In order to improve the electric car battery safety early warning ability and decrease the safe accident of electric vehicle charging, the electric vehicle charging safety factors are analyzed, the security early warning index selection ways and methods are studied, and a security early warning evaluation index system is set, the evaluation method of weight for safety index is discussed. Besides, a kind of electric vehicle charging security early warning model is designed using genetic wavelet neural network training principles with multi-scale multi-resolution characteristics, and five levels of security early warning are classified. Through simulation analysis, early warning system for a charging station safety data is used to verify the proposed electric vehicle charging correctness and feasibility of the early warning model.