Static Security Assessment with Accuracy Evaluation Based on Neural Networks

Focusing on assessing the safety status of power systems, static security assessment involves solving huge amounts of nonlinear power flow equations for analyzing the successive processes of power redistribution and component outages after an initial contingency of the grid, a neural network model that makes fast approximations of power flows during the successive process is proposed. This model also predicts the cascading processes of line failures with an indicative function, which considers overflows of lines and the corresponding line failures. Thus, this model is also used to predict the grid topologies during the cascading line failure process, whose accuracy is also estimated. The effectiveness of the proposed model is verified on the IEEE 5, 24, and 118-bus systems.