2026)
-
A.N. Alquennah, T. Zamzam, A. Kouzou, A. Kermansaravi, M. Trabelsi, S. Bayhan, H. Abu-Rub, A. Ghrayeb, H. Vahedi
This paper proposes an innovative model-free deep reinforcement learning-based controller (RL-C) for a grid-connected 5-level packed-U-cell (PUC5) multilevel inverter (MLI). The controller is designed to deliver a high-quality grid current while maintaining the PUC5 floating capacitor voltage at its reference level. In addition, the proposed controller supports both active and reactive power exchanges, adapts to variations in voltage and current references, and remains robust under grid voltage variations. The RL agent learns optimal switching actions through direct interaction with the PUC5 system, eliminating the need for data collection or reliance on existing control models. An Actor-Critic architecture is adopted, and the Proximal Policy Optimization (PPO) algorithm is applied for training (offline) using MATLAB/Simulink, where the RL-C is evaluated under diverse PUC5 configurations and operating conditions in the testing phase. The trained agent has been implemented on an Opal-RT real-time system and validated experimentally using a laboratory-made PUC5 prototype. The performance of the proposed RL-C approach is compared to both traditional approaches including finite control set model predictive control, sliding mode control, and PI control, and other state-of-the-art RL algorithms, demonstrating superior generalization and training efficiency. Moreover, a sensitivity analysis quantifying the impact of reward design, state space, network size, and key hyperparameters on convergence and performance is carried out.
...
This paper proposes an innovative model-free deep reinforcement learning-based controller (RL-C) for a grid-connected 5-level packed-U-cell (PUC5) multilevel inverter (MLI). The controller is designed to deliver a high-quality grid current while maintaining the PUC5 floating capacitor voltage at its reference level. In addition, the proposed controller supports both active and reactive power exchanges, adapts to variations in voltage and current references, and remains robust under grid voltage variations. The RL agent learns optimal switching actions through direct interaction with the PUC5 system, eliminating the need for data collection or reliance on existing control models. An Actor-Critic architecture is adopted, and the Proximal Policy Optimization (PPO) algorithm is applied for training (offline) using MATLAB/Simulink, where the RL-C is evaluated under diverse PUC5 configurations and operating conditions in the testing phase. The trained agent has been implemented on an Opal-RT real-time system and validated experimentally using a laboratory-made PUC5 prototype. The performance of the proposed RL-C approach is compared to both traditional approaches including finite control set model predictive control, sliding mode control, and PI control, and other state-of-the-art RL algorithms, demonstrating superior generalization and training efficiency. Moreover, a sensitivity analysis quantifying the impact of reward design, state space, network size, and key hyperparameters on convergence and performance is carried out.
Conference paper
(2024)
-
Azadeh Kermansaravi, Alamera Nouran Alquennah, Aleksandra Lekić, Mohamed Trabelsi, Ali Ghrayeb, Haitham Abu-Rub, Hani Vahedi
In this paper, a Reinforcement Learning controller (RLC) is designed and implemented on a 5-level Packed U-Cell (PUC5) grid-connected inverter to control the injected current flowing into the electric network.The RL agent is trained using a Proportional-Integral (PI) reward function to optimize its control strategy. Moreover, the voltage balancing of the auxiliary capacitor in PUC5 is separated from the RL controller and integrated into the switching algorithm to reduce the training burden. This modification reduces the observation inputs required for RL training, significantly shorten the training time. Simulation studies conducted in Matlab/Simulink evaluate the performance of the proposed RL controller, demonstrating robust dynamic response and accurate tracking of reference signals across different operational conditions.
...
In this paper, a Reinforcement Learning controller (RLC) is designed and implemented on a 5-level Packed U-Cell (PUC5) grid-connected inverter to control the injected current flowing into the electric network.The RL agent is trained using a Proportional-Integral (PI) reward function to optimize its control strategy. Moreover, the voltage balancing of the auxiliary capacitor in PUC5 is separated from the RL controller and integrated into the switching algorithm to reduce the training burden. This modification reduces the observation inputs required for RL training, significantly shorten the training time. Simulation studies conducted in Matlab/Simulink evaluate the performance of the proposed RL controller, demonstrating robust dynamic response and accurate tracking of reference signals across different operational conditions.