Dynamic State Estimation of Distribution Network Base on Pseudo Measurement Modeling and UPF

Abstract

The economic and reliable operation of distribution network relies on state variables information from state estimator. However, since the amount of distributed generation (DG) and electric vehicles connected to distribution network is steadily growing, the distribution network is full of randomness and volatility, which increases the difficulty of accurate state estimation. In this paper, a dynamic state estimation (DSE) algorithm based on unscented particle filter (UPF) is proposed. This algorithm uses the unscented Kalman filter (UKF) estimated results as the proposal distribution of particle filter (PF), and updates the latest measurements information in the process of generating predictive particles, which makes the distribution of particles much closer to the posterior probability distribution of true states. At the same time, a pseudo measurement model based on spiking neural network (SNN) is established, which outputs the pseudo measurement load at each sampling moment to enhance the accuracy of state estimation. The simulation results show that the proposed algorithm can effectively track the true value of distribution network states and is superior to the traditional DSE algorithm in estimation accuracy and noise robustness.