Condition monitoring and prognostic health management for electronics

With the current strong focus on the Internet of Things (IoT), autonomous vehicles and Industry 4.0, reliable and safe electronics are becoming increasingly important. By predicting the physical condition of complex electronic systems and displaying it in good time, failures with sometimes catastrophic effects can be avoided. Prognostic and Health Monitoring (PHM) are methods for monitoring electronic systems and predicting service life and expected failures. PHM brings together various of our competencies in the field of electronics reliability. This ranges from recording operating loads with customer-specific sensor technology (field load characteristics, mission profiles), to hot-spot analyses to identify critical areas of assemblies using finite element simulation and 3D X-ray inspection, through to linking machine learning algorithms with service life data.

Fraunhofer IKTS has developed a data-driven and Physics-of-Failure based modeled PHM approach for determining service life. This focuses on damage estimation of electronic components using a machine learning model that is trained with field data and synthetic failure data. In conjunction with developed service lifetime models, this enables the prediction of the remaining useful lifetime (RUL) of the components.

The PHM approach and its adaptability were successfully demonstrated using the example of motor electronics in e-bikes and power electronics in trams:

• Detection of operating loads (temperature, current, voltage, mechanical shock and vibration)
• Feature analysis (feature reduction, feature extraction)
• Hotspot analysis and development of the digital twin (virtual representation of real electronics)
• Accelerated ageing tests, service life modeling
• Development of AI/machine learning models for fault prediction
• Troubleshooting and fault diagnosis, physics-of-failure analysis
• Calculation of the remaining useful lifetime (RUL)