Systems control, instrumentation and characterisation

Context and research activities

The COSMIC group has multidisciplinary activities and a recognized experience in controlling and optimizing energy systems. Since several years, the group takes part in national and international research projects together with key players in the energy industry (ENGIE Cofely, EDF R&D, Acciona Energia, GE Oil & Gas…) and local companies (Pyrescom, Selecom …). On the basis of clearly identified needs and scientific issues, COSMIC has considered working on the instrumentation and control of solar power plants (such activity was initiated in early 2013 through the CSPIMP research project during which the group developed a sky imager able to provide HDR images of the sky as well as algorithms dedicated to the infra-hour forecasting of direct normal irradiance, with the aim of improving operation of parabolic trough power plants) and the management of distributed generation, at the microgrid scale. COSMIC's activities also deal with characterizing components and materials, with key applications in the development of new measuring devices, the optimization of multi-junction solar cells (CPV), and in aging of solar receivers. As a result, COSMIC's activities are organized according to three main thematic axis.

Axis 1. Control and optimization of energy production systems, in particular solar systems

COSMIC contributes to the optimization of ENGIE Cofely's (formerly Cofely GDF-Suez) multi-energy district boilers, via the design and optimal control of thermal storage tanks (i.e. hot water tanks). These plants are connected to local heat networks for thermal energy distribution (heating and hot water) (projects OptiEnR 1 & 2). Regarding the instrumentation and control of concentrated solar power plants (Figures 1 to 4), the development of a closed-loop aim-point strategy (for solar towers), allowing the disturbances impacting the solar field to be rejected, is on the way (collaboration with the TRECS group). A sky-imaging system composed of several synchronized fisheye cameras will be used for the spatial forecasting of solar resource (i.e. the system will allow obtaining a cartography of direct normal irradiance). As another short-term perspective, the algorithms developed during the CSPIMP research project will be adapted to the infra-hour forecasting of global horizontal irradiance and the management of energy mixes including photovoltaic power plants.

Axis 2. Microgrid efficiency and distributed generation management

Due to the rarefaction of fossil fuels and, as a result, an increasing distributed power generation, new management and supervision tools, allowing energy efficiency to be promoted, are necessary. At the microgrid scale (i.e. a building, a smart district or an industrial estate) COSMIC researches address load shifting mechanisms, control of appliances, in particular HVAC (Heating, Ventilation and Air-Conditioning) systems (project Batnrj), energy storage, and the way the microgrid and the electricity grid interact. In particular, the group develops "intelligent" management strategies based on forecasts of key parameters and disturbances.

Axis 3. Components and materials characterization

Regarding the characterization of components and materials, COSMIC's activities aim at improving energy production and conversion. Characterizing optic fibers in radiative environment, by measuring the radiation-induced attenuation (Figures 5 to 8), in particular to develop new measuring devices, is a key contribution (ANR project DROID). Characterizing antimonide cells dedicated to solar concentrating PV aims at improving solar cell efficiency (collaboration with the PPCM group). In addition, new algorithms for the thermophysical characterization of materials, dealing with the detection of invisible defects and the aging of solar receptors, are currently being developed (project Raiselife) jointly with the TRECS group. AI tools and inverse methods are used.

Tools and scientific goals

A wide range of skills (the COSMIC group members belong to the sections 61, 62 and 63 of the French National Committee of Universities), in particular in the fields of system identification and control, soft sensors development, and components and materials characterization, is put to good use to meet scientific goals. The considered approaches are based on artificial intelligence, model-based predictive control, and signal and image processing. So, knowledge models, semi-empirical models, as well as black-box models are developed, according to the complexity of the studied systems and available information. Several control structures are used, ranging from easily-implantable approaches to advanced approaches allowing constraint problems to be solved. Due to its efficiency and because it is clearly in line with such problems, model-based predictive control, which requires anticipating the behaviour of the considered system, is prioritized. Embedded systems are also developed, as well as new measuring devices and fault detection and diagnosis tools. Energy resource forecasting, in particular the solar resource, using models based on the concept of time series or image processing algorithms is a key point in the development of management and supervision approaches.

Permanent staff

Stéphane GRIEU, PR2 UPVD, Team leader
Bernard CLAUDET, PR2 UPVD (time-sharing)
Stéphane THIL, MdC UPVD
Hervé DUVAL, Technicien UPVD



PhD Students

Nouha DKHILI (since 2017)
Antoine GALLET (since 2019)
Shab GBEMOU (since 2018)
Youssef KAROUT (since 2019)
Kadar MAHAMOUD (since 2019)
Kadar MAHAMOUD-DJAMA (since 2019)
Romain MANNINI (since 2020)


Julien NOU
Hanany TOLBA