Dear All,
Please disseminate and promote the following PhD thesis position:
Title: Reinforcement-learning-based determination of the predominant parameters in model predictive control tuning. Application to the optimization of aqua energy-based system performance.
Keywords: Control system, Artificial intelligence, parameter tuning, performance, optimisation, energy transition.
Context: The PhD thesis project is presented as a scientific contribution to the "Waterwarmth" project (www.interregnorthsea.eu/waterwarmth) financially supported by “INTERREG North Sea Region” program. Involving a consortium of around twenty industrial, academics and public institutions, this European project aims to promote aqua energy as a source of heating. By tackling a number of challenges, the consortium aims to propose solutions that will contribute to the development of hot water production units incorporating surface heat pumps responsible for drawing on the thermal energy present in rivers, lakes and the sea. This energy would then be transmitted to a collective heating system, or even an individual heating system in the case of isolated dwellings. The consortium plans for these units to be supplied with electricity from a local renewable energy source (primarily solar), coupled with an electricity storage unit. To be consistent in its approach, the consortium's ambition is to set up pilot units in France (in Ouistreham, Normandy), Belgium, the Netherlands and Denmark. As a partner in the "Waterwarmth" project, the IRSEEM UR4353 laboratory is tasked with proposing automatic control laws to manage these energy systems.
Objectives: Due to the many advantages of model predictive control (MPC), the first objective of the proposed work will be to contribute to the development of a predictive controller for such energy systems. The second objective will be to show whether it is possible, using reinforcement learning, to determine the parameters of the MPC that will be preponderant with regard to the performance criteria considered, and then to adjust all the parameters of this MPC in order to obtain the expected performance. The third and final objective will be to apply the theoretical results obtained to the control of the hot water production unit developed by one of the partners in the 'Waterwarmth' project.
Applicant profile : A student wishing to become involved in the energy transition, currently preparing a Master 2 degree (or already holding such a degree), with skills in artificial intelligence (preferably in reinforcement learning methods) and a grounding in automatic control (preferably in predictive control). Skills in power electronics will be appreciated.
Start date: October 2024.
Application form: Send a CV, a copy of university diplomas obtained, a copy of the Master 2 pass certificate (or a valid school leaving certificate), a covering letter, L3 and M1 transcripts and the M2 intermediate transcript (if available) to nicolas.langlois@esigelec.fr. Incomplete applications will not be examined.
Best regards.