• Refine Query
  • Source
  • Publication year
  • to
  • Language
  • 1
  • Tagged with
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • About
  • The Global ETD Search service is a free service for researchers to find electronic theses and dissertations. This service is provided by the Networked Digital Library of Theses and Dissertations.
    Our metadata is collected from universities around the world. If you manage a university/consortium/country archive and want to be added, details can be found on the NDLTD website.
1

Classifica??o da capacidade produtiva de povoamentos de eucalipto por meio de m?todos tradicionais e redes Kohonen

Silva, Eul?lia Aparecida 18 July 2017 (has links)
Submitted by Jos? Henrique Henrique (jose.neves@ufvjm.edu.br) on 2017-12-18T20:32:46Z No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) eulalia_aparecida_silva.pdf: 2432219 bytes, checksum: e8c925e3e5ce27f2aa2cd5015580bd28 (MD5) / Approved for entry into archive by Rodrigo Martins Cruz (rodrigo.cruz@ufvjm.edu.br) on 2018-01-03T17:15:50Z (GMT) No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) eulalia_aparecida_silva.pdf: 2432219 bytes, checksum: e8c925e3e5ce27f2aa2cd5015580bd28 (MD5) / Made available in DSpace on 2018-01-03T17:15:50Z (GMT). No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) eulalia_aparecida_silva.pdf: 2432219 bytes, checksum: e8c925e3e5ce27f2aa2cd5015580bd28 (MD5) Previous issue date: 2017 / Coordena??o de Aperfei?oamento de Pessoal de N?vel Superior (CAPES) / O objetivo do trabalho foi avaliar a efici?ncia da classifica??o da capacidade produtiva de povoamentos florestais de eucalipto (Eucalyptus ssp.) por meio de rede neural artificial (RNA). Os dados utilizados foram provenientes de invent?rios florestais cont?nuos conduzidos em povoamentos de clones de Eucalyptus ssp. localizados no estado de Minas Gerais. A classifica??o da capacidade produtiva foi realizada por meio de quatro m?todos: curva-guia, predi??o dos par?metros, equa??o das diferen?as e rede neural artificial. Em todos os m?todos foi adotada uma idade de refer?ncia de 72 meses e foram obtidas tr?s classes de capacidade produtiva (superior, m?dia e inferior). Para os m?todos da curva-guia e equa??o das diferen?as foi empregado o modelo de Schumacher linearizado e para o m?todo da predi??o dos par?metros foi utilizado o modelo log?stico. Na classifica??o por meio de RNA utilizou-se a rede auto-organiz?vel de Kohonen, sendo o agrupamento realizado em dois est?gios. Na primeira etapa os dados foram utilizados para treinar a rede e na segunda etapa os vetores de pesos sin?pticos foram agrupados utilizando o m?todo do vizinho mais distante. Foram testadas diferentes entradas (E) para as RNA: E1- volume total com casca (V); E2- ?rea basal (B); E3- altura total (Ht); E4- altura dominante (Hd); E5- di?metro quadr?tico m?dio (q); e E6- V, B, Ht, Hd, q e n?mero de ?rvores por hectare. A sele??o da entrada foi realizada por meio da an?lise discriminante, sendo selecionada a entrada E6 com 83,6% de acerto geral. Os m?todos foram comparados em termos de porcentagem de coincid?ncia na aloca??o dos talh?es, ?rea e volume por classe de capacidade produtiva. As classes obtidas pelos m?todos da curva-guia e equa??o das diferen?as foram muito semelhantes de acordo com os crit?rios de compara??o adotados. A classifica??o pelo m?todo da predi??o dos par?metros n?o foi semelhante aos outros m?todos. A classifica??o por meio de rede neural artificial foi eficiente quando comparada aos demais m?todos em termos de porcentagem de coincid?ncia na aloca??o dos talh?es, ?rea e estoque volum?trico por classe de capacidade produtiva. / Disserta??o (Mestrado) ? Programa de P?s-Gradua??o em Ci?ncia Florestal, Universidade Federal dos Vales do Jequitinhonha e Mucuri, 2017. / The objective of this work was to evaluate the efficiency of the classification of the productive capacity of eucalyptus forest stands (Eucalyptus ssp.) through artificial neural network (ANN). The data used came from continuous forest inventories conducted in stands of Eucalyptus ssp. located in the state of Minas Gerais. The classification of the productive capacity was accomplished through four methods: guide curve, prediction of the parameters, equation of the differences and artificial neural network. In all methods, a reference age of 72 months was adopted and three productive capacity classes (upper, middle and lower) were obtained. For the methods of the guide curve and equation of the differences was used the linearized Schumacher model and for the method of the prediction of the parameters was used the logistic model. In the classification by ANN was used the self-organizing network Kohonen, which the grouping was performed in two stages. In the first step the data were used to train the network and in the second step the vectors of synaptic weights were grouped using the method of the most distant neighbor. Different entries (E) for RNA were tested: E1- total volume with bark (V); E2- basal area (B); E3- total height (Ht); E4- dominant height (Hd); E5- square mean diameter (q); and E6- V, B, Ht, Hd, q and number of trees per hectare. The selection of the input was performed through the discriminant analysis, and the E6 input was selected with 83.6% of the general hit. The methods were compared in terms of coincidence percentage in the allocation of stands, area and volume by class of productive capacity. The classes obtained by the guide curve methods and equation of the differences were very similar according to the adopted criteria of comparison. Classification by the method of parameter prediction was not similar to the other methods. The classification by artificial neural network was efficient when compared to the other methods in terms of coincidence percentage in the allocation of stands, area and volumetric stock by productive capacity class.

Page generated in 0.0616 seconds