Abstract: To address the issues of the lack of real-time measurement, low accuracy of pseudo measurement, and the assumption of constant system state process noise in existing dynamic state estimation (DSE) methods for medium voltage distribution networks, this paper proposes a robust DSE method for medium voltage distribution networks based on an improved adaptive unscented Kalman filter (UKF) algorithm. Firstly, the method uses the smart meter measurement and transformer model at the low voltage side of the distribution transformer in the medium voltage distribution network to derive the equivalent medium voltage measurement to enhance the measurement redundancy of the medium voltage distribution network. Then, this article draws on signal processing technology to update the covariance matrix of system state process noise in real-time and integrates it into the UKF algorithm to reduce the uncertainty of state prediction and measurement filtering, thus proposing a robust DSE method for medium voltage distribution networks based on the improved adaptive UKF algorithm. Finally, based on the constructed 15-buses medium voltage distribution network, simulation results show that the proposed method can effectively perform DSE on the medium voltage distribution network and obtain more accurate situational awareness information.