In order to achieve the remote, cost-effective, and efficient evaluation of the metering performance of DC charging piles, a DNN model for calculating accumulated electric energy of DC charging piles is established based on deep neural network (DNN) and a large amount of data of on-site electric vehicles charging on DC charging piles. A method of remote metering performance verification for DC charging piles is proposed, and the correlation between variables and accumulated electric energy during charging is analyzed. The instance verification results show that the state of charge has the greatest influence on the calculation of accumulated electric energy, and the current has the least influence. The established DNN model can accurately calculate the “actual” output energy of the pile to be measured, and the absolute value of the difference between the indication error of model calculation result and the actual verification indication error is less than 1%. Therefore, the proposed method can achieve an efficient metering performance evaluation.