Özet:

Purpose: In order to prevent breakdowns in production systems, predictive maintenance planning using Industry 4.0 infrastructure has been the focus of companies. In this study, a predictive maintenance decision support system integrated with internet-of-things (IoT) was developed and a pilot study was carried out on a machine to minimize the breakdowns and production downtime. Methodology: Temperature, humidity, and sound sensors have been used in order to provide continuous monitoring of the system. The data obtained with these sensors is transferred to a database via a network using IoT. In order to predict the system state (“breakdown may occur”, “good”) from this data, the machine learning techniques (Support Vector Machine and Decision Tree) are used. Historical breakdown records and production plan information are merged in order to find out the effect of production schedule on machine breakdowns. Findings: The proposed decision support system is able to make self-maintenance decision. Thus, it would be possible to minimize 1419 min. downtime of the machine that the pilot study was performed on. Originality: The effect of production sequence on system breakdowns has been investigated with sequential pattern mining algorithms. A maintenance decision support system including the integration of IoT, machine learning, predictive maintenance, sequential pattern mining and dynamic scheduling has been developed.

Anahtar Kelimeler: