Real-Time Preventive Control of Cascading Outage Risk Based on Generative Adversarial Network

The impact of power outages caused by cascading outage is severe. In order to deal with the time- varying power flow conditions and reduce the risk of cascading outage of power system, this paper proposed a real-time preventive strategy of cascading outage risk based on the generation of adversarial network. Firstly, the preventive control of cascading outage risk was modeled as an optimization problem considering risk threshold and defense cost. Then, based on the WGAN-GP algorithm, the discriminator and generator were trained with condition-risk samples, and the constraints of output balance, generator capacity and line capacity were considered in the training process. Finally, after the training, with the change of the initial condition, the generator could optimize the output of the generator online and reduce the cascading outage risk of the system in real time. The simulation results of IEEE 39-bus system show that, the proposed method has good convergence, can effectively reduce the risk of cascading outage, and the solution speed is fast, which meets the requirement of real-time control.