Reference : UMR7503-DOMMAR-001
Workplace : VANDOEUVRE LES NANCY
Date of publication : Monday, November 21, 2022
Scientific Responsible name : Dominique Martinez
Type of Contract : PhD Student contract / Thesis offer
Contract Period : 36 months
Start date of the thesis : 1 March 2023
Proportion of work : Full time
Remuneration : 2 135,00 € gross monthly
Description of the thesis topic
Classification of dynamic systems by combining AI and automatic control approaches
Classical automatic control uses (implicitly or explicitly) a model of the system to be controlled in order to estimate state variables, build a control law or perform classification and diagnosis from observations gathered from multiple sensors. For a class of systems, under certain assumptions such as observability and/or controllability, we can demonstrate the stability of these approaches. Nevertheless, in many cases, even when the model of the system is known in advance, it is difficult to find analytical solutions based on the model alone. Artificial Intelligence (AI), on the other hand, has experienced considerable success in classification tasks during the last decades, as evidenced by numerous works in the academic and industrial world (e.g. Google Colab). Nevertheless, AI techniques are mainly developed to classify static data and their use for dynamic systems is a relatively unexplored field. The subject of this thesis concerns the classification of time series data gathered from sensors by combining the automatic control approach, based on modeling the underlying dynamical system, with the AI approach, based on learning the classifier from the available data. The major interest in combining the two approaches is to take into account the model of the dynamic system for learning the classifier even when the model is known partially or time varying.
The subject of the thesis is divided in two parts:
-A methodological part whose objective is to develop generic approaches, based on AI and automatic control, for the classification of dynamic systems from sensor data.
-A validation part dedicated to the experimental validation of the theoretical results obtained. For the validation, we will use chemical sensors of two types: biosensors developed in the framework of the French ANR Pherosensor project (https://anr.fr/ProjetIA-20-PCPA-0007) and metal oxide sensors developed by our partner in Qatar.
The PhD thesis is part of a cotutelle doctoral program between the University of Lorraine and the University of Qatar. The PhD student will spend half of the thesis at LORIA (www.loria.fr) in Nancy France and the other half at Hamad Bin Khalifa University (www.hbku.edu.qa) in Doha Qatar.
Constraints and risks
The PhD thesis sould start on March 2023 at the latest
We talk about it on Twitter!