Improved GRU prediction of paper pulp press variables using different pre-processing methods

Predictive maintenance strategies are becoming increasingly more important with the increased needs for automation and digitalization within pulp paper manufacturing sector.Hence, this study contributes to examine most efficient pre-processing approaches predicting sensory data trends based on Gated Recurrent Unit (GRU) neural networks. To validate model, from two presses several methods utilized units’ conditions. The results of validation criteria show that using a LOWESS in combination Elimination discrepant filter achieves stable results, prediction error decreases, predicted values easier interpret. model can anticipate future MAPE, RMSE MAE 1.2, 0.27 0.30 respectively. errors below significance level. Moreover, it is identified best hyperparameters found each press must be different.