A Comparative Study on Machine Learning Models for Fault Detection and Classification in Manufacturing Integration of IoT and Artificial Intelligence in Industry 4.0 management

Abstract

The present study presents an examination of fault detection and classification in manufacturing that is possible due to IoT and Artificial Intelligence in the Industry 4.0 structure. The study is based on IoT and  compares three Artificial Intelligence algorithms such as Naive Bayes , Extreme Gradient Boosting , and k-Nearest Neighbors . The results displayed that in ten trials of the experiment achieving the accuracy from 0.82 to 0.89 . The interpretability tool allowed for the understanding of the models’ predictions. The consistent accuracy in both SHAP and LIME results, i.e. 0.80 and 0.89, suggest that both models performed effectively in model explanation. Thus, end users need to combine advanced machine learning algorithms with interpretable explanations to enhance fault detection and classification in manufacturing. This will consequently allow for better production processes and increased innovation in Industry 4.0 manufacturing.