Winding deformation is one of transformer faults of the main types, which threatens the normal operation of power system. To effectively increase the accuracy of detecting the winding deformation on-line, combining with the characteristics of transient overvoltage the surge of which the transformer suffers during its operation, an on-line fault location method for transformer winding deformation based on traveling wave analysis with transient overvoltage is proposed. Voltage signal of traveling wave is measured when transient overvoltage invades the transformer winding. Intrinsic mode function(IMF) is obtained through complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN) of the signal. Correlation coefficients of the IMF components are calculated, and some correlation coefficients of IMF are chosen as fault feature of winding deformation through the comparison and analysis. Finally, the BP neural network is used to establish the mapping relationship between fault characteristics and fault locations, and the on-line fault location of winding deformation is achieved. The simulation results show the feasibility of the proposed method.