Performance Comparison of dSPACE and MCU MPC Implementation in PV Power Optimizer

Finite control set model predictive control (MPC) can be computationally intensive especially when scaled up for more complicated converters since each state variable in the system is evaluated for all possible switching configurations. Multi-objective optimization of the cost function adds to the complexity of MPC. Much of the MPC literature in power electronics relies on expensive rapid prototyping (RP) platforms (RPP) such as dSPACE and OPAL RT. This paper aims to bridge the gap between expensive RPP and low-cost MCUs by exploring the feasibility of rapid prototyping (RP) on the low-cost F28379D launchpad. This paper is aimed toward the practitioner or researcher who desires to reduce to practice the MPC methods developed on a RPP to a lower-cost embedded controller. The suitability of a low-cost Texas Instrument (TI) launchpad to perform the same rapid-prototyping functions needed for MPC regulator development in a full-featured, high-end RPP is discussed. Experimental results using the dSPACE 1006 and the F28379D Launchpad showing steady-state performance and dynamic response are illustrated and compared. Comparative results show the F28379D launchpad and dSPACE 1006 offer similar performance markers.