In order to quickly and effectively diagnose the metal corrosion state of power equipment，the corrosion image of the metal material Q235 steel in the hanging piece experiment of the substation was analyzed，and a metal corrosion state evaluation method based on the analysis of the corrosion image characteristics was proposed. First，multi-dimensional feature parameters were extracted through image preprocessing，chromatics，statistics， wavelet and fractal analysis methods; then，a metal corrosion state evaluation method was proposed based on neural network algorithm，and the effectiveness of the method was verified. The results show that the color，statistics， wavelet and fractal characteristic parameters of the corrosion image can fully reflect the evolution law and corrosion state of the metal corrosion morphology，and the corrosion evaluation model constructed by the neural networkalgorithm and the multi-dimensional characteristic parameters can accurately evaluate the metal corrosion degree. The evaluation results of the corrosion status of different corrosion samples on site are consistent with their actual corrosion degree.