A Power Electronic Converter Control Framework Based on Graph Neural Networks -- An Early Proof-of-Concept

TL;DR

This paper introduces a topology-agnostic control framework for power electronic converters using graph neural networks, enabling transferability across different designs and demonstrating near-optimal performance in simulations.

Contribution

It presents a novel meta-control approach that encodes converter netlists as graphs and trains a task-conditioned GNN for universal control across heterogeneous converter topologies.

Findings

Achieves near-optimal tracking in simulations

Demonstrates transferability across converter designs

Uses differentiable predictive control for training

Abstract

Power electronic converter control is typically tuned per topology, limiting transfer across heterogeneous designs. This letter proposes a topology-agnostic meta-control framework that encodes converter netlists as typed bipartite graphs and uses a task-conditioned graph neural network backbone with distributed control heads. The policy is trained end-to-end via differentiable predictive control to amortize constrained optimal control over a distribution of converter parameters and reference-tracking tasks. In simulation on randomly sampled buck converters, the learned controller achieves near-optimal tracking performance relative to an online optimal-control baseline, motivating future extension to broader topologies, objectives, and real-time deployment.