The damp of contamination on an insulator surface is one of key factors leading to flashover. It is of great importance to understand the moisture condition of contamination for the sake of assessing the insulation status of insulators. In order to solve the problem that it is difficult to measure the moisture condition of an insulator surface directly

in this paper

a nondestructive and non-contact hyperspectral technique is used to obtain the spectral information of the surface contamination

and then to detect moisture content in contamination quantitatively. Firstly

the spectrum of damp contamination was got with a hyperspectral imager

and the collected spectral lines were pretreated by black-white correction and multiplicative scatter correction（MSC） to reduce the effect of noise and scattering. Secondly

the successive projections algorithm（SPA） was used to extract the feature bands of pretreated spectral lines. Then

a back propagation neural network（BPNN） prediction model was established based on all bands and feature bands

in which the mean square root error of the prediction is 0.110 1 and 0.076 8

respectively. The results show that hyperspectral technology can be used for non-contact moisture detection of the contamination on the insulator surface. In addition

the model of using feature bands possesses the superiority of high recognition rate and less amount of computation compared with that of using all bands. The method is applied to the moisture detection of contamination on the surface of naturally contaminant insulators

and the result shows the model is applicable. This paper provides a new method for the online measuring of insulation status.