` MODELO DE PREDIÇÃO DE DADOS BASEADO EM REDES NEURAIS RECORRENTES INTEGRADO COM HISTORIADOR INDUSTRIAL

Abstract

This paper presents the development and evaluation of a data prediction model of generated electrical power based on LSTM-type recurrent neural networks integrated with industrial historian graphical user interface. The data used were obtained directly from the industrial historian PI System using the PI Web API interface. The codes for the prediction models were developed in Python language. The model using LSTM four-layer neural network performed well (R² = 0.718) and the neural network model was integrated into a PI Vision dashboard facilitating the prediction of data on demand by the end user.