Method for identifying false data injection attacks in power grid based on improved CNN-LSTM

In contemporary power systems, the interaction between informational and physical dimensions significantly increases vulnerability to network attacks, particularly false data injection attacks (FDIAs). These attacks are characterized by their stealth and potential for severe disruption, posing a threat to the stability and security of power grid operations. The complexity and high dimensionality of operational data in power systems further exacerbate computational challenges, leading to reduced accuracy in traditional attack detection models. To address these issues, this paper introduces an improved CNN-LSTM approach for FDIA detection in power grids. It incorporates an attention mechanism in the autoencoder structure for refined feature extraction and utilizes a sparrow search algorithm for optimizing model parameters. Evaluational results show over 95% accuracy rate, demonstrating the effectiveness of the proposed method in diverse power grid scenarios.