Publications by authors named "Bernhard Humm"

Article Synopsis
  • Anomaly detection is crucial for maintaining the security and functionality of modern cyber-physical production systems, helping to identify and fix issues before they escalate into larger failures.
  • This study proposes an unsupervised, decentralized approach to real-time anomaly detection, utilizing 1D convolutional autoencoders and a sliding window method to enhance performance and adaptability.
  • The approach is tested in a real industrial setting, demonstrating its effectiveness in detecting anomalies across various processes within individual cyber-physical systems without requiring expert knowledge for implementation.
View Article and Find Full Text PDF

Smart factories are complex; with the increased complexity of employed cyber-physical systems, the complexity evolves further. Cyber-physical systems produce high amounts of data that are hard to capture and challenging to analyze. Real-time recording of all data is not possible due to limited network capabilities.

View Article and Find Full Text PDF