The Statistical Energy Analysis is a widely used tool for the vibro-acoustic analysis of dynamical systems in the high frequency regime. In the applicationof SEA to complex assembled structures, a purely predictive model often exhibits errors, which are mainly due to a lack of accurate modelling of the power transmitted between subsystems and the power dissipated by each of them. The description of the power terms and the SEA equations depend essentially on two key parameters: The coupling loss factor (CLF) and the damping loss factor (DLF). This thesis is included in the framework of Experimental SEA and aims at providing reliable input parameters to an SEA representation of complex systems, as well as procedures that allow a statistical description of these parameters. A new stochastic energy model (SEM) is adopted to enhance the estimation of structural loss factors and assess their sensitivity to subsystems energy variance. For this purpose, expressions are obtained to randomly perturb the energy matrix elements and generate individual samples for the Monte Carlo technique applied to derive the ensemble averaged CLF. The method is firstly applied to an academic example of two rigid coupled plates, then to a more complex aircraft fuselage section. The results indicate that the SEM approach has the potential to significantly enhance the CLF estimate compared to classical matrix inversion methods. Moreover, the new derived sensitivity criterion gives the opportunity to guide the selection of SEA subsystems. The experimental determination of the DLF of built-up aircraft structures and trim panels using the power injection method and the decay rate technique is a complementary topic of this dissertation. Installation effects on the derived DLF are addressed.weiterlesen