Abstract: The loss rate is an important indicator to measure the technical operation level and economic

operation level of the power system. In response to the actual needs of the distribution network in management and

technical loss reduction，based on loss statistics and theoretical calculations，a short-term loss characteristic

characterization method based on Gaussian mixture model was proposed. The expectation-maximization algorithm

was adopted to calculate the parameters both of the normal days and abnormal days of loss data，and it utilized the distribution extremes of the loss data，the calculation process was simplified and accuracy of the iterative algorithm was improved. Through the comparison of the obtained mixed distribution parameters，the loss abnormal phenomena would be identification and location，and it is helpful to location of loss abnormal place and formulation of loss reduction measures.

      Key words: distribution network；management loss reduction；technical loss reduction；Gaussian mixture

model（GMM）；expectation-maximization（EM）algorithm