Authors:
Padma Iyenghar
1
;
2
Affiliations:
1
innotec GmbH, Hornbergstrasse 45, 70794 Filderstadt, Germany
;
2
Faculty of Engineering and Computer Science, University of Applied Sciences Osnabrueck, 49009 Osnabrück, Germany
Keyword(s):
Data Quality, Data Accuracy, Functional Safety, Artificial Intelligence (AI), Machine Learning (ML), Predictive Maintenance, Reliability, Availability.
Abstract:
This paper focuses on the critical role of dataset accuracy in the context of machinery functional safety within an AI-based predictive maintenance system in a manufacturing setting. Through experiments introducing perturbations simulating real-world challenges, a decrease in performance metrics was observed—factors such as sensor noise, labeling errors, missing data, and outliers were identified as contributors to the compromise of the AI model’s accuracy. Implications for reliability and availability were discussed, emphasizing the need for high-quality datasets to minimize the risk of unplanned downtime. Recommendations include the implementation of robust data quality assurance processes and improved outlier detection mechanisms to ensure the reliability and availability of machinery in high-risk environments.