Research on Operation and Anomaly Detection of Smart Power Grid Based on Information Technology Using CNN+Bidirectional LSTM

Abstract

The accurate detection of abnormal users in the grid is conducive to maintaining the stability of the smart grid. This paper briefly introduces the smart power grid and the intelligent algorithm used to detect users with abnormal power consumption in the power grid. The intelligent algorithm combined the bidirectional long short-term memory (LSTM) and a convolutional neural network (CNN) to extract the features from the power consumption data of the users and then used the adaptive boosting (AdaBoost) model to classify the users. The field operation test was carried out in a small substation. The proposed method was compared with the single bidirectional LSTM and CNN methods. The findings showed that the proposed method had the best performance in the simulation experiment, with a precision of 98.7%, a recall rate of 97.9%, and a false drop rate of 3.6%, and its receiver operator characteristic (ROC) curve deviated the most from the diagonal line and had the largest area enclosed. In the field operation test, the proposed method obtained a lower and more stable false detection rate (approximately 3.6%).