It is of significance to detect the insulation defect of electrical equipment accurately and warn of dangerous discharge. And it is an important target and key technical problem for the electric power department to carry out the state operation and maintenance. In this paper

the oil wedge discharge model is used to simulate the partial discharge defects between the winding turns and the intercake of the power transformer. It is used ultra high frequency（UHF）and high frequency current（HF）methods to detect the signals in real time under constant voltage. The trend and regularity of the amplitude

frequency

entropy of the signals and PRPD pattern are studied. The consistency and difference of signal characteristics of HF and UHF are analyzed. The results show that the oil wedge discharge has stable development stage and accelerated deterioration stage under constant voltage

and at the same time

the rationality of two discharge stages is analyzed from the angle of discharge mechanism.The frequency of the signals in the period of stationary development is slowly increasing

and the statistical parameters of signals amplitude are certain fluctuation.In the accelerated deterioration stage

the frequency of the signals have obvious inflection point and the characteristic of violent concussion. The characteristic parameters of HF and UHF have obvious rise or decrease trend

and the information of the two can realize the accurate discrimination of stable development state and accelerated deterioration condition.The research results of this paper provide the test basis for the status detection and hazard warning.