Modelagem dinâmica de cabos com identificação paramétrica e validação experimental
Resumo
A presente dissertação aborda a identificação paramétrica em modelos dinâmicos de
estruturas flexíveis do tipo cabo. A modelagem dinâmica é feita a partir de um formalismo discreto,
no qual se considera a flexibilidade contínua do cabo como aproximada por uma equivalente
discreta, considerando-se elos rígidos conectados por articulações fictícias elásticas. Considera-se o
cabo se movimentando no plano vertical. Foram desenvolvidos algoritmos para gerar o modelo
dinâmico do cabo de forma automática, para um número qualquer de elos rígidos considerados na
aproximação discreta. Considerou-se inicialmente um modelo dinâmico com sessenta elos como
alvo, representando o resultado experimental. Outro modelo com doze elos deveria ter os
parâmetros identificados de forma a que os resultados de simulação tenham erros mínimos com
relação aos resultados com o modelo de sessenta elos. Utilizou-se o método Nelder–Mead Simplex
para a minimização dos erros e a consequente identificação dos parâmetros das matrizes de rigidez
e de atrito uma vez que os demais parâmetros referentes às massas e aos comprimentos dos elos são
facilmente mensuráveis. Os resultados foram considerados bons, de forma que a estrutura proposta
de identificação de parâmetros foi validada, fato que possibilitou a sua aplicação
experimentalmente. Um cabo de fibra sintética e bastante flexível foi fixado ao teto de uma sala,
deixando-se sua outra extremidade livre, na qual foi posta uma carga terminal. Ao longo do cabo
foram demarcadas doze posições igualmente espaçadas, a fim de facilitar a identificação da
configuração espacial do cabo a partir de uma câmera digital. Simulações e experimentos foram
realizados sob as mesmas condições iniciais. Utilizou-se também o método Nelder–Mead Simplex
para a identificação dos parâmetros do modelo. Os resultados de simulações foram muito próximos
dos experimentais, de forma que a técnica de modelagem proposta em trabalhos anteriores a este foi
validada experimentalmente.
The present dissertation addresses the parametric identification in dynamic models of flexible cable- type structures. The dynamic modeling is done from a discrete formalism, in which the continuous flexibility of the cable is considered as approximated by a discrete equivalent, considering the rigid links connected by fictitious elastic joints. The cable is considered moving in the vertical plane. Algorithms were developed to generate the dynamic cable model automatically for any number of rigid links considered in the discrete approach. It was initially considered a dynamic model with sixty links as a target, representing the experimental result. Another model with twelve links should have the parameters identified so that the simulation results have minimal errors with respect to the results with the sixty-link model. The Nelder-Mead Simplex method was used for the minimization of errors and the consequent identification of the parameters of the stiffness and friction matrices since the other parameters concerning the masses and the lengths of the links are easily measurable. The results were considered good, so that the proposed parameter identification structure was validated, which enabled its application experimentally. A very flexible synthetic fiber cable was attached to the ceiling of a room, leaving its other free end, into which a terminal charge was placed. Twelve equally spaced positions were demarcated along the cable in order to facilitate the identification of the spatial configuration of the cable from a digital camera. Simulations and experiments were performed under the same initial conditions. The Nelder-Mead Simplex method was also used to identify the model parameters. The results of simulations were very close to the experimental ones, so that the modeling technique proposed in previous works to this one was validated experimentally.
The present dissertation addresses the parametric identification in dynamic models of flexible cable- type structures. The dynamic modeling is done from a discrete formalism, in which the continuous flexibility of the cable is considered as approximated by a discrete equivalent, considering the rigid links connected by fictitious elastic joints. The cable is considered moving in the vertical plane. Algorithms were developed to generate the dynamic cable model automatically for any number of rigid links considered in the discrete approach. It was initially considered a dynamic model with sixty links as a target, representing the experimental result. Another model with twelve links should have the parameters identified so that the simulation results have minimal errors with respect to the results with the sixty-link model. The Nelder-Mead Simplex method was used for the minimization of errors and the consequent identification of the parameters of the stiffness and friction matrices since the other parameters concerning the masses and the lengths of the links are easily measurable. The results were considered good, so that the proposed parameter identification structure was validated, which enabled its application experimentally. A very flexible synthetic fiber cable was attached to the ceiling of a room, leaving its other free end, into which a terminal charge was placed. Twelve equally spaced positions were demarcated along the cable in order to facilitate the identification of the spatial configuration of the cable from a digital camera. Simulations and experiments were performed under the same initial conditions. The Nelder-Mead Simplex method was also used to identify the model parameters. The results of simulations were very close to the experimental ones, so that the modeling technique proposed in previous works to this one was validated experimentally.
Descrição
Dissertação (mestrado)
Palavras-chave
Modelagem dinâmica, Cabos, Identificação paramétrica, Minimização de Erros, Simulações, Dynamic modeling, Cables, Parametric identification, Minimization of Errors, Simulations
Citação
MENEZES, Guilherme Moraes de. Modelagem dinâmica de cabos com identificação paramétrica e validação experimental. 2019. 95 f. Dissertação (Mestrado) - Programa de Pós-Graduação em Modelagem Computacional, Instituto de Matemática, Estatística e Física, Universidade Federal do Rio Grande, Rio Grande, 2019.
