AI Method for Power MOSFET I-V Characterisation of High Voltage - High Current Region
Date Issued
2025
DOI
10.1109/WiPDA63755.2025.11303434
Abstract
This paper presents a novel data-driven methodology for characterising the full I-V characteristics of a Silicon Carbide MOSFET using an Artificial Neural Network (ANN). The method uses a Machine Learning (ML) algorithm to extract the high voltage-high current characteristics of the device. By training the network with both static and dynamic measurement data, the method achieves high accuracy across a wide voltage and current range. It eliminates the need for extensive manual parameter extraction traditionally required by physical models, offering a more accessible and efficient pathway to device modelling. The ANN-based model demonstrates robust interpolation performance within the measured data range and shows encouraging extrapolation capabilities towards the device’s rated limits.