Direct Model Predictive Control Using Model Based Design

The hardware implementation of efficient direct model predictive control (MPC) algorithms for power electronic systems is a challenging and time-consuming task. Nowadays, model-based design, together with automated C code generation for embedded deployment is widely used in industry and academia for classical control schemes like field-oriented control for induction machines. Due to recent advances in terms of microprocessor computational power this design approach becomes interesting for MPC as well. Automatic code generation for CPU-based systems has the advantage of being considerably faster compared to bitstream generation for FPGA-based systems, which leads to shorter iteration times between different firmware versions. This paper discusses how direct MPC with reference tracking can benefit from model-based design and automated C code generation tools. A case study highlights the effectiveness of this approach by showing the development process of a multistep direct model predictive current controller for a variable speed drive system, from offline simulation to real-time deployment on a rapid control prototyping unit.