Automatic model predictive control implementation in a high-performance microcontroller

In this paper, a framework for embedding Model Predictive Control (MPC) for Systems-on-a-Chip applications is presented. This contribution is especially interesting when dealing with high performances devices since it increases the number of addressable control applications in the industrial field and particularly in fast systems. Aiming to allow the implementation of such a computationally expensive controller on chip, we propose optimization hints satisfying trade-offs between code size and computation speed versus digital precision and effectiveness of the computed control action. The illustration of the proposed implementation is tested on a high performance STM32F407 microcontroller. Analysis of the accuracy of the digital implementation of MPC algorithm are given, and it shows that the proposed framework controls successfully the process with a good set-point tracking and a low computational burden with a high speed.