Pneumatic Pressure and Electrical Current Time Series in Manufacturing

Published: 9 April 2024| Version 2 | DOI: 10.17632/ypzswhhzh9.2


Obtained at a real-world discrete manufacturing shopfloor, the dataset contains measurements of pneumatic pressure and electrical current. The dataset contains 7 days of operation, spanning roughly 150 processed pieces. The dataset contains time series pre-segmented at points in time where the internal state-machine of the observed process changes, i.e. at points in time when the machine transitions from one operation to the next. The labels contained within the dataset enable the application of supervised learning algorithms such as time series classification, as well as validation of unsupervised approaches such as time series clustering. The observed process is an end-of-line testing machine for consumer-grade small electric drive assembly (device under test – DUT). The machine takes several actions to evaluate each DUT, the measurements present in this dataset observe the pneumatic pressure powering actuators and the electrical current activating them. The data is pre-segmented using the testing machines internal state machine. Each segment corresponds to single action performed by the testing station in the process of manipulating the piece under observation. The segments contained in the dataset indicate characteristic pressure and current signatures for each action performed. This enables the dataset to be used in the design of time series algorithms aimed at non-invasive monitoring of (discrete) industrial processes.



Univerza v Ljubljani Fakulteta za Elektrotehniko, Institut Jozef Stefan


Clustering, Time Series Analysis, Manufacturing, Sensor, Time Series, Machine Monitoring, Pressure Measurement, System Fault Detection


Javna Agencija za Raziskovalno Dejavnost RS