Leveraging Machine Learning for Predictive Maintenance: Insights from a Comparative Study

Author : Djahida Belayadi, Cherifi Lamia, Aissani Manel

Abstract :In recent years, the integration of Machine Learning (ML) within the realm of Predictive Maintenance (PdM) has garnered significant attention in both academic and industrial settings. PdM, which focuses on predicting equipment failures before they occur, presents a strategic approach to minimizing downtime and reducing operational costs. Leveraging ML tech niques, particularly in environments characterized by complex, non-linear, and structured or semi-structured data, enhances the predictive capabilities of maintenance systems. However, the multitude of available ML algorithms, each with unique strengths and weaknesses, complicates the decision-making pro cess regarding their application to specific PdM challenges. This paper seeks to streamline this process by evaluating various ML algorithms (including RF, DT and KNN) and identifying the most effective options for PdM scenarios. By employing a comparative analysis based on established criteria from existing literature. To improve the prediction performance of the models, comprehensive experiments are carried out on four different datasets. we aim to provide insights that will assist practitioners in selecting appropriate ML methodologies to optimize mainte nance strategies and improve overall equipment efficiency.

Keywords :Industry 4.0, Machine Learning, Predictive Maintenance, Algorithm Selection.

Conference Name :International Conference on Computer Science, Technology and Artificial Intelligence (ICCSTAI-25)

Conference Place Bordeaux, France

Conference Date 29th Jan 2025