Use este identificador para citar ou linkar para este item: http://repositorio.ufla.br/jspui/handle/1/58392
Título: Desenvolvimento de sistemas inteligentes para a predição de pragas e doenças no cafeeiro
Título(s) alternativo(s): Development of intelligent systems for the prediction of pests and diseases in coffee tree
Autores: Ferreira, Danton Diego
Silva, Rogério Antônio
Ferreira, Danton Diego
Silva, Rogério Antônio
Ferreira Júnior, Luiz de Gonzaga
Volpato, Margarete Marin Lordelo
Palavras-chave: Predição
Sistemas inteligentes
Doenças e pragas
Cafeeiro
Prediction
Intelligent systems
Diseases and pests
Coffee
Data do documento: 6-Out-2023
Editor: Universidade Federal de Lavras
Citação: ANDRADE, T. Desenvolvimento de sistemas inteligentes para a predição de pragas e doenças no cafeeiro. 2021. 75 p. Dissertação (Mestrado em Engenharia de Sistemas e Automação) - Universidade Federal de Lavras, Lavras, 2021.
Resumo: The pests and diseases in the coffee tree, caused by the coffee miner bug, coffee borer, coffee rust and cercosporiosis, reach up to 50% of a coffee crop, which can cause great damage to coffee growers. Thus, intelligent systems are of paramount importance to predict these damages to the coffee tree and also assist coffee farmers in decision making. This work proposes the use of artificial neural networks of the type Multi-layer Perceptron (MLP) and decision trees for the development of intelligent systems to predict the rate of occurrence of pests and incidence of diseases in the crop. A linear regression method was used in order to compare it with models of intelligent systems by means of statistical metrics. Meteorological data were used, such as: minimum and maximum temperatures, rainfall, relative air humidity, incidence of sunlight and the number of days without rain in the region as input variables for the models. The value of data relating to pests and diseases were collected at the EPAMIG Experimental Camp in São Sebastião do Paraíso, in the south of Minas Gerais. Statistical metrics Root Mean Square Error (RMSE) and the Termination Coefficient (R 2 ) were used to verify how the proposed models are adequately predicting pest and disease manifestations. The MLP neural networks showed the best results for disease and pest models with an RMSE in the range of 0.0220 to 0.1569 and an R 2 that varied between 0.7552 to 0.9803. The values for the decision tree models were a range for RMSE and R 2 between 0.0477 to 0.2900 and 0.2059 to 0.8752, respectively. The results of the models applying multiple linear regression varied between 0.0633 to 0.3154 and 0.1045 to 0.4822 for the metrics RMSE and R 2 , respectively. One of the advantages of using artificial neural networks of the MLP type is the high capacity to learn and generalize after training the algorithm, this was evident in this work. Finally, an application was developed for textit smartphones embedded with an intelligent system model with the aim of predicting and informing the coffeegrower in decision making regarding diseases and pests.
URI: http://repositorio.ufla.br/jspui/handle/1/58392
Aparece nas coleções:Engenharia de Sistemas e automação (Dissertações)

Arquivos associados a este item:
Arquivo Descrição TamanhoFormato 
DISSERTAÇÃO_Desenvolvimento de sistemas inteligentes para a predição de pragas e doenças no cafeeiro.pdf5,89 MBAdobe PDFVisualizar/Abrir


Os itens no repositório estão protegidos por copyright, com todos os direitos reservados, salvo quando é indicado o contrário.