Metodologia para o desenvolvimento de antenas baseada em intelig??ncia computacional

• Autor(a) principal: Melo, Marcello Caldano
• Data de Publicação: 2021
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações da INATEL
• Texto Completo: https://tede.inatel.br:8080/tede/handle/tede/220
• topic: Aprendizagem de m??quina; Desenvolvimento de antenas; Intelig??ncia Artificial; Intelig??ncia Computacional e Otimiza????o multiobjetivo.; Antennas design; Artificial Intelligence; Computational Intelligence; Machine learning; Multi-objective optimization; Engenharia - Telecomunica????es

The antenna design is a challenging task, which might be time-consuming using computational methods that typically require high computational capability, due to the need for several sweeps and re-running processes. This work proposes an efficient and accurate computational intelligence-based methodology for antenna design and optimization. The computational technical solution consists of a surrogate model application, composed of a Multilayer Perceptron (MLP) artificial neural network with backpropagation for the regression process. Combined with the model, two multiobjective optimization meta-heuristic strategies, Non-dominated Sorting Genetic Algorithm (NSGA-II) and Multiobjective Evolutionary Algorithm based on Decomposition (MOEA/D), are used to overcome the mentioned issues from the antenna design traditional method. As proof of the proposed methodology concept, three studies of case are reported: a half-wavelength dipole; a complete-wavelength dipole; Quasi-Yagi antenna. Comparisons between the models developed with the proposed methodology and results by the full-wave numerical simulations software from ANSYS HFSS are performed aiming to demonstrate the developed methodology application and efficiency. According to the proposed methodology, the matched port printed dipole antenna input-output sets were obtained in 53 seconds by the NSGA-II algorithm to maximize the bandwidth centered at 3.5 GHz, which resulted in 18 % operation band centered at 3.53 GHz. In the matching impedance structure printed dipole, the optimal dimensions were obtained in 5 seconds aiming to maximize the bandwidth centered at 3.5 GHz. The estimated output was 21.4% band centered at 3.5 GHz. Finally, for QuasiYagi design and optimization, the NSGA-II algorithm spent 2 minutes to find the best impedance matching structure dimensions that minimize simultaneously the reflection coefficient at 1.9, 2.6, and 3.5 GHz. The obtained results prove the methodology potential as a complementary alternative to the electromagnetic simulation software for antenna design and optimization.