Abstract: To solve the problem of small sample imbalance in state estimation of active distribution networks (ADNs), this paper proposes a robust forecasting-aided state estimation (FASE) method based on the improved synthetic minority oversampling technique (SMOTE) and particle filter (PF) of Prophet. The method enables state estimation of ADNs. Firstly, to handle the small-sample imbalance problem, a hash function is constructed based on the data features of the ADN and an optimization approach is proposed using the hash function for the Borderline-SMOTE+Tomek-Links algorithm. Secondly, considering the large amount of data and the stochastic output of distributed energy resources in ADNs, the Prophet prediction model is used for state estimation of ADNs, and a robust FASE method based on Prophet-PF is proposed for fast and accurate estimation of ADNs states. Finally, numerical simulations are conducted on standard IEEE 118-bus distribution network and a DTU 7k 47 distribution system to evaluate the proposed method. The results demonstrate that the proposed method has high accuracy and robustness, providing useful references for state estimation in ADNs.