Anomaly detection method for power system information based on multimodal data.

With the increasing complexity of modern power systems, effective anomaly detection is essential to ensure operational security. Conventional methods often depend on single-domain data, which limits their ability to fully capture the dynamic behavior of power systems. This study introduces a novel multimodal approach that integrates time-domain and frequency-domain data to improve anomaly detection accuracy and robustness. By leveraging this integration, our method captures both temporal patterns and spectral signatures, offering a more comprehensive analysis of system behavior-an advancement that significantly enhances detection performance compared to traditional techniques. Experimental results show that our approach achieves a detection accuracy of 97.6%, outperforming baseline methods. Beyond its technical merits, this method has practical implications for real-world power systems, enabling early identification of security threats, improving system reliability, and reducing the risk of operational failures. These findings contribute to the field of power system security and provide a versatile framework for anomaly detection in critical infrastructures.