Localization Detection of False Data Injection Attacks in Novel Energy and Power Systems Based on Correlated Feature-Multi-Label Cascading Boosted Forests

Abstract

Under the dual influence of power system transition to integrated energy and the evolution of cyberattack technology, a correlation feature-multilabel cascade boosted forest based false data injection attack localization and detection method is proposed for the new energy power system to accurately locate the attacked position of the power grid in response to the stealthy false data injection attack (FDIA). Considering the FDIA principle and characteristics of the new energy power system, as well as the fact that the new energy power system contains a large amount of measurement data and variable operation states, the proposed method enhances the fitting ability of multi-label cascade forests to complex power measurement data by incorporating the extreme gradient boosting algorithm to identify the anomalies of state quantities of each node of the system, and introduces the “correlation feature” algorithm to detect the original power measurement data. The “correlation feature” algorithm is introduced to extract highly informative features from the original power measurement data to enhance the generalization ability of the multi-label cascade forest, so as to obtain more accurate localization detection. Simulation tests are conducted in the IEEE-57 node system to verify the effectiveness of the proposed method, and compared with other methods, the proposed method has better accuracy, detection rate, sensitivity, and F1 score.