The evaluation of voltage sag of power grid is to analyze the situation of voltage sag, to guide the planning, construction and transformation of power grid, and further improve the power quality. It is an important content in the current power quality research. In this paper, the fault point is regarded as a new virtual node, and the fault location is described by fault location parameters λ. The system node impedance matrix is expanded by calculating the self impedance and mutual impedance between the virtual node and the concerned node. Through the modeling and simulation of distribution network, short-circuit faults are set to analyze the voltage sag. According to the three-phase voltage sag waveform recorded by power quality monitoring system (PQMS), the fault type can be identified, and the fault line can be determined according to the PDR of SCADA system. Without considering the fault resistance, the fault type and fault line are known, so the fault location can be found only by solving the fault location parameters λ. Because of its parallel processing and good robustness, particle swarm optimization algorithm can find the optimal solution of the problem with large probability and fast convergence speed, so this paper chooses to use particle swarm optimization algorithm to solve the fault location parameters λ.