Power System Region Partition Method with High-Penetration of Renewable Energy Considering Frequency Temporal-Spatial Distribution Characteristics

The widespread application of new energy sources challenges traditional power system partitioning methods. This paper proposes a novel partitioning method based on frequency spatiotemporal distribution to address complex frequency response and uneven inertia distribution caused by new energy integration. The method involves modeling the equipment and network sides to obtain a closed - loop frequency response model. The system's return difference matrix is used to decouple the frequency response into common - mode and differential - mode frequencies through spectral decomposition. Each frequency component's lowest point and average rate of change are analyzed to measure the spatiotemporal frequency distribution. A damped sine curve fits the common - mode frequency trajectory, enabling unified structural parameter calculation. PMU data under various disturbances is used to build similarity and Laplacian matrices, and extract modal frequency feature vectors. An improved k - means clustering algorithm then groups similar frequency curves for system partitioning. Tested on a simulation platform with synchronous generators, grid - connected inverters, and grid - forming inverters, the method effectively captures frequency spatiotemporal distribution and achieves reasonable partitioning, offering a new approach for power system stability and control.