Model-Based Design and AI for Monitoring Systems in Automotive Power Electronics

The article presents an innovative methodology for the design and validation of monitoring and anomaly detection algorithms, focused on the aging phenomenon linked to the anomalous modification of the Rd in the switching devices of electronic systems of power in high-performance electric vehicles. The case study concerns an electric traction system with a three-phase axial flux synchronous motor integrated into a drive wheel (Elaphe) and a high-efficiency three-phase inverter with SiC (silicon carbide) technology. The methodology is developed in four phases: 1) creation of a real-time model of electric traction, validated with experimental data from WLTP tests; 2) Generation of a virtual dataset representing aging, via anomaly injection emulating the phenomenon with a scaling factor based on the Rd on the motor phase current; 3) Design of an estimator of the Rd using an Artificial Neural Network (ANN) regression model, including feature extraction and reduction techniques. 4) Experimental validation of the method through PIL (Processor-In-the-Loop) tests, integrating the monitoring algorithm on the NXPs32k144 embedded platform, making it interact with the electric traction model with anomaly injection. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.