Short-term Voltage Stability Assessment of AC/DC Power Grid based on SDAE and ADAM Algorithms

With the development of HVDC transmission and changes of load composition and characteristics, the short-term voltage problem seriously threatens the safety and stable operation of power systems. A short-term voltage stability assessment method based on deep learning neural network for AC/DC receiving-end power grid is proposed in this paper. The stacked denoising autoencoder (SDAE) and adaptive moment estimation (ADAM) algorithms are used to build a rapid evaluation model, and the training sample set is used to train the transient voltage stability rapid evaluation model. The steady-state power flow features are used as inputs and an index of quantifying short-term voltage stability of commutation bus is utilized as the output. Simulation results of real multi-infeed AC/DC power grid demonstrate the effectiveness of the proposed method.