A Multibranch Observer Strategy for Coherent Bus Clustering in Voltage-Stability Analysis

Voltage collapse remains a critical threat to modern power systems operating under increasingly stressed conditions, and new practical techniques are required to visualize operational risks. Therefore, this paper proposes a method for identifying coherent bus groups in power systems by combining reactive power flow sensitivity analysis with a multibranch observer strategy. For each selected observer branch, buses were clustered according to the similarity of their sensitivity values within a tolerance range of 0–10%, enabling a systematic evaluation of groups across varying thresholds. Coherent groups were confirmed by intersecting candidate clusters obtained from multiple-branch observers and retaining only those with a support index of 70% or higher. The method was validated on two benchmark systems, the Bonneville Power Administration (BPA) and Nordic Power System (NPS), demonstrating its applicability to different networks. The results show that the identified coherent buses share similar reactive power reserve margins and generator depletion patterns, confirming their physical significance for voltage stability assessment. In the BPA system, two buses exhibited the strongest coherent buses, with a recurrence of 92.9% across the evaluated tolerance range, while other groups of buses also showed relevant support levels. In the NPS, the strongest coherent bus pairs reached support values between 85.3% and 90.2%. Another pair of buses showed the highest support, increasing from 80.4% at a 3% tolerance to 90.2% at 9–10%, while other groups of buses reached support levels between 85.3% and 87.3%. Additional groups appeared mainly under higher tolerance levels, with support values close to the 70% acceptance threshold. Random observer validation also showed that the dominant coherent structures could be preserved even when only a reduced number of branch observers were used. Compared with existing sensitivity-based approaches, the proposed method can visualize groups of buses according to different branch observers, evaluate the classification according to a tolerance criterion, and distinguish strong coherent buses in regions with similar reactive behaviors. This approach offers power system operators an interpretable tool for identifying coherent buses prior to a more detailed study.