Aplicação de técnicas de Deep Learning para estimar o coeficiente de uniformidade de microaspersores em sistemas de irrigação por aspersão

• Ano de defesa: 2021
• Autor(a) principal: Santos, Everton Schneider dos
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Tecnológica Federal do Paraná; Medianeira; Brasil; Programa de Pós-Graduação em Tecnologias Computacionais para o Agronegócio; UTFPR
• Palavras-chave em Português: Irrigação agrícola; Irrigação por aspersores; Redes neurais (Computação); Irrigation farming; Sprinkler irrigation; Neural networks (Computer science); CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO; Engenharia/Tecnologia/Gestão
• Link de acesso: http://repositorio.utfpr.edu.br/jspui/handle/1/30307

Uncertainties related to the availability and price of water resources caused by population growth and climate changes are some of the challenges faced by irrigated agriculture. The optimization of irrigation processes must be done so that there is a decrease in water usage by agriculture without interfering with food production. The simulation of distribution uniformity, which is one of the main ways to assess the quality of water usage in micro-sprinkler irrigation systems, can help reduce the cost of field tests used to measure this variable. This work used artificial neural networks and Deep Learning techniques to predict the water precipitation and effectively estimate the uniformity coeficiente of micro-sprinklers. Using bayesian search and hyperparameter optimization, the predictive model created was able to estimate Crhstiansen’s uniformity coefficent with an average error of 0,79% on the validation set. The same model achieved a RMSE of 64,2487 and a R2 of 0,9287, showing that the methodology used in this work is capable of simulate micro-sprinklers irrigation processes, used to reduce costs associated with experimentation and support the decision-making process made by irrigation system managers.