Failure prediction towards predictive maintenance using machine learning

Authors:
Open Access
Publisher website
Google Scholar
Abstract
Machine learning-based predictive maintenance systems have demonstrated their reliability, operational efficiency, and usefulness for informed decision-making. In this work, we evaluate the performance of various supervised machine learning algorithms to predict failures in industrial machinery, using sensor data. Two analysis approaches are applied: one without feature selection and another using the Neighborhood Component Analysis (NCA) method to reduce dimensionality and highlight the most relevant variables for classification. Discriminant Analysis (DA) model achieved the best overall performance, with superior precision, accuracy, and sensitivity metrics in both approaches. Feature selection using NCA improves the efficiency of several models, especially KNN and Linear Classifier, reducing complexity without compromising accuracy. This work demonstrates the potential of machine learning models for implementation in predictive maintenance industrial environments.