Aplicação de técnicas de inteligência artificial na agricultura de precisão para estimar a produtividade da soja
The precision agriculture is a management technique of agricultural areas that aids in the detection of the factores that causes spacial variability of soil and plant. With the information obtained by the practice of precision agriculture, it is possible to know which points of the tillage need some...
| Autor principal: | Michelon, Gabriela Karoline |
|---|---|
| Formato: | Trabalho de Conclusão de Curso (Graduação) |
| Idioma: | Português |
| Publicado em: |
Universidade Tecnológica Federal do Paraná
2020
|
| Assuntos: | |
| Acesso em linha: |
http://repositorio.utfpr.edu.br/jspui/handle/1/12511 |
| Tags: |
Adicionar Tag
Sem tags, seja o primeiro a adicionar uma tag!
|
| Resumo: |
The precision agriculture is a management technique of agricultural areas that aids in the detection of the factores that causes spacial variability of soil and plant. With the information obtained by the practice of precision agriculture, it is possible to know which points of the tillage need some correction of soil to increase the productivity, what consequently avoids waste of agricultural inputs, leading to decreased costs of production and environmental degradation. The precision agriculture techniques have been favored by the appearance of electronic equipment, softwares and information technologies. Among the many computer technologies, there are the artificial intelligence techniques, which have been widely exploited to assist in agricultural production as techniques that allows, using characteristic data of the plant and/or soil, predict crop yields. Therefore, in this study was chose to apply artificial intelligence techniques in soybean, because it is a very important culture for the human and animal consumption, and as a biofuel alternative in all the world. It is proposed to this study, through the soybean leaf macronutrients and through artificial intelligence techniques to predict the soybean yield, so that the producer is in time to improve the productivity of the crop planted. The artificial intelligence techniques that were used in this study were artificial neural networks and support vector machines for regression. The soybean sheet data were Nitrogen, Phosphorus, Potassium, Calcium and Magnesium collected in three stadiums of plant development at two sites. The results are that was obtained a good predictivity model of artificial neural network, which is able to explain 74% of the real data, using the soybean leaf nutrients of the first data collection in the two sites. It conducted training of sampling areas separately, and in the second collection of nutrients from soybean area A, obtained a second best performance of all the tests with artificial neural networks and support vector machines, explaining 61% of the real data. Selecting CfsSubsetEval attributes was performed to seek a prediction model of soybean yield using fewer nutrients of leaf, and consequently reduce the cost of leaf analysis for the technical application. However, the best model is obtained from the second collection soybean leaf nutrients from area B by means of a support vector machine, explaining 58% of real data. Therefore, this study found good models of soybean yield prediction, using to this, only the leaf nutrients. |
|---|