Wide-band oscillation modal analysis for renewable-integrated grids using wavelet transform and dynamic mode decomposition

This paper proposes a hybrid algorithm integrating Discrete Wavelet Transform (DWT) and Dynamic Mode Decomposition (DMD) to address the challenge of wide-band oscillation modal extraction in power systems with high renewable penetration. A key feature is the use of a fixed Db5 wavelet basis, which avoids the need for scenario-specific adjustments while covering the target frequency band (0.1–1kHz). The algorithm follows a two-stage sub-band isolation and spatiotemporal identification framework. First, DWT decomposes the signal into sub-bands for pre-separation of frequency components and noise suppression. Subsequently, DMD is applied independently to each sub-band to extract dominant spatiotemporal dynamic modes. This approach effectively converts the problem of identifying densely packed modes across the full spectrum into extracting sparser modes within individual sub-bands, thereby significantly improving modal discernibility. Simulation case studies demonstrate the method's accuracy in extracting oscillation frequency, damping ratio, and spatial mode shape. The results confirm the efficacy of the fixed wavelet basis, offering a practical, minimal-tuning solution for online wide-band oscillation monitoring.