Pixelization of Steady-state Variables Based Power System Transient Stability Assessment

Rapid and accurate implementation of transient stability assessment (TSA) in power systems can improve grid perception capabilities and reduce the likelihood of stability disruptions, making it crucial for the safe operation of power systems. In order to further enhance the speed and accuracy of transient stability assessment, this paper proposes a method for power system transient stability assessment based on pixelization of steady-state characteristic variables and deep machine learning. Firstly, the steady-state power flow data and electrical topology connection information are mapped to a multi-dimensional pixel feature map of active power-reactive power-voltage by using the proposed transfer mechanism, fully utilizing the advantages of convolutional neural networks (CNN) in feature map processing. Multiple deep convolutional neural networks are then used to extract the operating state features of the power grid for transient stability assessment. Finally, the proposed method is validated by using the New England 10-machine 39-bus standard system, and the results show that the data-model fusion optimization model, which fully retains the system topology and the magnitude and direction information of each electrical quantity, improves the performance of transient stability assessment compared to other machine learning methods, providing a new approach for online assessment and analysis of power systems.