A data-driven distributed and easy-to-transfer method for short-term voltage stability assessment

In this paper, a novel data-driven method named Gated Recurrent Graph Attention Network (GRGAT) for STVS assessment is developed by learning the relationship between system dynamics during faults and the corresponding transient voltage security index (TVSI). GRGAT can capture the spatial-temporal correlation in the power system, because the attention operations of bus information are performed directly on the system topology and the system dynamics are captured with gated recurrent units. Particularly, all operations are independent between buses. Therefore, GRGAT is not only distributed during online application, but also easy-to-transfer, which can adapt to the change of topological structures. To show the feasibility, adversarial adaptation is adopted to transfer learned knowledge for another modified network. The effectiveness and efficiency of GRGAT are demonstrated on the New England 10-Generator-39-Bus system and its modified systems. Simulation results also show the potential of this learning technique in knowledge transfer.