PRESENTATION

Intelligent Power Semiconductor Computing

To enhance the reliability of power electronics, deep EdgeAI is becoming an essential approach. One challenge is to scale AI massively to achieve a full distributed deployment, ideally into the power electronics itself. Therefore, a lightweight machine learning approach for estimating the Remaining Useful Life of IGBT and Silicon Carbide transistors is a useful contribution. By leveraging Radial Basis Function networks, we achieve state-of-the-art accuracy with a fraction of the computational cost. Our method is fully deployable on ultra-low memory DSPs, supporting on-chip learning and inference. This breakthrough would allow AI integrated in power transistors to move beyond simple switching, incorporating unprecedented, embedded intelligence and real-time diagnostic capabilities within a single package.