Predictive Maintenance Berbasis Machine Learning dalam Smart Manufacturing

Abstract

The concept of predictive maintenance represents a significant change in traditional maintenance methods. The use of machine learning in manufacturing machine maintenance has the potential to offer unprecedented opportunities for predicting problems by uncovering hidden patterns in vast data sets. This study aims to examine four machine learning models in classifying maintenance needs in a smart manufacturing environment. Machine learning models such as Logistic Regression, Random Forest, XGBoost, and Multi-layer Perceptrong (MLP) are trained with 5-fold cross-validation. The dataset used is a public dataset from the kaggle website, which consists of 10000 rows and 13 features with the maintenance_required feature as the target feature. The model training results are evaluated using various metrics, such as accuracy, precision, recall, f1-score, and ROC-AUC. The test results show that Random Forest provides the best performance with an accuracy of 98.37%, precision of 99.97%, recall of 91.72%, f1-score of 95.67%, and ROC-AUC of 95.95%. The tree-based ensemble method Random Forest is able to capture patterns in the data better than linear and neural models. This indicates that Random Forest is a reliable model for detecting machine maintenance requirements. Further research can consider increasing dataset capacity, integration with deep learning techniques, examining the perspective of multivariate time-series structures.