Dataset from a Smart Factory to evaluate a Semi-automated Approach to Detecting Process-Level Activities from Sensor Data

This dataset contains sensor and process data related to the execution of storage and production processes collected from a Fischertechnik Industry 9.0V smart factory model available at the University of St.Gallen. This dataset is used in the evaluation of the paper "A Semi-automated Approach to Detecting Process-level Activities from Sensor Data", authored by Luciano García-Bañuelos, Mauricio Jacobo González González, Ronny Seiger, Marco Franceschetti and Alejandra Guadalupe Silva Trujillo, submitted to the 8th International Conference on Emerging Data and Industry (EDI40).

In this dataset, we used a custom Python-based software stack to control the smart factory via a business process system (Camunda Platform) that calls the functionality of the smart factory via web services implemented in Python flask. The sensor data from the smart factory is collected at a frequency of 10 Hz.

The data set contains the following files with low-level IoT (sensor) data:

training_tenhertz_log_20230411-095748.txt: contains the training data comprising 5 executions of each, the storage and production process.

test_tenhertz_log_20230411-103455.txt: contains a larger dataset for testing with several executions of the storage and production processes.

Each entry in these files corresponds to one message (as JSON object). Each line contains all the readings of all the sensors, actuators and additional data from one CPS component (i.e., production station) at one point in time.

Moreover, the dataset contains the following process (BPM)-related files:

storage_process.bpmn: executable BPMN 2.0 model of the storage process that has been executed several times via the Camunda Platform BPM system to control the smart factory

production_process.bpmn: executable BPMN 2.0 model of the production process that has been executed several times via the Camunda Platform BPM system to control the smart factory

camunda_process-instance.json: event log generated by the BPM system regarding the process instance execution (ground truth)

camunda_activity-instance.json: event log generated by the BPM system regarding the activity instance execution (ground truth)