Under the action of strong electric field

the streamers in transformer oil can get step development at a specific threshold voltage and induce insulation failure in the large oil gap in the actual operation of transformer. However

the presently used common ‘4 modes’ classification method can only reflect the variation extent of average development speed in case of abrupt variation of the streamer discharge

but cannot effectively distinguish the abrupt variation pattern and mechanism of streamers in transformer oil with different physical and chemical characteristics. In this paper

the gas paraffin oil and natural ester commonly used in transformers is used as the study object

the streamer discharge pattern and the accompanying photoelectric signal in a 25 mm point-plane electrode structure are measured under the standard lightning impulse voltage. Based on the consideration of traditional parameter of the average development speed of the streamer discharge

such two types of characteristic components as the fractal dimension of discharge and amplitude of photoelectric signal are introduced to perform a three-dimensional clustering analysis of the streamer discharge development. The study results show that the improved classification of the streamer propagation mode

compared with the ‘4 types of mode’ classification method

has more obvious effect on the differentiation of the 1st and 2nd mode slow streamers and the 3rd and 4th mode fast streamers. Furthermore

the understanding of the variation of the streamer discharge pattern in case of voltage variation and the influence mechanism of the variation trigger as well as the streamer pattern jump phenomenon are strengthened.