Analyzing Manufacturing Process By Enabling Process Mining on Sensor Data

Typical manufacturing processes involve various machines, each of which may be equipped with a variety of sensors. Digital twins can be used to model how the machines operate and support analysts in issue identification and identifying potential improvements in the process. For a complete view of the status of a machine, however, models need to be enriched to identify patterns over changes in the measurements of sensors and correlations between these sensors. Process mining techniques could be usefully applied in this context, given that they provide descriptive analyses to explain and simulate physical objects based on event logs storing multi-perspective data about the process. However, although sensors generate a vast amount of data about the status of machines on the production floor, they cannot be directly used by process mining techniques. To tackle this issue, we introduce a method that creates a custom event log from sensor data based on the process analysts interests. To this end, we propose different encodings for the sensor data. An exploratory experiment using real-life data from an industrial partner shows the effectiveness of our approach.