The design of complex systems, especially of embedded complex micro systems, causes problems such as component integration, power consumption, reliability, time-to-market ....Mechatronics design appears to be particularly suitable for these systems because it integrates closely simulations, experiments, interactions between subsystems and redesign cycles at all levels. The resulting product is more optimized, more efficient and time-to-market is reduced.This thesis led to methods of characterization and parameter identification but also to methods for optimizing active mechatronic systems through numerical model building and different bench types, i.e. digital, physical, and hybrid. The framework is specifically that of an amplified piezoelectric actuator, its control as well as the general constitution of the closed loop of the related mechatronic system. The conclusions are generalizable.In this thesis, different concepts have been successfully introduced: - The "Free Signal". A new control signal of the piezoelectric actuator, based on splines, maximizes the speed of the actuator movement and minimizes the energy consumption. - Two improvements of the particle swarm optimization algorithm. The first one introduces a stopping criteria by measuring the swarm radius; the limit radius is defined by the measurability limit of the parameters to be optimized (“Radius”). The second one adds a swarm ability: it can jump to a better location keeping its geometry. This allows a faster convergence rate (“BSG-Starcraft”). - The experimental optimization. The numerical model being very uncertain, it is directly replaced by the real system in the optimization process. This leads to better results than those obtained using numerical simulation.
