A Novel Fault-Identification Method for Micro Coils of EMECs Based on a Composite Analytical Model Combining a 2D...

Area of Science:

• Electrical Engineering
• Materials Science
• Computational Modeling

Background:

• Relays are critical electromechanical components (EMECs).
• Faults in micro-coils can compromise relay performance and safety.
• Existing fault identification methods may lack accuracy or efficiency.

Purpose of the Study:

• To develop a novel, accurate fault-identification method for micro-coils in relays.
• To integrate analytical and numerical thermal models with deep learning for fault detection.
• To address common micro-coil faults including thermal aging, delamination, and short circuits.

Main Methods:

• A composite analytical and 2D thermal model was developed, coupled via iterative correction.
• Physically constrained temperature sequences were generated for fault simulation.
• Orthogonal centerline temperature distributions were extracted as 1D feature vectors.
• A 1D Convolutional Neural Network (1D-CNN) was employed for classification.

Main Results:

• The hybrid thermal model demonstrated a low error of approximately 1.7% compared to finite-element analysis.
• The trained 1D-CNN achieved a high accuracy of 98.13% on 160 test samples.
• Experimental validation confirmed the method's ability to distinguish normal and various fault states.

Conclusions:

• The proposed method effectively identifies micro-coil faults in relays.
• The hybrid modeling and 1D-CNN approach offers a robust solution for fault diagnosis.
• This technique enhances the reliability and safety of electromechanical components.