A Study of Machine Learning Techniques for Predicting Equipment Failures in Industrial Maintenance

A significant financial commitment is needed for the upkeep and replacement of highway equipment, and a thorough inspection of every piece of equipment is quite costly. This article aims to trace the development of research and the advancements achieved in using AI/ML approaches for equipment malfunction prediction. To forecast industrial maintenance equipment failures on the AI4I 2020 Predictive Maintenance Dataset, this research compares the accuracy of three ML models: GB, MLP, and KNN. Each model’s performance was assessed using accuracy, precision, recall, and F1-score. With a $90 \%$ accuracy rate, $92 \%$ precision, $90 \%$ recall, and an F1-score of $85.69 \%$, the Gradient Boosting model clearly excels at detecting true positives and striking a good balance between recall and precision. The result shows that the GB model accurately predicts failure prediction. Future work will focus on further enhancing model performance by exploring advanced feature engineering techniques, incorporating additional data sources, and experimenting with other machine learning algorithms to improve precision-recall balance and overall predictive accuracy.