To main content

Application of Black-Box State-Space Model Identification Methods to Power Electronic Systems

Abstract

This paper evaluates three classes of system identification methods when utilized to extract multiple-input multiple-output (MIMO) black-box models of power electronic systems for use in small-signal stability studies. The studied methods include Vector Fitting (VF), Subspace Identification (SID), and the Prediction Error Method (PEM). Using measured frequency responses, these methods are evaluated in terms of accuracy and computational efficiency when identifying a state-space model of a two-level (2L) voltage source converter (VSC) in both gridforming mode, controlled as a Virtual Synchronous Machine (VSM), and grid-following mode (GFL). System identification is performed on several different control schemes considering different levels of modeling complexity, including one case based on an experimental setup with a 2L VSC. The results highlight some trade-offs between identification accuracy and computational cost: SID-based methods offer the fastest computation time and best scaling with increasing system and frequency sampling size, while VF-based methods offer superior accuracy in eigenvalue identification and frequency-response matching. The presented discussions provide guidance for a complete model identification process and for selecting suitable techniques for stability and interaction studies based on black-box models that preserve the intellectual property rights of power electronic systems.

Category

Academic article

Language

English

Author(s)

Affiliation

  • SINTEF Energy Research / Energy Systems
  • Research Centre Jülich
  • Norwegian University of Science and Technology

Year

2026

Published in

IEEE Journal of Emerging and Selected Topics in Power Electronics

ISSN

2168-6777

Page(s)

1 - 1

View this publication at Norwegian Research Information Repository