Predição recursiva de diâmetros de clones de eucalipto utilizando rede Perceptron de múltiplas camadas para o cálculo de volume

• Ano de defesa: 2012
• Autor(a) principal: Soares, Fabrízzio Alphonsus Alves de Melo Nunes
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Uberlândia; BR; Programa de Pós-graduação em Engenharia Elétrica; Engenharias; UFU
• Palavras-chave em Português: Redes neurais; Inventário florestal; Clones de eucaliptos; Perceptron multicamadas; Predição de diâmetros; Redes neurais (Computação); Árvores - Inventários; Neural networks; Forest inventory; Eucalyptus clones; Multilayer perceptron; Diameter prediction; CNPQ::ENGENHARIAS::ENGENHARIA ELETRICA
• Link de acesso: https://repositorio.ufu.br/handle/123456789/14300; https://doi.org/10.14393/ufu.te.2012.35

The volume of timber planted is essential information in a rational and sustainable direction of resources available in the forests. Thus it is very important to quantify it as precisely as possible. The Forestry activity that deals with the quantification of wood in the forest is the Forest Inventory. This inventory is the basis for planning the use of forest resources. Through it is possible to characterize a given area and the qualitative and quantitative knowledge of the species that compose it. Forest inventories are used in various types of surveys for recognition, diagnosis and evaluations in the field of forests. The assessment of timber stock in feasibility studies, planning and preparation of plots of exploitation, as well as exploratory post-diagnosis, require specific inventories. Neural networks, especially multilayer perceptron networks with back-propagation algorithm have been used in several areas due to its high capacity to deal with nonlinear relationships of input-output, highlighting the learning ability and the ability to generalization, association and parallel search. These networks have been applied in various areas such as finance, time series forecasting, pattern classification, among others. Neural networks have also been used in several works modeling to estimate forest trees of various parameters such as diameter, height, volume, and others. In this work we performed the analysis and prediction of diameters and calculation of volumes of trees of genus Eucalyptus. The model uses only three diameters measure from the base of the tree, and recursively, the diameters are predicted. For the proposed model experiments were performed to approaches with the overall height and the minimum commercial diameter. The model performance was compared in the experiments and the results showed that the proposed model showed satisfactory performance in relation to the traditional models used in Forestry.