Abstract: In order to improve the operational reliability of electric vehicle charging piles under typhoon weather, an improved charging pile fault prediction method integrating typhoon influence factors is proposed. First, the charging pile data is preprocessed. Secondly, in view of the insufficient prediction performance of the traditional C4.5 decision tree algorithm under the feature coupling effect, meteorological attributes are introduced and the feature selection strategy is optimized based on the information gain ratio to reduce the interference of inter-attribute correlation on classification accuracy. Finally, the analysis of the example results shows that the accuracy of this algorithm is improved by 21% compared with the traditional algorithm, which verifies the optimization effect of the introduction of typhoon influence factors on the performance of the prediction model.

Key words: electric vehicle charging pile, decision tree, typhoon impact, fault prediction