FPGA implementation of online finite-set model based predictive control for power electronics

Recently there has been an increase in the use of model based predictive control (MBPC) for power-electronic converters. MBPC allows fast and accurate control of multiple controlled variables for hybrid systems such as a power electronic converter and its load. The computational burden for this control scheme however is very high and often restrictive for a good implementation. This means that a suitable technology and design approach should be used. In this paper the implementation of finite-set MBPC (FS-MBPC) in field-programmable gate arrays (FPGAs) is discussed. The control is fully implemented in programmable digital logic by using a high-level design tool. This allows to obtain very good performances (both in control quality, speed and hardware utilization) and have a flexible, modular control configuration. The feasibility and performance of the FPGA implementation of FS-MBPC is discussed in this paper for a 4-level flying-capacitor converter (FCC). This is an interesting application as FS-MBPC allows the simultaneous control of the output current and the capacitor voltages, yet the high number of possible switch states results in a high computational load. The good performance is obtained by exploiting the FPGA's strong points: parallelism and pipe-lining. In the application discussed in this paper the parallel processing for the three converter phases and a fully pipelined calculation of the prediction stage allow to realize an area-time efficient implementation.