1.School of Electrical Engineering Beijing Jiaotong University, Beijing 100044, China;2.Electric Power Research Institute of China Southern Power Grid Company Limited, Guangzhou 510080, China

The process noise statistical parameters of state equations are unknown and time-varying in the dynamic state estimation of distribution network. Thus, the real-time estimation of the process statistical parameters is essential for the state estimation process. Inaccurate noise parameters will cause the decrease of performance or divergent filtering of unscented Kalman filter. In this paper, a dynamic state estimation method for distribution network based on improved robust adaptive unscented Kalman filter is proposed. The noise parameter statistical estimator in the method is made up of a biased and an unbiased estimator, which will improve the accuracy of noise parameters estimation during the state estimation process as well as guarantee the semi-positive definiteness of process noise variance matrix. As a result, this method can ensure the robustness of the algorithm. The simulations carried out on the IEEE 33-node system demonstrates that the proposed method can ensure high precision of state estimation when the system runs smoothly, the load changes violently or the initial noise parameters are set improperly.