A Remote Out-of-Tolerance Classification Method of Gateway Electric Energy Meters Based on Feature Selection and Deep Belief Network

Abstract

Accurate identification of the out-of-tolerance of the gateway electric energy meter plays a vital role in ensuring the stability of the new power system and maintaining the fairness of the transaction. As a key metering device in power generation enterprises, network connections between power grids, inter-provincial network connections, and other metering devices, the importance of checkpoint energy measurement is self-evident. Therefore, studying the causes of metering errors and the classification of outlier checkpoint metering devices is significant. A remote out-of-tolerance gateway electric energy meters classification method based on feature selection and deep belief network is proposed. This paper first uses Spearman and Pearson correlation coefficients to select the main influencing factors of checkpoint energy metering error and then trains a deep belief network model to complete the detection of outlier checkpoint energy metering. Finally, experimental results verify the superiority of the proposed method.