[en] This thesis deals with the optimization of carbon dioxide consumption by microalgae. Indeed, following several current environmental issues primarily related to large emissions of CO₂, it is shown that microalgae represent a very promising solution for CO₂ mitigation. From this perspective, we are interested in the optimization strategy of CO₂ consumption through the development of a robust control law. The main aim is to ensure optimal operating conditions for a Chlorella vulgaris culture in an instrumented photo-bioreactor. The thesis is based on three major axes. The first one concerns growth modeling of the selected species based on a mathematical model reflecting the influence of light and total inorganic carbon concentration. For the control context, the second axis is related to biomass estimation from the real-time measurement of dissolved carbon dioxide. This step is necessary for the control part due to the lack of affordable real-time sensors for this kind of measurement. Three observers structures have been studied and compared: an extended Kalman filter, an asymptotic observer and an interval observer. The last axis deals with the implementation of a non-linear predictive control law coupled to the estimation strategy for the regulation of the cellular concentration around a value which maximizes the CO₂ consumption. Performance and robustness of this control law have been validated in simulation and experimentally on a laboratory-scale instrumented photo-bioreactor. This thesis represents a preliminary study for the optimization of CO₂ mitigation strategy by microalgae. (author)