A Python package for the Virtual Reference Feedback Tuning, a direct data-driven control method

Abstract

In this paper, thepyvrft, a Python package for the data-driven control method known as Virtual Reference Feedback Tuning (VRFT), is presented. Virtual Reference Feedback Tuning is a control designtechnique that does not use a mathematical model from the process to be controlled. Instead, it uses input and output data from an experiment to compute the controller’s parameters, aiming to minimizean H2 Model Reference criterion. The package implements an unbiased estimate of the controller for MIMO (Multiple-Input Multiple-Output) processes using both least-squares and instrumental variabletechniques. The package also provides accessory functions to import data and to perform MIMO systems simulations, together with some examples.