As more and more devices sensitive to voltage sag are widely connected to the power grid, the direct and indirect economic losses caused by voltage sag interference become more and more serious, which puts forward higher requirements for power supply quality. Accurate identification of sag sources is an essential step in the treatment of voltage sag.It analyzes the types of voltage sags caused by various short-circuit faults and the changes of voltage sags waveforms after transformer transmission, and establishes the standard sample waveforms of voltage sags according to theoretical analysis. A voltage sag waveform matching method based on mutual approximation entropy principle is proposed. The measured waveforms are compared with the sample waveforms and the similarity matching is carried out directly. Finally, the effect of identifying voltage sag disturbance source is achieved. The simulation results show that the method has high accuracy in identifying the measured voltage sag waveforms in power grid, and the feasibility and effectiveness of the method are verified.