Thermal Management and Anomaly Detection in Electric Machines

The field of electric machines is moving towards more advanced thermal management and anomaly detection techniques. Machine learning approaches are being explored for estimating internal temperatures and monitoring cooling efficiency, offering a promising alternative to conventional modeling methods. These data-driven methods have shown satisfactory performance even under transient conditions. Graph-based frameworks are also being developed for robust and interpretable fault diagnosis in rotating machinery, achieving high diagnostic accuracy and strong noise resilience. Furthermore, anomaly detection methods are being proposed for industrial control systems, focusing on timely detection of anomalies and providing explainability. Noteworthy papers include:

  • A paper proposing a graph-based framework for fault diagnosis, which achieves high diagnostic accuracy and strong noise resilience.
  • A paper presenting an anomaly detection method that involves accurate linearization of non-linear forms, providing millisecond time response and explainability.

Sources

Temperature Estimation in Induction Motors using Machine Learning

Reliable Thermal Monitoring of Electric Machines through Machine Learning

Graph-Based Fault Diagnosis for Rotating Machinery: Adaptive Segmentation and Structural Feature Integration

GiBy: A Giant-Step Baby-Step Classifier For Anomaly Detection In Industrial Control Systems

Built with on top of