Data-Driven Stability Assessment of Power Electronic Converters with Multi-Resolution Dynamic Mode Decomposition

TL;DR

This work addresses harmonic instability in grid-tied power-electronic converters by adopting data-driven stability assessment through multi-resolution dynamic mode decomposition (MR-DMD). By recursively applying DMD across time-bin levels and screening slow modes, MR-DMD uncovers dynamic modes across multiple frequency ranges and time scales, improving robustness to transient behavior and data gaps. The authors derive explicit parameter relationships for MR-DMD (subsample size $\mu$, termination level $L$, and slow-mode threshold $\rho$) and validate the approach on a downscaled electrified-railway single-phase converter, successfully identifying an 8.6 Hz low-frequency oscillation and related modal content in both DC-side and AC-side signals. The results demonstrate that MR-DMD can reconstruct measurements and extract stability-related eigenvalues more reliably than standard DMD under imperfect data conditions, offering a practical, data-driven stability diagnostic for real-world power-electronic systems.

Abstract

Harmonic instability occurs frequently in the power electronic converter system. This paper leverages multi-resolution dynamic mode decomposition (MR-DMD) as a data-driven diagnostic tool for the system stability of power electronic converters, not requiring complex modeling and detailed control information. By combining dynamic mode decomposition (DMD) with the multi-resolution analysis used in wavelet theory, dynamic modes and eigenvalues can be identified at different decomposition levels and time scales with the MR-DMD algorithm, thereby allowing for handling datasets with transient time behaviors, which is not achievable using conventional DMD. Further, the selection criteria for important parameters in MR-DMD are clearly defined through derivation, elucidating the reason for enabling it to extract eigenvalues within different frequency ranges. Finally, the analysis results are verified using the dataset collected from the experimental platform of a low-frequency oscillation scenario in electrified railways featuring a single-phase converter.

Figures (9)

• Figure 1: System diagram of a single-phase converter in electrified railways.
• Figure 2: Experimental setup of a down-scaled single-phase converter in an electrified railway system.
• Figure 3: Schematic of dynamic mode decomposition (DMD) (SVD: singular value decomposition, ED: eigen-decomposition).
• Figure 4: Measured signal reconstruction and eigenvalues identification by DMD using (a) normal dataset and (b) dataset with missing data — DMD fails to accurately assess stability based on the latter.
• Figure 5: Schematic of multi-resolution dynamic mode decomposition (MR-DMD) — DMD is performed recursively in varying time scales and slow modes are screened in each level.