Motor control and diagnostic software for six-phase three-level inverter

Two compatible variants of the inverter were considered, the IGBT inverter at BUT and the GaN inverter constructed by ZF. The Infineon AURIX 3rd generation (not yet publicly available) was targeted as a microcontroller in the AI4CSM project. Its GTM module is sufficient to generate all 36 PWM control signals of the inverter. The internal structure of the microcontroller is prepared for precise synchronization of the sensor measurements to the inverter PWM, which was implemented in low-level drivers. A key part of the motor control software is the DC bus middle point balancing, which was implemented in two parts. The first part is responsible for feed-forward compensation each PWM period. The second part controls both DC bus voltages to the same value. High-level motor control software was designed and tested in a MATLAB/Simulink environment. The three-level inverter was modeled as a Simscape schematic to verify the behavior of the DC bus middle point balancing algorithm. The algorithms were integrated into microcontroller firmware by C code generation from Simulink models. The diagnostic algorithms based on convolutional artificial neural networks were implemented into the Parallel Processing Unit (PPU) of the AURIX microcontroller using vector DSP instructions. For measurement and calibration on the inverter, the XCP protocol over the CAN FD communication bus was integrated into the microcontroller software. To interface the inverter with the testbench control system, the CAN communication stack based on the DBC files description was implemented using the Vehicle network toolbox in Simulink.