Small-signal stability assessment based on virtual regional eigenvalue

Small-signal stability assessment (SSSA) plays a significant role in large-scale power systems operation. Since the computational burden and unavailability of the detailed time-varying model, it is practically difficult to use model-based approaches to meet the real-time demands of SSSA. Considering the neighbour eigenvalues share similar oscillation characteristics, this paper presents an efficient method for SSSA based on looking at the regions that include the neighbour eigenvalues of interest rather than the exact position of eigenvalues. After dividing the insufficient damping area into several regions, the limitations of existing data-driven SSSA resulting from the uncertainty of critical eigenvalues are overcome by virtual regional eigenvalue (VRE), which is introduced to estimate the prominent eigenvalue of neighbour eigenvalues in each divided region. A composite structure based on Long short-term memory is designed to accommodate the different number of eigenvalues, generate VRE and the boundary of eigenvalues in divided regions. The effectiveness of the proposed method is demonstrated by two different scale test systems.