Vibration-based damage identification using hybrid optimization algorithms

• Autor(a) principal: Reynier Hernández Torres
• Data de Publicação: 2017
• Tipo de documento: Tese
• Idioma: eng
• Título da fonte: Biblioteca Digital de Teses e Dissertações do INPE
• Texto Completo: http://urlib.net/sid.inpe.br/mtc-m21b/2017/08.18.16.12

The inverse problem of structural damage identification is addressed in this thesis. The inverse solution is obtained by solving an optimization problem using different hybrid algorithms. The forward structural model is solved by Finite Element codes. FORTRAN code developed by the research group of the Laboratório Associado de Computação e Matemática Aplicada (LAC) of the Instituto Nacional de Pesquisas Espaciais (INPE) was applied to some problems, and for other numerical experiments the NASTRAN software was employed. The acceleration, velocity or displacement time history could be used as experimental data in this methodology. The objective function is formulated as the sum of the squared difference between the measured displacement and the data calculated by the forward model. Different hybrid metaheuristics are tested, using a two-step approach. The first step performs the exploration, and the second one carries out the exploitation, starting from the best solution found in the first step. One optimization approach combines the Multi-Particle Collision Algorithm (MPCA) with the Hooke-Jeeves (HJ) direct search method. MPCA is improved using different mechanisms derived from the Opposition-Based Learning, such as Center-Based Sampling and Rotation-Based Learning. Other applied optimizer is the novel q-gradient, and it is also hybridized with HJ method. The methodology is tested on structures with different complexities. Time-invariant damage was assumed to generate the synthetic measurements. Noiseless and noisy data were considered in tests using models implemented in FORTRAN. Most of the experiments were performed using a full set of data, from all possible nodes, and an experiment was done using a reduced dataset with a low level of noise in data. Noiseless data were considered with experiments using NASTRAN. In this case, the experiments were performed using a full set of data. In general, good estimations for damage location and severity are achieved. Some false positives have appeared, but damages were well identified.