Data-Driven Approach for Current Control in DC-DC Boost Converters

Abstract

This work addresses the control of the inductor current in DC-DC boost converter using a data-driven approach. Model-based approaches tend to be either conservative or complex tasks when considering intrinsic non-linearities. In that sense, data-driven solutions to the control design of such processes come in handy as they do not need a plant model and, given a sufficiently rich experiment, all information needed from the system at a given operation point is obtained from data. Even for a specific data-driven method - in this work, the Virtual Reference Feedback Tuning (VRFT) - there are different ways to implement it, which allow tuning of different controller structures. In this work, we compare and apply these approaches on the control of the inductor current in a DC-DC boost converter directly on an actual plant and show that, for this kind of plant, the VRFT approach in which the inverse of the controller is identified along with the "derivative" pole outperforms the standard one.