Indústria 4.0 : inteligência artificial aplicada ao processo de curtimento na indústria do couro.

Abstract

The concepts of Industry 4.0 propose to bring digital transformation to the industry and one of the enabling technologies is artificial intelligence. Thus, the objective of the present work is, using artificial intelligence techniques and historical data, to generate and evaluate Machine Learning models for the prediction of leather yield in tanneries. To this end, the Design Science Research methodology was used. Through a systematic review of the literature, the relationships between the enabling technologies of Industry 4.0 and the leather industry were identified, investigating strategic themes, challenges, and opportunities, with the help of the SciMAT (Science Mapping Analysis Tool) software. The main challenges lie in machine-to machine integration, process traceability, data quality, Internet of Things implementations and simulation. The opportunities identified are related to studies focused on artificial intelligence, in order to detect production patterns, optimize processes, predict failures and simulate the production process. Thus, using the Orange Data Mining software, a leather yield prediction artifact was developed, and different Machine Learning models analyzed. The results show that the Machine Learning model using the AdaBoost algorithm can predict leather yield with the smallest error (Mean Absolute Error - MAE (0.042), Mean Squared Error - MSE (0.003), Root Mean Squared Error - RMSE (0.057) and Determination Coefficient - R2 (0.331)) among the other investigated algorithms kNN, Tree, Random Forest, SVM, Gradient Boosting, Neural Network and Linear Regression. The Random Forest and Gradient Boosting models were shown to be equivalent in terms of predicting leather yield.