In consideration of the problem that equal salt deposit density of insulator is complexly influenced by climate factors and accurate model is difficult to construct

a novel prediction model of insulator’s ESDD under different climate conditions is proposed based on least square support vector machine（LS-SVM）.With five main climate factors including temperature

humidity

wind velocity

and so on as inputs

ESDD of insulator as output

the nonlinear mapping between input and output is fitted through LS-SVM.Collecting and processing field data as learning samples to train the model

the insulator’s ESDD under certain climate condition is hence predicted by the trained model.Experimental results demonstrate that the model based on LS-SVM is constructed more rapidly than the standard SVM-based model

and its prediction error is smaller.Moreover

the present model gains better prediction accuracy and speed compared with the BP model.