Abstract. With the increase of voltage grade in power transmission system and the aggravation of industrial pollution, flashover of contaminated insulators is getting more and more serious in high-voltage (HV) and ultra-high-voltage (UHV) systems, this threatens the security of power transmission severely. The key to improve the security of power system is to explore methods of monitoring states of insulator's surface by some eigenvalues. A new idea, using chirplet as the ruler to define fractal dimension, was proposed in this paper. Chirplet-fractal dimension is defined as the sum of residues of decomposed signals in formula 8. The fractal characteristics of leakage current (LC) are gotten from this dimension. The results show in figure 11, which the chirpletfractal dimension efficiently describes the information of arcs discharge in LC, and it is also a good eigenvalue for flashover discrimination and risk prediction.