Predictive Maintenance for Industrial Equipment: Using XGBoost and Local Outlier Factor with Explainable AI for analysis

In industrial operations, the need to minimize downtime and enhance productivity has produced the need for predictive maintenance techniques. Using artificial intelligence (AI) in this domain has revolutionized maintenance practices, but the lack of transparency of many AI models has obstructed thei...

Full description

Saved in:

Bibliographic Details
Published in:	2024 14th International Conference on Cloud Computing, Data Science & Engineering (Confluence) pp. 25 - 30
Main Authors:	Ghadekar, Premanand, Manakshe, Aman, Madhikar, Sarthak, Patil, Sushrut, Mukadam, Mehvish, Gambhir, Tejas
Format:	Conference Proceeding
Language:	English
Published:	IEEE 18-01-2024
Subjects:	Explainable AI Industrial materials Local Outlier Factor Merging Prediction algorithms Predictive maintenance Predictive models Productivity Reliability Task analysis XGBoost
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!

Description
Summary:	In industrial operations, the need to minimize downtime and enhance productivity has produced the need for predictive maintenance techniques. Using artificial intelligence (AI) in this domain has revolutionized maintenance practices, but the lack of transparency of many AI models has obstructed their widespread acceptance and trustworthiness. This research paper explores the application of XGBoost and Local Outlier Factor algorithms in the domain of Predictive Industrial Maintenance for industrial materials. Using these advanced machine learning techniques, this research aims to predict equipment failures and improve the overall reliability of industrial processes. In addition, this research employs Explainable AI methods to provide clear and interpretable insights into the predictive models' decision-making processes. By combining the power of XGBoost and Local Outlier Factor with explainability. In this study, for predictive classification, the XGBoost gave an F1 score of 96% and for early prediction Local Outlier Factor gave an F1 score of 94% this research also explained the impact of features on Output class using the SHAP model. This research not only enhances predictive accuracy but also ensures transparency, enabling users to make informed decisions for timely maintenance and system optimization.
ISSN:	2766-421X
DOI:	10.1109/Confluence60223.2024.10463280