System Partitioning Primary Frequency Regulation Capability Evaluation Based on DBSCAN Clustering

The high proportion of new energy sources and large-capacity DC feeds have led to a continuous weakening of system inertia support and primary frequency regulation capabilities. This has highlighted the temporal and spatial dispersion of system frequency response. This dispersion introduces certain errors in evaluating primary frequency regulation capabilities, resulting in unnecessary economic losses for units participating in frequency response and significantly reducing their initiative. To mitigate the impact of frequency dispersion on the primary frequency regulation capability of units, this paper proposes a system partitioning method based on the DBSCAN clustering algorithm. Firstly, to effectively reduce the impact of frequency response dispersion, the DBSCAN clustering algorithm is introduced to partition the power system, with the optimal partitioning results evaluated using the DB index. Then, sensitivity analysis is used to perform disturbance analysis on nodes within each region to determine the optimal frequency measurement point. Finally, a partial regional system of the East China Power Grid is established in the PSD-BPA simulation software to verify the effectiveness of the proposed method. Simulation results show that this method can effectively reduce the impact of frequency dispersion on primary frequency regulation assessment results, making the evaluation more accurate and reliable.