A New Authentic Cloud Dataset from a Production Facility for Anomaly Detection

Emil Blixt Hansen, Nadeem Iftikhar, Simon Bøgh

Research output: Chapter in Book/Report/Conference proceedingConference contribution to proceedingpeer-review

Abstract

As technology advances and modern Industry 4.0 solutions are becoming more widespread, the need for better-suited datasets is rising. The commonly used datasets for training machine learning focus on simple data of often publicly available information. Within the industry, there is only a handful of datasets publicly available to use. In this paper, we present a new authentic industrial cloud data (AICD) dataset collected from an actual operating pick-and-place machine handling items with variations in shape, size, and weight. The AICD dataset contains various analogue sensor values and states of the machine, collected from an existing cloud solution. Within the data, an error is present when the machine fails. Therefore, this dataset is suited for testing and developing predictive maintenance and anomaly detection algorithms to be used in the industry. Moreover, the paper also presents a baseline implementation as a performance indicator for future models.
Original languageEnglish
Title of host publicationTowards Sustainable Customization : Bridging Smart Products and Manufacturing Systems - Proceedings of the 8th Changeable, Agile, Reconfigurable and Virtual Production Conference CARV 2021 and 10th World Mass Customization and Personalization Conference MCPC 2021
EditorsAnn Louise Andersen, Rasmus Andersen, Thomas Ditlev Brunoe, Maria Stoettrup Schioenning Larsen, Kjeld Nielsen, Alessia Napoleone, Stefan Kjeldgaard
Number of pages8
PublisherSpringer
Publication date2021
Pages415-422
ISBN (Print)978-3-030-90699-3
ISBN (Electronic)978-3-030-90700-6
DOIs
Publication statusPublished - 2021
Event8th Changeable, Agile, Reconfigurable
and Virtual Production Conference &10th World Mass Customization and Personalization Conference, 2021
- Aalborg Universitet, Aalborg, Denmark
Duration: 1 Nov 20212 Nov 2021

Conference

Conference8th Changeable, Agile, Reconfigurable
and Virtual Production Conference &10th World Mass Customization and Personalization Conference, 2021
LocationAalborg Universitet
Country/TerritoryDenmark
CityAalborg
Period01/11/2102/11/21
OtherThe CARV2021 conference is co-located with the 10th World Conference Mass Customization, Personalization & Co-Creation (MCPC2021).
SeriesLecture Notes in Mechanical Engineering

Fingerprint

Dive into the research topics of 'A New Authentic Cloud Dataset from a Production Facility for Anomaly Detection'. Together they form a unique fingerprint.

Cite this