You are in the accessibility menu

Please use this identifier to cite or link to this item: http://acervodigital.unesp.br/handle/11449/101867
Title: 
Desempenho das redes neurais artificiais na estimativa das variáveis físicas e químicas do solo
Author(s): 
Angelico, João Carlos
Institution: 
Universidade Estadual Paulista (UNESP)
Abstract: 
  • Métodos estatísticos de interpolação são freqüentemente utilizados para se obter as características dos solos em locais não amostrados, visando diminuir o número de amostras necessárias para um bom mapeamento do campo. Nesse trabalho, a estimativa da variabilidade espacial de atributos do solo foi realizada de duas maneiras: primeiramente utilizando-se os métodos estatísticos da krigagem e da co-krigagem e posteriormente as redes neurais artificiais. Os resultados obtidos pelos dois métodos foram comparados, com a finalidade de se verificar a eficiência das redes neurais artificiais na estimativa de atributos do solo. Os resultados mostraram que as redes neurais artificiais, em particular as redes Perceptron, com uma e com duas camadas de neurônios, são capazes de estimar as variabilidades espaciais dos solos com precisão maior do que os métodos estatísticos da krigagem e da co-krigagem. As redes neurais artificiais também se mostraram eficientes na estimativa de uma determinada variável do solo em função de sua classe textural.
  • Statistic methods of interpolation are often used to get the soil characteristics in non-sampled places in order to decrease samples numbers, which are necessary to obtain a good field mapping. In this project, the estimation of soil spatial variability attributes was done in two different ways. First, it was used statistic methods of kriging and cokriging, and in second instance, it was used artificial neural networks. The results computed by both techniques were compared each other in order to verify the efficiency of the artificial neural networks in estimating soil attributes. The results indicated that artificial neural networks, especially Perceptron networks, both with one and two layers of neurons, are able to estimate the soil spatial variability much better than the kriging and the cokriging methods. The artificial neural networks have also showed very efficient in estimating soil variables with respect to its textural class.
Issue Date: 
29-Apr-2005
Citation: 
ANGELICO, João Carlos. Desempenho das redes neurais artificiais na estimativa das variáveis físicas e químicas do solo. 2005. xi, 160 f. Tese (doutorado) - Universidade Estadual Paulista, Faculdade de Ciências Agronômicas, 2005.
Time Duration: 
xi, 160 f. : il. color., grafs., tabs.
Publisher: 
Universidade Estadual Paulista (UNESP)
Keywords: 
  • Mapeamento do solo
  • Redes neurais (Computação)
  • Propriedades do solo
  • Variáveis do solo
  • Neural networks
  • Spatial variabilily
  • Kriging method
  • Artificial intelligence
URI: 
Access Rights: 
Acesso aberto
Type: 
outro
Source:
http://repositorio.unesp.br/handle/11449/101867
Appears in Collections:Artigos, TCCs, Teses e Dissertações da Unesp

There are no files associated with this item.
 

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.