Abstract: DC bias of large power transformer will cause core vibration,which seriously threatens the safe operation of power system. At present,there are still some problems in DC bias monitoring of transformers. The index commonly used to measure DC bias intensity of transformer is vibration spectrum complexity,but the spectrum complexity will increase with the magnetic saturation of transformer core itself,even if there is no DC bias,the spectrum complexity will increase,so it will be disturbed by the magnetic saturation of the iron core. In order to effectively detect and evaluate the DC magnetic bias degree of transformer,a 3-dimensional transient calculation model of transformer electric-magnetic-mechanical multi-physical field is established,and the magnetic bias vibration signals of each detection point of transformer core under different DC bias coefficients and different excitation voltages are analyzed. According to the characteristics of DC bias vibration signal,an analysis method of DC bias severity of transformer based on time domain integration method is proposed. In this method,the energy difference between the positive and negative half cycles of the vibration signal in a single cycle caused by DC bias is used to integrate the transformer vibration signal in time domain,and the characteristic parameters λ of each point with different DC bias coefficients are calculated. It shows that the simulation characteristic quantity λ increases with the increase of DC bias coefficient,which can track and reflect the severity of transformer bias. The excitation voltage will not interfere with the characteristic quantity λ,and the interference of magnetic saturation of transformer core can be eliminated. The research results provide effective theoretical guidance for monitoring and controlling DC bias of transformers.