Previsão da temperatura interna de um forno de carvão de alvenaria utilizando rede neural artificial do tipo LSTM

Abstract

In this work, the ability of a Long Short-Term Memory (LSTM) Artificial Neural Network was evaluated for predicting the temperature in a charcoal furnace. This type of furnace is widely used in Brazil due to its ease of construction and its good performance, with a yield that can reach up to 30% of charcoal. LSTM is a type of recurrent artificial neural network widely used for time series prediction and has demonstrated good results. Three different Artificial Neural Network models were tested, configured respectively with the window method, batch memory, and stacked LSTM with batch memory. The metric used to evaluate the models was the RMSE (Root Mean Square Error). This metric measures the distance between the data from the original time series and the data predicted by the LSTM. In the end, it was possible to verify that the LSTM configured with the window method was the one that best suited the task of predicting the temperature in a charcoal furnace.