Steps and Challenges in Analyzing Real Sensor Data from a Productive Press Shop and its Value for Predictive Maintenance Application


Safa Evirgen1 and Maylin Wartenberg2, 1Volkswagen AG, Germany, 2Hochschule Hannover, Germany


This paper highlights the significance of AI-powered maintenance strategies in modern industry for operational optimization and reduced downtime. It emphasizes the crucial role of sensor data analysis in identifying anomalies and predicting failures. The research specifically examines sensor data from an automotive press shop, addressing questions related to data selection, collection challenges, and knowledge generation. By utilizing unsupervised learning on compressed air data from a press line, the study identifies patterns, anomalies, and correlations. The results offer insights into the potential for implementing an effective predictive maintenance strategy. Additionally, a systematic literature review underscores the importance of data analysis in production systems, particularly in the context of maintenance.


Predictive Maintenance, smart database, sensor data analytics, data science, production.