A Domain Knowledge-based Approach for Fault Diagnosis

Sorry you have no rights to view this entry!

5 Comments

  1. Mr. Alexander, thank you for the presentation! Is your error limit a stable value, or dynamic one which is updated using related machining parameters and domain knowledge? When domain knowledge is included, how do you determine its effect to calculate current error limit?

    • Thank you for your question. Both limit values mentioned are initially assigned statically and were determined experimentally in order to introduce variability into the system. During operation, the limit values are self-learned and vary depending on the process. Higher uncertainties can be mapped and an early match with domain knowledge can take place. The domain knowledge is only used for the later classification, not for the initial detection. The detection is based on similarity metrics in relative comparison to the self-learned intervention boundaries.

  2. Thanks for your presentation.

  3. The signals were taken directly from DMG machine ? How do you collect them ?

    • Thank you. Yes, we took it directly from the machine with an additional edge device at the control unit. Actually, there is no possibility to get the data in real-time. We measured a time delay of <1 sec, due to the hardware. For our cases this is permitted.

Leave a Reply

Your email address will not be published. Required fields are marked *