Abstract
Increasing scale of power grids induces higher computational complexity of static contingency analysis, which is in contradiction with the real-time requirements. Towards fast credible contingency analysis, a power flow calculation method based on graph neural network was proposed. This method can realize fast fitting of nonlinear relationship between new energy output, load data and branch power flow, node voltage in multiple scenarios. To avoid the disappearance of edge graph nodes caused by branch disconnection and to ensure the robustness of the model, a matrix reflecting the topology changes is designed in response to the network structure changes caused by various anticipated accidents. Besides, the load-source data is separated and the input eigenvector considering the change of the load-source data is constructed. The test of IEEE 30-bus and 118-bus system shows that the proposed model can adapt to changes of network topology brought by N-1 faults and the fluctuations of new energy, realize fast fitting of power flow, and provide a new tool for online fast credible contingency analysis.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
