RT23 | On-line Monitoring of Maximum Temperature and Loss of a MFT using FEM, ANNs and OPAL-RT

Losses and maximum temperature are considered as critical indicators of the health status of medium frequency magnetic components. However, obtaining accurate estimates for both losses and maximum temperature presents significant challenges due to the complex nature of these components. Factors contributing to this complexity include the variation in core losses with respect to average temperature, variations in winding losses due to litz wire manufacture, and the impact of high-frequency effects. In this presentation, we introduce a novel Real-Time estimation approach developed and integrated into OPAL-RT. This approach uses Artificial Neural Networks (ANN) trained using Finite Element Simulations (FEM). Experimental results reveal that ANNs, which return maximum internal temperatures and forced convection losses, were successfully integrated into OPAL-RT, achieving an error rate of less than 2% over a wide range of scenarios while working in real time. Furthermore, this accuracy is maintained even when OPAL-RT simultaneously handles the complex modulation of the DAB converter.