A cyber-resilient framework for detection and identification of false data injection attacks in PV plants

The increasing digitalization of renewable energy resources, particularly through Power Plant Controllers (PPCs), has introduced significant cyber-vulnerabilities. False Data Injection Attacks (FDIAs) represent a severe threat to these systems, as they can mislead PPC control actions and result in equipment disconnection or substantial economic losses through unnecessary generation curtailment. While FDIA detection has been extensively studied at the transmission and distribution levels, its application within the specific context of photovoltaic (PV) power plants remains largely unexplored. This paper proposes a novel, systematic framework for the detection and identification of FDIAs in PV plants using a hybrid state estimation approach. The methodology combines an Equality-Constrained Weighted Least Squares Estimator (EC-WLSE) for detecting gross anomalies with an Equality-Constrained Schweppe-Huber Generalized M-Estimator (EC-SHGME) to identify subtle, stealthy manipulations. By leveraging the static topology of PV plants to optimize estimator parameters, the framework addresses realistic, sophisticated attack vectors. Experimental validation demonstrates that the integrated tool achieves high detection accuracy, with F1-scores exceeding 85% even in challenging low-generation scenarios and coordinated multi-measurement attacks, providing a computationally efficient solution for securing modern PV plant operations.