This paper designs a regional distribution network state estimator based on Bayesian theory to solve the difficulties in the distribution network state estimation caused by the randomness in the output of the distributed generations accessing to the regional distribution network and the limit of the number of measuring devices. Firstly, the distributed network voltage estimation model is established based on the back/forward sweep power flow calculation method, then the state estimator based on the Bayesian theory is proposed, the impacts of changing the positions and using the information of different measuring devices on the voltage estimation accuracy are theoretically analyzed. The correctness and effectiveness of the proposed method is verified by the simulation results on the IEEE 33 nodes system. The influences of the variance of the node injection power and the positions of the measuring devices on the estimation accuracy are analyzed in further simulations, and the optimal positioning of the measuring devices are also Provided. This paper provides a theoretical and practical method for the state estimation of regional distribution network with high permeability of distributed generations accessed.