To monitor the pollution level of the insulator surface in time for EHV and UHV power grids and accurately evaluate the potential harm of pollution flashover

voltage was applied upon a ZS-35 insulator to extract the amplitude

RMS value and equivalent pulses of leakage current as well as the environmental parameter relative humidity as input parameters of neural network

so a pollution level evaluation model of polluted insulator and a flashover warning model were established based on radial basis function neural network. The results of the models coincide well with the data measured experimentally

verifying the accuracy of the models.