J. Sobota

Raspberry Pi and Arduino Boards in Control Education

Abstract Nowadays the control education usually heavily relies on the available simulation packages and virtual laboratories, both of which have their irreplaceable position in the educational process. But unfortunately the closed loop experiments are way too often limited to these virtual domains and the students lack the physical feedback about the impact of control algorithms and its parameters. The reasons might be the price of the hardware setup to demonstrate the control algorithms physically or a complicated transition from simulation to real-time platform. This paper describes an extremely inexpensive, straightforward and surprisingly powerful platform for implementation of real-time control algorithms. The platform consists of an Arduino board and a Raspberry Pi running the REX Control System. The Arduino board is used for interaction with the physical world via its inputs and outputs. The REX Control System allows the students to develop and verify the control algorithms in Simulink and then run it in real-time by a few mouse clicks. However, the REX Control System is by no means dependent on Simulink, it is fully functional even if Simulink license is not available. The platform further bridges the gap between the virtual and physical worlds as it is tightly connected to PIDlab.com and Contlab.eu portals, which makes it an ideal choice for control education purposes.

Simple Pulse-Step Model Predictive Controller

Abstract This paper describes a novel simple model based predictive controller with manipulated value constraints. This controller is suitable for substitution of the classical PID controller used in industrial practice. It is assumed that the controlled system is stable, linear and t-invariant FIR system. The discrete step response sequence is used as the process model. Alternatively it is possible to use three-parameter model. To make the open-loop optimization easier the set of admissible control sequences is restricted to stepwise pulse-step sequences. The optimization procedure is then executable in reasonable time. A single tuning parameter is available for manual fine-tuning of the controller - the control moves penalty coefficient.

Automatic Tuning of the Pulse-Step Model Predictive Controller

Abstract This paper describes an autotuning procedure for single-loop model based predictive controller with manipulated value constraints focused on applications in process control. It is assumed that the controlled process is stable, linear and t-invariant FIR system described by three-parameter model derived from a simple identification experiment. Controller parameters are computed from the identified model. A single parameter intended for fine-tuning of the controller is available. Exact intervals are given to vary this parameter within.