Dynamic state estimation of power system with stochastic delay based on neural network

Abstract With the development of the wide-area power systems, time delays, including inherent and malicious network attacks time delays, are introduced into control signal communication channel, which jeopardize power system operation, control and protection. To eliminate the threats, state estimation was used to monitor the power system operating state. However, traditional state estimation methods are difficult to carry out due to the time delays have randomness. Therefore, a dynamic state estimation (DSE) method for power system based on delay and its stochastic characteristics was proposed in this paper. First, dynamic equations of important components, such as generators, exciters and power system stabilizers (PSSs), are introduced to describe complex power systems. In order to facilitate the use of Kalman filter (KF) methods, the power system state equations are discretized. Then, the correlation between stochastic delays and power system operating state indicated by neural network (NN). On the basis of the correlation, a KF method combined with NN has been proposed to adapt to the power system with stochastic delays. Finally, to verify the effectiveness of the proposed method, simulation experiments has been carried out on an actual power system. The simulation results show that the proposed method performs well in monitoring generators’ dynamic performance when the power system was disturbed by stochastic delays.