An Artificial Neural Network for Automated Fault Detection

Julian Bitterwolf, Evgenia Rusak, Sebastian Reiter, Alexander Viehl, Oliver Bringmann
ITIDS Conference in Ufa, Russia
Intelligent and interconnected cyber physical systems are a key enabler for future cost-efficient, automated and flexible industrial production systems. Predictive maintenance and condition monitoring are important techniques in order to reduce costs associated with unnecessary maintenance or premature breakdowns. In this paper, we propose techniques from supervised learning for automated malfunctioning detection. For that purpose, we train an artificial neural network on time series data representing the internal system behavior. We present experimental results from an industrial motor control system. We use a digital twin of the electronic component that models the relevant features of the physical system. The obtained information can be used during the runtime of technical systems and installations for a criticality analysis and the subsequent selection of measures.
Online Sources
Research focus
Virtual Prototyping and Life Cycle Management, Big Data and Service Science, Safety and Security
Download .bib
Download .bib
Published by