Abstract：The number of network attacks targeting the power system is increasing, and information physical security issues have attracted high attention from power companies and academia. In order to accurately detect false data injection attacks in the power grid, a one-dimensional convolutional neural network (1DCNN) based on residual neural network (ResNet) structure, and long short-term memory (LSTM) network based multi-channel fusion network model which called Res- CNN-LSTM is proposed. This algorithm utilizes the efficient extraction ability of 1DCNN and LSTM in time series information, and fuses the extracted information in different channels to further enhance the extraction effect of data features. At the same time, the main body of the model adopts a residual jump connection structure to solve the problem of overfitting in the training process of the neural network. Simulation is conducted based on IEEE-14 and IEEE-118 node testing systems, and the proposed method is compared with other neural network model algorithms. The results verified the effectiveness and accuracy of the proposed method in the paper.
方正刚. 基于通道融合的Res-CNN-LSTM电网虚假数据注入攻击检测[J]. 电气技术, 2024, 25(3): 11-17.
FANG Zhenggang. Detection of false data injection attacks in power grid based on Res-CNN-LSTM with channel fusion. Electrical Engineering, 2024, 25(3): 11-17.
2024, Vol. 25 
