Computational optimization of a MIMO constrained embedded predictive controller implementing a novel software-based architecture

Abstract

Embedded controllers for multivariable processes have become a powerful tool in industrial implementations. Due to the portability that they can offer. However, embedded controllers face lower computational resources, than those implemented in PCs. This work presents a novel software-based architecture for a state-space MIMO constrained predictive embedded controller. The goal of this work is to improve the performance (Timing and resource usage) of the MPC. The performance validation for the proposed architecture was evaluated using the board NI myRIO 1900 as embedded system. A comparison amount a regular implementation and the new architecture is also discussed. Timing and processor usage show a reduction after the new architecture is implemented, with minimal impact on the behavior after comparing the implementation of the new architecture to the results obtained from the original algorithm.