Explainable eigenvalue-based identification of inverter dynamics
Explainable eigenvalue-based identification of inverter dynamics
The proliferation of inverter-based resources (IBRs) presents a significant modeling challenge, creating a dichotomy between intractable first-principles (white-box) models and uninterpretable, purely data-driven (black-box) models. This paper introduces a framework that bridges this chasm by identifying an explainable, grey-box model for inverters from passive operational data. The methodology employs a Koopman autoencoder to discern the system's underlying linear dynamics in a latent space, from which dominant system eigenvalues are subsequently extracted. This process yields a physically meaningful, control-oriented linear model without requiring knowledge of the inverter's proprietary internal structure. The efficacy of the proposed framework is rigorously validated by comparing its captured eigenvalues and open-loop transient response against a high-fidelity benchmark, demonstrating high accuracy in capturing the true system dynamics.
Qing, Hanshan
3e2f5e64-d095-495b-8c68-236d53f9c3ac
Singh, Abhinav Kumar
6df7029f-21e3-4a06-b5f7-da46f35fc8d3
Batzelis, Stratis
2a85086e-e403-443c-81a6-e3b4ee16ae5e
Qing, Hanshan
3e2f5e64-d095-495b-8c68-236d53f9c3ac
Singh, Abhinav Kumar
6df7029f-21e3-4a06-b5f7-da46f35fc8d3
Batzelis, Stratis
2a85086e-e403-443c-81a6-e3b4ee16ae5e
Qing, Hanshan, Singh, Abhinav Kumar and Batzelis, Stratis
(2026)
Explainable eigenvalue-based identification of inverter dynamics.
2026 IEEE POWER & ENERGY SOCIETY GENERAL MEETING, PALAIS DES CONGRÈS DE MONTRÉAL, Montreal, Canada.
19 - 23 Jul 2026.
(In Press)
Record type:
Conference or Workshop Item
(Paper)
Abstract
The proliferation of inverter-based resources (IBRs) presents a significant modeling challenge, creating a dichotomy between intractable first-principles (white-box) models and uninterpretable, purely data-driven (black-box) models. This paper introduces a framework that bridges this chasm by identifying an explainable, grey-box model for inverters from passive operational data. The methodology employs a Koopman autoencoder to discern the system's underlying linear dynamics in a latent space, from which dominant system eigenvalues are subsequently extracted. This process yields a physically meaningful, control-oriented linear model without requiring knowledge of the inverter's proprietary internal structure. The efficacy of the proposed framework is rigorously validated by comparing its captured eigenvalues and open-loop transient response against a high-fidelity benchmark, demonstrating high accuracy in capturing the true system dynamics.
Text
Explainable_Eigenvalue_Based_Identification_of_Inverter_Dynamics
- Accepted Manuscript
More information
Accepted/In Press date: 1 March 2026
Venue - Dates:
2026 IEEE POWER & ENERGY SOCIETY GENERAL MEETING, PALAIS DES CONGRÈS DE MONTRÉAL, Montreal, Canada, 2026-07-19 - 2026-07-23
Identifiers
Local EPrints ID: 510822
URI: http://eprints.soton.ac.uk/id/eprint/510822
PURE UUID: 65777701-ba65-44a8-b5ae-ece5ab33ea28
Catalogue record
Date deposited: 22 Apr 2026 16:50
Last modified: 23 Apr 2026 02:07
Export record
Contributors
Author:
Hanshan Qing
Author:
Abhinav Kumar Singh
Author:
Stratis Batzelis
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics