Multivariate Sensor Dataset of an Industrial Rotogravure Printing Press (MSDIRPP)

Abstract

We present a multivariate sensor dataset for machine learning research in context of industrial print application. The dataset contains 7608 rolls of pre-processed multivariate sensor data of a single production scale rotogravure printing press. The data volume corresponds to 43.181 km of printed paperboard and paper. For each roll we provide high-resolution sampled inline sensor data, machine condition labels and several meta information. Besides basic information like machine speed the dataset contains web movement data such as web edge and web tension measurements, material measurement like web moisture and print quality data such as register measurements in cross and machine direction for 11 print units. We publish the dataset to provide data researchers a strong baseline dataset for several applications in industrial printing.