It is always difficult to timely locate the location of the network attack and achieve rapid deployment of defense strategies when the smart grid is attacked by the network.

In order to solve this problem, this article proposes a Convolutional Neural Network (CNN) model that integrates Convolutional Block Attention Modules (CBAM) (CNN-CBAM) to detect False Data Injection Attack (FDIA) positions. The attack identification problem of FDIA is modeled as a multi label classification problem, where CNN is used to extract spatial features of the data. The CBAM module can be directly integrated into the convolution operation of the CNN module, which not only focuses on important parameter information from the perspective of spatial domain, but also considers feature relationships in the channel domain, and allocates attention to the input data from two dimensions to improve the performance of the model.

The performance of the proposed CNN-CBAM network FDIA position detection model is verified on Institute of Electrical and Electronics Engineers (IEEE) 14 and IEEE118 node systems. The experimental results show that the FDIA position detection rates of CNN-CBAM on IEEE14 and IEEE118 node systems are 98.25%and 96.72%, respectively.

Compared with other methods, the CNN-CBAM network model proposed in this paper can effectively extract the spatiotemporal characteristics between data, with improved existence of FDIA. It also im-proves the accuracy of attack location identification with better robustness.