Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Use Digital Twins to Support Fault Diagnosis From System-level Condition-monitoring Data

Killian Mc Court, Xavier Mc Court, Shijia Du, Zhiguo Zeng

TL;DR

This paper proposes to use a digital twin to support developing data-driven fault diagnosis model to reduce the amount of failure data used in the training process and results showed that the deep learning model trained by digital twins is able to diagnose the locations and modes of 9 faults/failure from 4 different motors.

Abstract

Deep learning models have created great opportunities for data-driven fault diagnosis but they require large amount of labeled failure data for training. In this paper, we propose to use a digital twin to support developing data-driven fault diagnosis model to reduce the amount of failure data used in the training process. The developed fault diagnosis models are also able to diagnose component-level failures based on system-level condition-monitoring data. The proposed framework is evaluated on a real-world robot system. The results showed that the deep learning model trained by digital twins is able to diagnose the locations and modes of 9 faults/failure from $4$ different motors. However, the performance of the model trained by a digital twin can still be improved, especially when the digital twin model has some discrepancy with the real system.

Use Digital Twins to Support Fault Diagnosis From System-level Condition-monitoring Data

TL;DR

This paper proposes to use a digital twin to support developing data-driven fault diagnosis model to reduce the amount of failure data used in the training process and results showed that the deep learning model trained by digital twins is able to diagnose the locations and modes of 9 faults/failure from 4 different motors.

Abstract

Deep learning models have created great opportunities for data-driven fault diagnosis but they require large amount of labeled failure data for training. In this paper, we propose to use a digital twin to support developing data-driven fault diagnosis model to reduce the amount of failure data used in the training process. The developed fault diagnosis models are also able to diagnose component-level failures based on system-level condition-monitoring data. The proposed framework is evaluated on a real-world robot system. The results showed that the deep learning model trained by digital twins is able to diagnose the locations and modes of 9 faults/failure from different motors. However, the performance of the model trained by a digital twin can still be improved, especially when the digital twin model has some discrepancy with the real system.

Paper Structure

This paper contains 15 sections, 1 equation, 4 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Motivating example
  3. Digital failure twin for the robot
  4. Digital model
  5. Failure behavior model
  6. Other elements in the digital twin model
  7. Fault diagnosis based on digital failure twin
  8. Generate training data from the digital failure twin
  9. Collect test data from the real robot
  10. Training and evaluating the fault diagnosis model
  11. Results and discussions
  12. Training process
  13. Average accuracy over all the classes
  14. Performances on different classes
  15. Conclusions and future works

Figures (4)

  • Figure 1: The developed digital failure twin for the robot.
  • Figure 2: An illustration of one trajectory: Response of motor 1.
  • Figure 3: Training process with different termination epochs.
  • Figure 4: Simulation results from the digital twin model on one test dataset.