Abstract: The state estimation of complex distribution networks is the most basic function of power grid intelligent operation. The state estimation algorithm of transmission networks usually cannot be directly applied to the distribution networks, it is needed to provide accurate and efficient state estimation in the face of the basic data of distribution networks with different qualities on the condition of power grid operation states. The complex distribution network considered in this paper is modelled according to the characteristics of three-phase unbalance. A state estimation method based on load classification and complex network pruning is presented in detail, the problems of lack of measured data in the distribution network is better solved. Through the field test in an actual distribution network, it is proved that the proposed algorithm is robust, accurate and efficient, and can meet the actual application needs.