Data-driven modelling ofAC losses in high-temperature superconducting coils

• Autor(a) principal: Teixeira, Miguel Alexandre Amaral
• 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/10362/155437
• topic: Superconductor; HTS; Power Devices; AC Loss; Artificial Neural Network; Domínio/Área Científica::Engenharia e Tecnologia::Engenharia Eletrotécnica, Eletrónica e Informática

Predicting the loss in superconductive power devices is of utmost importance when designing such devices. This is because the cooling system needs to be designed accordingly. The current methods for predicting AC Loss are either inaccurate or very time consuming. These conventional methods for predicting loss are of two types in which one is faster but inaccurate, while the other is very accurate but also very time consuming. While currently they are both employed in different stages of the design process, there is an interest in a faster, but still accurate, form of predicting AC Loss. Studies have time and time again shown that Artificial Neural Networks are capable of taking on complex tasks and handling them faster than regular computing. Because of this, in this work, an Artificial Neural Network based approach is proposed as to predict AC Loss in various configurations of HTS coils. This approach aims to replicate the accuracy of standard numerical models while being much faster than said models. This results in a final framework comprised of two distinct sequential Neural Networks that are capable of predicting the AC Loss for different configurations of HTS coils nearly instantaneously while still being very accurate and reliable in their predictions.