Traditional methods to determine whether the work of electric vehicle charging equipment is abnormal is limited to in-lab confirmation and on-site the abnormal over- limit attribute of single threshold, which have such problems as long time, low quality, easy development of defects, and easy to make mistakes. Breakthrough the limitation of laboratory and field conditions, in this paper, an effective abnormal prediction method for electric vehicle charging equipment is established by comparing various machine learning technologies and combining with the working mechanism of EV charging equipment components based on multi-scale massive data accumulated in time and space during the operation of EV charging equipment. The results show that the XGboost algorithm has a high accuracy in the abnormal prediction of EV charging equipment components and their characteristics, and the accuracy of anomaly prediction reaches 90.34%, which has an absolute advantage.