Development of an artificial intelligence model for classifying short-duration machine stoppages: A case study of a beverage manufacturing plant

Abstract

In the beverage packaging industry, even short-duration machine interruptions can significantly affect overall production performance (Overall Equipment Effectiveness: OEE). This study presents the development of an artificial intelligence model using a Multilayer Perceptron (MLP) Artificial Neural Network to classify short-duration machine stoppage (Minor Stoppage) events on a can-filling line. The model was evaluated using 5-fold walk-forward time-series cross-validation to ensure robust performance under sequential data conditions. The evaluation results indicate that the proposed ANN model achieved a high overall performance, with a mean macro-F1 score of 92.00% (  = 8.40%) and a mean Mi-F1 score of 84.48% (  = 15.97%). The model was also able to systematically categorize stoppage events, revealing that most cases were classified as short, non-repetitive interruptions (General-Minor Stoppage: GM), accounting for 62.83% of all events. Meanwhile, severe short stoppages (Major-Minor Stoppage: Mi) comprised only 6.67% but were identified as a major source of production efficiency losses. Traditional monitoring methods could not clearly distinguish between routine and severe short-duration stoppages. These findings demonstrate that the developed ANN model has strong potential to serve as an intelligent tool for automatic monitoring and analysis of machine stoppage events. It can detect early signs of abnormality and alert operators to take corrective action before serious failures occur. Consequently, the system enables proactive maintenance planning, reduces unplanned downtime, and enhances OEE in a tangible way, and ultimately improves the long-term efficiency and competitiveness of beverage manufacturing plants.