Abnormal recognition of electric energy metering data at charging stations based on S-transform algorithm

The electric vehicle charging pile should deal with the grid-connected/independent dual operation mode, with obviously high-power and nonlinear load characteristics, high error probability of electric energy metering, and low Youden index of electric energy metering. Therefore, a method of abnormal identification of electric energy metering data of charging piles based on the S-transform algorithm is proposed. Based on Blackman window S-transform (BST) method, the abnormal characteristics of electric energy metering error are extracted, and the effective analysis of abnormal characteristics of electric energy metering data is realized by setting different window width ratios. The abnormal characteristic matrix is constructed, and the complexity of the left one method (loo) is reduced by solving press, and the regularization coefficient of the kernel extreme learning machine (kelm) classifier is quickly optimized, so as to realize the recognition of abnormal data in the electric energy metering of the charging pile. The features extracted by Blackman window transform are input into the optimized loo-kelm classifier to obtain the classification results of abnormal electric energy metering data, so as to realize the accurate identification and analysis of abnormal electric energy metering data of the charging pile. The experimental results show that the proposed method can not only effectively identify various anomalies such as voltage fluctuation anomaly, current harmonic anomaly, power factor anomaly, electricity metering error anomaly and data communication anomaly, but also the Youden index is higher than 0.90, showing high sensitivity and specificity. At the same time, with the increase of electric energy measurement data, the recognition time used in this method is always less than 30 ms, which reflects its accurate recognition ability and excellent recognition efficiency.