Fault Diagnosis of Electronic Components: Switch-mode AC/DC power supply

This project presents an integrated approach for condition monitoring of SMPS, which are essential in modern, space-constrained electronic devices. The focus is on diagnosing failures primarily associated with switching devices and capacitors under adverse working conditions. By leveraging dual sensing of both current and voltage signals, the proposed framework extracts statistically significant features to capture electrical stress effectively.

A correlation-based feature selection process was employed to identify the most salient features, which were then used in machine learning classifiers for improved fault detection and isolation (FDI). Among the classifiers tested, random forest and gradient boosting demonstrated high reliability, though they incurred greater computational costs compared to others. The decision tree classifier, on the other hand, provided a cost-efficient alternative with dependable diagnostic results.

Overall, the framework offers a comprehensive and effective methodology for diagnosing and classifying faults in SMPS, ensuring enhanced reliability in condition monitoring for electronic devices with limited space.

Grants

• Grand Information Technology Research support (IITP-2020-2020-0-01612) provided by the MSIT (Ministry of Science and ICT) South Korea (2020 -2021)

Journal Publication from the project

• Kareem, A.B.; Akpudo, U. E.; Hur, J.-W. An Integrated Cost-Aware Dual Monitoring Framework for SMPS Switching Device Diagnosis. Electronics 2021, 10, 2487. https://doi.org/10.3390/electronics10202487