In order to realize the early warning and accurate positioning for the detection of inter-turn short circuit fault by recurrent surge oscilloscope（RSO） method

the improved minimum energy method is presented in this paper to determine the start time of the recurrent surge pulse and estimate signals propagating time difference so as to improve the accuracy of fault location. The detection in generator rotor winding is taken as an analysis example. In the four cases that there are no defects

slight defects

severe defects and serious defects in windings

the influence of the minimum energy factor α on the accuracy of wave front estimation is analyzed. The result indicates that the larger the protruding of the characteristic waveform

in other words the higher the signal to noise ratio

the more accurate the time difference calculation will be. Then thirty sets of RSO data collected in the test of different cases are statistically analyzed based on the Monte Carlo method. The results show that no matter what kind of fault is

the factor α always has a reasonable selection range which can improve the calculation accuracy of the wave front as well as the defect location of the inter-turn shortcircuit. Finally

the defect location method that judges the severity of the defect firstly

selects the appropriate computing factor secondly and calculates of time difference at last

is proposed.