This paper proposes a new algorithm to evaluate the voltage sag in unbalanced three-phase distribution network with distributed generations DG . The paper models fault characteristics (fault line, location, type, duration, impedance) and DG output. By using two point estimation method (2PEM), the random probability problem in voltage sag simulation is transformed into multiple deterministic problems. Through computing the mean, the standard deviation and the third central moment of each node in distribution network, the probability density functions (PDF) of each node can be got based on Cornish-Fisher expansion (CFE) series, and then the voltage sag index System Average Root Mean Square Variation Frequency Index (SARFI) can be computed. Weak branches in distribution network are analyzed. Experiments have been made on the IEEE 13-bus test case with the consideration of two stochastic DG units. Thirteen scenarios simulations about voltage sag and comparison with Monte Carlo Simulation have been made. The comparisons with different wind turbine models have been provided. The feasibility and effectiveness of the proposed algorithm for voltage sag evaluation have been proved.

No abstract