Predicting rectal temperature of broiler chickens with artificial neural network

Abstract

Poultry production, facing modernization and increasing competitiveness, shows itself to be enterprising in the adoption of new technologies which enable increased productivity. Knowing that poultry productivity and rectal temperature (Tr ) are affected by environmental conditions, this research was done with the objective of developing and evaluating artificial neural networks (ANNs) for the prediction of Tr in function of thermal conditions (air temperature, Tair; relative humidity, RH; and air velocity, V). The architecture chosen for this purpose was a single hidden layer Multilayer Perceptron (MLP), which was developed and trained under Scilab 4.1.1 aimed with ANN toolbox 0.4.2. The total data available, 139 data points obtained from literature, was divided into two sets, training (94) and validation (45). The selected MLP presented excellent results, providing estimates with an average error of 0.78% for the training set and 1.02% for the validation set. Thus, artificial neural networks constitute an appropriate and promising methodology to solve problems related to poultry production.