Mi SciELO
Servicios Personalizados
Revista
Articulo
Indicadores
- Citado por SciELO
- Accesos
Links relacionados
- Citado por Google
- Similares en SciELO
- Similares en Google
Compartir
Archivos de Zootecnia
versión On-line ISSN 1885-4494versión impresa ISSN 0004-0592
Resumen
BRENNECKE, K. et al. Prediction of protein fractioning of Brachiaria brizantha cv Marandu through Artificial Neural Networks. Arch. zootec. [online]. 2011, vol.60, n.232, pp.1271-1279. ISSN 1885-4494. https://dx.doi.org/10.4321/S0004-05922011000400042.
This paper aims to connect morphogenetic variables of forage and climatic data, with protein fractions (A, B1, B2, B3 and C) through the artificial neural networks known as Multi-Layer Perceptron, with three layers and algorithm training based on back-propagation of error gradient, in order to create a model to predict the protein fractions of fodder from their morphogenetic characteristics and climate. The data used for training and the test were collected in an experiment that was conducted on a 25.2 ha experimental area, located at 47o26'W, 21o59'S, and with pasture composed of Brachiaria brizantha cv Marandu, in a completely randomized block, with four replicates and a forage allowance of 5% (5 kg of dry matter per 100 kg animal.day-1). Each block was divided into four experimental units of 1.575 ha, with five paddocks of 0.315 ha each. Samples were taken two days before the entry of animals, protein was analysed at laboratory to subsequently be compared with the values estimated by the network. Thus, by comparing the output of the network and those obtained by laboratory analysis, it was possible to calculate the average error for fractions A, B1, B2, B3 and C proteins and, thus, can be concluded that the model MLP is able to efficiently predict protein fractions of Brachiaria brizantha.
Palabras clave : Forage.