Abstract

SIQUEIRA, Luciano Henrique Alves de et al. Forecast of the Total Recoverable Sugar (TRS) Index of Sugarcane Using Artificial Neural Networks. RISTI [online]. 2025, n.57, pp.85-107.  Epub Mar 31, 2025. ISSN 1646-9895.  https://doi.org/10.17013/risti.57.85-107.

Sugarcane harvesting significantly impacts the productivity of sugar mills and should be carried out at the optimal maturity point or Period of Industrial Utilization (PIU). The PIU is estimated through laboratory analyses that measure the Total Recoverable Sugar (TRS) index. The challenge lies in the fact that the planning of sugar mills depends on production forecasts (sugar and ethanol), and there are no tools capable of replacing the high costs of laboratory TRS analyses. Therefore, this article presents a case study aimed at proposing a computational model to estimate the TRS index of the production of a sugar mill in the countryside of São Paulo, Brazil. This model is based Artificial Neural Networks (ANN) and was applied to databases of 48,151 plots from the 2016/2017 to 2022/2023 harvests. The experiments presented a mean absolute error (3.49%) much lower than laboratory analyses (12.17%). The main contribution of this study lies in providing a low-cost model that estimates the ATR index with very high accuracy of the PIU.

Keywords : Sugar Mills; Prediction; Productivity; Artificial Intelligence.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )