Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Detecting Causalities in Production Environments Using Time Lag Identification with Cross-Correlation in Production State Time Series

: Saller, Dirk; Kumova, Bora I.; Hennebold, Christoph


Rutkowski, Leszek:
Artificial Intelligence and Soft Computing. 19th International Conference, ICAISC 2020. Proceedings. Pt.II : Zakopane, Poland, October 12-14, 2020, held virtually
Cham: Springer Nature, 2020 (Lecture Notes in Artificial Intelligence 12416)
ISBN: 978-3-030-61533-8 (Print)
ISBN: 978-3-030-61534-5 (Online)
ISBN: 978-3-030-61535-2
International Conference on Artificial Intelligence and Soft Computing (ICAISC) <19, 2020, Online>
Conference Paper
Fraunhofer IPA ()
Produktionsoptimierung; Fertigungsprozess; Korrelationsanalyse; Wissenserschließung; Zeitverzögerung

One objective of smart manufacturing is to resolve complex causalities of production processes, in order to minimize machine idle times. We introduce a methodology for mining from raw state time series of machines, possible causal relations between the machines of a given production environment. By applying the similarity measure cross-correlation on binary production state time series of two machines pairwise, we obtain a probability distribution, whose characteristic properties imply possible causal orderings of the two machines. In case of complex causalities, the measure may be applied to all possible machines pairwise, in order to extract a complete web of statistically significant causalities, without any prior context information of the environment. In this paper, we analyze the characteristic properties of such probability distributions and postulate four hypotheses, which constitute the steps of our methodology. Furthermore, we discuss the stochastic and temporal conditions that are necessary for the transitive propagation of causal states.