Research
Brief description of the research topics
Energy systems such as fuel cells, microgrids, and electric vehicles raise fundamental challenges in automatic control. They are inherently distributed, multi-domain, and time-varying, with complex dynamics across multiple temporal and spatial scales. As a result, traditional analytical modelling is often impractical, and many existing control theories—typically built on well-structured models—become difficult to apply.
To address these challenges, my research focuses on data-driven or scientific machine learning and control methods for energy systems. Large volumes of data and knowledge are continuously generated during system operation, creating opportunities to learn system behaviour from historical and real-time data, as well as evolutionary knowledge, often in the form of equations. Beyond modelling, data and knowledge can be directly exploited for controller design, state estimation and prediction, performance evaluation, and real-time optimization.
In this sense, the data-driven and scientific machine learning methodologies in my research cover modelling, diagnosis, prognosis and control. Four specific systems are currently focused on:
- fuel cell systems
- electrolyzer systems (low temperature electrolysis mainly)
- multi-source energy systems
- hybrid electric vehicles.
Key words:
Fuel cell systems: design, modeling, diagnosis, prognosis, and fault tolerance control
Fault diagnosis and prognosis methodologies
Energy storage: modeling and control
Design and control of power electronics
Control of converters interfacing renewable energy and electric grid
Energy management for micro-grids and hybrid electric vehicles
Artificial intelligence and advanced control methods and their applications in energy domain
Equipment and Facilities
The team is affiliated with the FCLAB research facility service centre, enabling my research to benefit from the cutting-edge fuel cell and electrolyser test facilities. The details can be seen on the website of FCLAB https://www.fclab.fr/facilities/.
In addition, the team is also equipped with a complete rapid control prototype and hardware-in-the-loop devices and commonly used devices in electrical engineering, mechanical engineering and thermal engineering fields.
Softwares and models
1- AlphaPEM: A control-oriented fuel cell multiphysics model coded in Python.
Within my ANR JCJC project, we have developed an efficient multiphysics fuel cell model with Python. In the model, we have verified and updated the state-of-the-art equations dedicated to two-phase water transport and gas transport across fuel cell layers, and carefully calibrated the model with data from specifically designed experiments. By running the model, we can see clearly how internal variables, such as water saturation, evolve in various practical scenarios. We can also use the model to recover the electrochemical impedance spectroscopy (EIS).
The model is deliberately tailored for online diagnosis and control purposes. Dr Raphaël GASS, the principal contributor of AlphaPEM, and other followers, are continuing to update the model and the software.
2- Dual-scale system-level fuel cell degradation model.
Within my ANR JCJC project, we have developed a modelling framework involving models at different scales to systematically model the degradation behaviours and their interactions with system operations and performance. Following the PhD work of Dr Walid TOUIL, we are still working on this model framework, aiming to publish the code package soon. This work has been presented in detail in the following paper:
Walid Touil, Zhongliang Li, Rachid Outbib, Daniel Hissel, Samir Jemei, A system-level modeling framework for predicting Pt catalyst degradation in proton exchange membrane fuel cells, Journal of Power Sources, Volume 625, 2025, 235628, ISSN 0378-7753, https://doi.org/10.1016/j.jpowsour.2024.235628.