Multi-node abnormal data identification in distribution networks based on deep learning

Abstract

With the continuous advancement of the construction of the power grid industry, the lean management of the distribution network is also being fully implemented. Due to unstable communications, system defects, equipment failures, and human factors, data center problems such as missing data and abnormal data collection in the process of power grid deployment and deployment are becoming more and more prominent. To improve the quality and availability of abnormal data in transmission networks, this paper proposes a deep learning-based method for identifying abnormal data of multiple nodes in distribution networks. The dual-channel deep CNN network is used to extract the time series power data features, and the abnormal data is corrected by redistributing the power loss of the transmission line by using the balance calculation relationship. Experiments show that the model proposed in this paper improves the detection rate of abnormal data identification and reduces the false alarm rate compared with traditional machine learning and statistical analysis detection methods. In addition, the abnormal data of the grid is checked and corrected to improve the availability of grid data.