Subject Training

Context

A key enabler in combating climate change is the transition toward renewable energy sources such as solar and wind power. This transition has driven a major paradigm shift over the last decade: energy production and storage are becoming increasingly distributed, and alternating current (AC) components are progressively being replaced by direct current (DC) devices. In this context, DC microgrids have emerged as a promising paradigm for integrating renewable sources and local storage. Their operation relies heavily on power electronic converters, which manage power flows among multiple sources, storage elements, and loads. Ensuring efficient, stable, and reliable operation of these converters is therefore essential. These challenges form the core of the ANR JCJC project CASIMIR. The converters used in DC microgrids are commonly referred to as multiport converters because they interface several energy sources and loads through multiple ports. A key characteristic of these systems is that they exhibit degrees of freedom in their control inputs with respect to the primary control objective, namely maintaining the DC bus voltage. The main challenge addressed in this internship is to exploit these degrees of freedom to enhance performance while preserving stability and voltage regulation. The approach envisioned in CASIMIR relies on dynamic control allocation, which decomposes the actuation into three coordinated components: (i) a stabilizer guaranteeing voltage regulation, (ii) an optimizer distributing currents among the different ports to minimize losses or meet secondary criteria, and (iii) an annihilator ensuring that these secondary objectives do not interfere with the main regulation task. While such techniques are well established for linear systems, extending them to nonlinear systems⬔such as many multiport converter architectures⬔remains largely unexplored. The strategy adopted here is to focus on a well-understood multiport topology as a first step toward generalizing dynamic control allocation to nonlinear systems.

Objectives

This internship focuses on the study of a specific multiport converter architecture: parallel Boost converters. The goal is to design and evaluate a dynamic control allocation scheme and compare its performance with classical control approaches. The work is structured into the following tasks: - Build and validate a nonlinear dynamical model of the parallel Boost converter architecture. - Design a stabilizer ensuring voltage regulation at the DC bus, with an emphasis on Lyapunov-based methods. - Design an optimizer that distributes current among the converters to minimize conduction losses at steady state. - Design a steady-state annihilator guaranteeing that the voltage regulation objective is preserved. - Simulate the complete dynamic control allocation scheme in MATLAB/Simulink and compare its performance with standard control strategies (e.g., PI-based voltage control, droop control). - Compare the performances of the dynamic control allocation strategies with respect to existing approaches from the power electronics community.