Controle híbrido de vibrações em temperaturas distintas : uma abordagem numérica e experimental
Resumo
Resumo: Sistemas mecânicos que apresentam vibrações excessivas requerem controle, uma vez que elas podem comprometer o correto funcionamento dos sistemas e diminuir sua vida útil, além de possivelmente prejudicar a saúde das pessoas envolvidas com eles. Entende-se por controle de vibrações o conjunto de medidas que visam manter a resposta dinâmica de um sistema mecânico abaixo de níveis máximos considerados permitidos. O presente trabalho apresenta um sistema de controle híbrido de vibração, implementado e investigado através de uma abordagem que usa simulações numéricas e ensaios experimentais. Esse sistema híbrido apresenta uma composição que inclui um subsistema ativo-adaptativo e um subsistema passivo. O subsistema ativo-adaptativo abrange sensores, atuadores e uma unidade de controle equipada com filtros digitais, todos organizados em uma arquitetura por antecipação (feedforward). Por outro lado, o subsistema passivo consiste em um neutralizador dinâmico viscoelástico (NDV) de 2 graus de liberdade, com lâminas viscoelásticas não constritas e massas concentradas. O estudo da aplicação desse sistema de controle híbrido é conduzido em uma viga engastada-livre, sujeita a excitações harmônicas e de frequência variável, além de variações de temperatura. Sabe-se que a integração desses dois subsistemas de controle proporciona robustez, flexibilidade e adaptabilidade à ação de controle. As contribuições substanciais deste trabalho englobam: o estabelecimento de uma metodologia de simulação numérica para o controle híbrido de vibrações no domínio do tempo, em condições de temperatura variável; a proposição de um algoritmo adaptativo denominado CVA-FxNLMS (Controller-Vibration-Aware-FxNLMS) para mitigar vibrações ressonantes, o que constitui um significativo avanço em relação ao algoritmo FxLMS (Filtered-x Least Mean Squares) convencional; e a apresentação da metodologia de análise e projeto do neutralizador abordado. A eficácia do sistema de controle híbrido proposto é minuciosamente avaliada na tese, de modo a evidenciar sua relevância. Com base nos resultados obtidos, pode-se concluir, em particular, que o algoritmo CVA-FxNLMS apresenta melhor performance do que o FxNLMS enquanto que, em geral, o caráter benéfico e valioso da associação entre os subsistemas ativo/adaptativo e passivo se destaca. Abstract: Mechanical systems that exhibit excessive vibrations require control, as they can compromise the proper functioning of the systems and reduce their lifespan, as well as it can harm the health of individuals involved with them. Vibration control is understood as a set of measures aimed at maintaining the dynamic response of a mechanical system below maximum levels considered permissible. This work presents a hybrid vibration control system implemented and investigated through an approach that uses numerical simulations and experimental tests. This hybrid system comprises an active-adaptive subsystem and a passive subsystem. The active-adaptive subsystem encompasses sensors, actuators, and a control unit equipped with digital filters, all of these organized in a feedforward architecture. On the other hand, the passive subsystem consists of a 2-degree-of-freedom dynamic viscoelastic neutralizer (DVN) with unconstrained viscoelastic layers and concentrated masses. The application study of this hybrid control system is conducted on a cantilever beam subjected to harmonic and variable frequency excitations, as well as temperature variations. It is known that the integration of these two control subsystems provides robustness, flexibility, and adaptability to control actions. The substantial contributions of this work include: the establishment of a numerical simulation methodology for hybrid vibration control in the time domain under variable temperature conditions; the proposal of an adaptive algorithm called CVA-FxNLMS (Controller-Vibration-Aware-FxNLMS) to mitigate resonant vibrations, which constitutes a significant advancement compared to the conventional FxLMS (Filtered-x Least Mean Squares) algorithm; and the presentation of the analysis and design methodology of the addressed neutralizer. The effectiveness of the proposed hybrid control system is thoroughly evaluated in the thesis to demonstrate its relevance. Based on the results obtained, it is concluded, in particular, that the CVA-FxNLMS algorithm presents superior performance than that of the FxNLMS while, in general, the beneficial and valuable nature of the association between the active/adaptive and passive subsystems stands out.
Collections
- Teses [73]