A MAINTENANCE SCHEDULING OF SPIRAL PIPE MILL MACHINES BASED ON CORRECTIVE MAINTENANCE DATA USING THE RANDOM FOREST MODEL

Abstract

The Spiral pipe mill machine is a vital asset in the steel pipe manufacturing industry that often experiences downtime due to maintenance strategies that are still reactive or corrective maintenance. This approach leads to operational uncertainty and high repair costs. This study aims to create a maintenance schedule for the spiral pipe mill machine using a random forest machine learning algorithm model. This research analyzes 168 historical machine failure data points from January to December 2024 to calculate machine failure intervals. The model was trained using 500 decision trees with an 80% training data and 20% testing data split. Evaluation results show precise model performance with an R-Squared (R²) value of 0.9916, Mean Absolute Percentage Error (MAPE) of 1.84%, and Mean Absolute Error (MAE) of 0.3993 days. Based on these calculations, an annual maintenance schedule can be compiled to provide accurate maintenance time recommendations to minimize sudden machine failures and increase production efficiency.