On the use of virtual sensing for damage identification on wind turbine blades

• Autor(a) principal: Mendes, Andreia Sofia da Silva
• Data de Publicação: 2022
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10362/159587
• topic: Wind turbine blades; Vibration-based Structural Health Monitoring; Damage detection; Virtual sensing; Domínio/Área Científica::Engenharia e Tecnologia::Engenharia Mecânica

With the recent growth in size and complexity of wind turbine blades, continuous monitoring has become imperative in reducing costs and downtime by preventing difficult maintenance and repair works. Using ambient excitation, dynamic responses are analysed with Vibration-based Structural Health Monitoring (VSHM) techniques to obtain modal parameters such as natural frequencies, mode shapes and damping ratios. For that, this study uses two output-only modal identification methods to identify those modal parameters from numerical and experimental responses, aiming to detect changes in those parameters attributed to damage. Using a cantilever beam, damage is introduced as a localized change in mass where, in the presence of damage, natural frequencies may present a decrease in value. Given the limitations of placing physical sensors on wind turbine blades, specially the ones in-service, this study uses measured responses from the structure and estimated ones at unmeasured locations, using data estimation techniques to create virtual sensors. The present objectives are to compare two modal identification methods and two response estimation ones, with the goal of understanding whether the use of virtual sensors improves damage detectability. The numerical comparison between the modal identification methods showed that both are capable of identifying modal parameters close to each other and to the numerical model ones. The results from both methods indicate that using virtual responses increases damage detectability for the same set of measurements. One of the methods provides more consistent results but requires higher computational efforts, whereas the other is much faster and simpler to use, although its results present slight variations. The comparison between the response estimation methods showed that, even with limitations, bothwere capable of identifying close modal parameters to each other. The experimental application validated the conclusions found with the numerical results, from all the methods’ behaviour to the enhancement of damage detectability.