Control of an Active Gate Driver for an Electric Vehicle Traction Inverter Using Artificial Neural Networks

Electric vehicle drivetrains using wide-bandgap semiconductors face challenges regarding EMI and ac-celerated machine aging due to the fast switching transients. This paper presents a method of controlling a variable-resistance active gate driver with the help of a neural network in order to reduce the drawbacks of fast switching while increasing efficiency. Measurements covering the whole MOSFET operating range and a sinusoidal inverter output current prove that the proposed method effectively reduces losses while also reducing switching speed and, in this way, reduces EMI issues and machine damage.