Rus, GuillermoPalma, RobertoPérez-Aparicio, José L.2018-03-162018-03-1620120888-3270https://riunet.upv.es/handle/10251/99442[EN] A model-based inverse problem strategy is proposed for damage characterization, starting from the electromechanical response measurement as input data, and incorporating a numerical model that simulates the piezoelectric response. Furthermore, a sensitivity analysis is developed to provide a rational basis to correctly design the excitation/measurement system. The model-based inverse problem is solved by minimizing a cost functional using genetic algorithms. The cost functional or discrepancy is defined as the L-2 norm of the difference between experimental and simulated measurements. The latter are obtained by solving the forward problem, using a novel 2D dynamic piezoelectric finite element. The effects of measurement noise and model uncertainties are studied in detail through a sensitivity analysis, where a sensitivity factor is defined and implemented. The proposed inverse problem strategy reconstructs the defect characteristics with sufficient precision, under realistic levels of noise. © 2011 Elsevier Ltd. All rights reserved.Reconocimiento - No comercial - Sin obra derivada (by-nc-nd)Damage identificationFinite element methodGenetic algorithmsInverse problemPiezoelectricsCost functionalsDamage characterizationElectromechanical responseFinite ElementForward problemInput datasMeasurement NoiseModel uncertaintiesNumerical modelsPiezoelectric responseSensitivity factorsInverse problemsPiezoelectric ceramicsPiezoelectricitySensitivity analysisUncertainty analysisDamage detectionMECANICA DE LOS MEDIOS CONTINUOS Y TEORIA DE ESTRUCTURASArtículo10.1016/j.ymssp.2011.06.023Abierto