Koopman Based Control of a Boost Converter Feeding a Constant Power Load

In this paper, a Koopman based control technique is presented for voltage regulation of a boost converter feeding a constant power load (CPL). The nonlinearity of the boost converter and CPL can make the system unstable, with closed loop stability typically ensured only locally. To address these challenges, a data-driven Koopman operator approximation is developed for output voltage regulation using Koopman based model predictive control (KMPC). The tracking performance is demonstrated when the system is subject to changes in voltage reference and load power. Closed-loop simulation results are shown offline using Matlab/Simulink, and in real time using Opal RT. The KMPC problem is solved using mpcqpsolver, which allows for code generation and real time simulation, to demonstrate feasibility of online implementation of the controller.

I. Introduction

Due to the recent advances in power electronics and increased deployment of dc microgrids, there is an increase in power-electronics-enabled loads employing point-of-load power conditioning and control. These loads, sometimes acting as constant power loads (CPLs), introduce negative incremental resistance [1] when more current is drawn and the voltage decreases. This negative impedance can make the CPL behave as a source that supplies power back into the system, leading to stability issues [2]. Hence, the study of CPLs is not only important, but also practical in applications such as Plug-in Hybrid Electric Vehicle (PHEV) [3], More Electric Aircrafts (MEA) [4], All Electric Ships (AES) [5], etc. In order to ensure proper functioning of these dc systems feeding CPLs, control strategies are required for the converters. In particular, the design of controllers for boost converters with CPLs is challenging because of the economical and computational cost involved in addressing the nonlinearity and instability of the system.