Modélisation et analyse structurelle du fonctionnement dynamique des systèmes électriques

This thesis was motivated by the need to better understand the connection between the models used in simulation of the power systems dynamics and the phenomena which have to be analyzed or reproduced by the simulation. Indeed, to study and to simulate the behavior of the interconnected power systems, the sophisticated models such as the one of the transmission lines are generally replaced by simple ones. Usually, a dynamic structure of the whole system is not taken into account in a simplification step. As a consequence, experimental validations are generally needed to assess the result of the approximation. For this reason, a structural framework and a systemic viewpoint are proposed to make the solutions more general and more appropriate to the approximation of power systems. First, this allows explaining the link between the distributed parameters model of the transmission lines and the finite dimensional one called π model based on the model reduction. Next, a novel mixed approximation methodology for large-scale dynamic models is proposed which allows one to better rate the dynamics of the system in different situations, and to take into account several practical needs. This methodology is based on a mixture between the modal truncation and the energy of the impulse response so that the dynamical and the physical structures of the system remain unchanged. Moreover, in the context of the automatic control theory, the issue related to a number and a choice of input variables for distributed parameters systems is discussed. To address this issue, an algebraic approach is applied. Here, the main contribution is the detailed study conducted on the basis of the state of the art by which a new way is proposed. Because the issue is not fully solved, more investigations have to be focused on the so called boundary control variables. For practical applications, all the results presented in this study can be exploited to further improve numerical simulations and behavioral studies of large-scale power systems.