LSTM-CNN-Based Transient Stability Assessment

In the face of huge challenges in security and stability analysis brought by the growing scale and complexity of modern power systems, the application of deep learning to the transient stability assessment (TSA) of a power system has become a topic with ever-increasing concerns from both academic and industrial communities. Aiming at improving the speed and accuracy of TSA as well as addressing the space-time features of electrical parameters in the transient process after disturbances, this paper proposes a TSA model based on the Long Short-Term Memory (LSTM) network and Convolutional Neural Network (CNN). This model can not only extract the temporal features during the transient process, but also can attain the spatial features at different nodes. Also, the model is optimized through performance comparisons among different structures and parameters. The presented model is demonstrated by the IEEE 39-bus power system with high accuracy, and is of the potential for on-line TSA.