A damage characterization approach is developed adopting a model-update method. A derivative-free global optimization based on the Particle Swarm method was adopted, combining data computed from a FEM model and experimental signals retrieved from accelerometers in order to estimate degradation of the materialÇs properties. The method is able of simultaneously locating, determining the type of damage (changes on the stiffness, mass, damping) and outputting its severity (size and variation in material properties) using a vibrational approach. The procedure is fully automated (using pre/post-processing operations, scripting capabilities, data exchange) and implemented into PRODDIA¬ - Structural Systems Health Management platform. The proposed approach results in a closed loop optimization process driven by an objective function that considers appropriate damage metrics. The damage assessment capability is validated in a sensorised CFRP plate, 1x1 m and 3 mm thick, with two Z-shape stringers. A data acquisition system with low number of sensors (4 accelerometers, 1 temperature sensor) is deployed. Several experimental signals are obtained by simulating damage through the placement of a mass at 25 different locations on the plate. Different damages metrics are also evaluated. The proposed approach is able of characterizing damage in complex composite plates at different locations.