Programmable logic controller implementation for model predictive control

Abstract

Model Predictive Control (MPC) is a method of advanced control and optimization, focused on minimizing the differences between measured, predicted, and reference values while staying within a set of predefined constraints. However, MPC algorithms scale exponentially in size as the prediction horizon grows. This means that large-scale MPC problems can require a huge amount of processing power and memory from the computers that solve them, making them impractical for many applications. Programmable Logic Controllers (PLCs), on the other hand, have been industry standard for automation and manufacturing since their creation, making them low-cost and readily available embedded systems. In an effort to make MPC more accessible, we propose the use of a PLC to solve MPC algorithms in real-time.