Chip

Algorithm

Filter (signal processing)

Geodesy

Sampling (signal processing)

Model predictive control

Benchmark (surveying)

Artificial intelligence

Geography

Microcontroller

Control (management)

STM32

Computer science

Computer hardware

Computer vision

Control theory (sociology)

Telecommunications

IEEE Transactions on Industrial Informatics

This paper describes fast versions of model predictive control (MPC) algorithms and their practical implementation using the STM32 ARM microcontroller. Two MPC algorithms are considered: Dynamic Matrix Control (DMC) and Generalized Predictive Control (GPC). Computationally efficient Fast DMC (FDMC) and Fast GPC (FGPC) algorithms are derived in which the values of the manipulated variables are calculated from uncomplicated explicit formulas. To demonstrate effectiveness of the discussed algorithms, two emulated benchmark processes and a laboratory servo are considered. The influence of some tuning parameters on algorithms' calculation time is studied. It is shown that for short sampling periods the classical DMC and GPC algorithms fail to work since calculations last longer than the sampling period, which is unacceptable. Conversely, the FDMC and FGPC algorithms require only hundreds or tens of microseconds and single microseconds, respectively. Hence, proposed algorithms may be used for very fast processes, with very short sampling time.

Fast Analytical Model Predictive Controllers and Their Implementation for STM32 ARM Microcontroller