A Two-steps Splitting Strategy Based On Laplacian Eigenmap Algorothm For Large Power Grid With Power Wind Connected

This paper proposes a two-steps splitting strategy for large power grid with high wind power permeability. In the first phase, the wind power is processed by modifying the shrinkage admittance matrix of the system. Then the classification result of the coupling degree of the generators can be calculated off-line, according to which the power angle Laplacian matrix is corrected online. After that, the current homology clustering results are obtained. In the second phase, the problem of determining the splitting surfaces is transformed into the problem of solving generalized eigenvalues by using constrained spectrum clustering algorithm based on graph theory. And the real-time optimal splitting surfaces will be quickly obtained using the improved k-means algorithm. Finally, the strategy is proved to be correct, rapid and effective through the simulation of IEEE-39 bus system.