Gestão da produção de frangos de corte por meio de redes neurais artificiais

Abstract

Currently, the globalized and highly competitive environment imposes on broilerproduction companies two major challenges: optimizing resources and reducing costs. In this context, this work aims to use artificial neural networks and multiple linear regression for the analysis and forecast of productive variables of broilers in a Paraná agribusiness. It is also analyzed the applicability of recurrent neural networks in the prediction of the price of thekilo of frozen and chilled broiler. The database provided by the company has a 2-year history of movement, containing the main production variables, for 4650 batches of broilersof the Coob, Coob Fast and Coob Slow lines. In the analysis of the applicability of recurrent neural networks, in this work, two databases were used, with monthly prices, between January 2008 and December 2019, representing 132 observations. The results obtained show that the forecasting models provide reliable estimates for the response variables: Average Weight and Productive Efficiency Index and demonstrate the effectiveness of the forecasts, from the recurring LSTM network, for the price of the kilo offrozen and chilled broiler, for a short-term horizon.