Hybrid Physics-Based and Data-Driven Mitigation Strategy for Automatic Generation Control Under Cyber Attack

The fusion of information, data, control and electric power facilitates the electric cyber-physical system (ECPS). In the ECPS, though the high-level cyber-physical interaction and integration increase the flexibility and efficiency of power system operation, the accompanied cyber perils gradually endanger the system security. In fact, several energy sector-targeted attacks, including the infamous Ukraine power grid hack, have shown the power system vulnerability in cybersecurity incidents. Known as the critical power-balancing operation in real-time control systems, automatic generation control (AGC) is a typical ECPS application. Measurements from remote sensors may be manipulated by attackers when telemetered to the AGC center, thereby disrupting the balance of power and frequency stability. In this situation, cyber-attack- tolerant AGC plays an important role in the face of cybersecurity threats. To achieve the cyber-attack-tolerant AGC, some strategies including the physics-based and data-driven attack mitigation schemes have been employed. In this chapter, inspired by the fault detection, diagnosis and reconfiguration in fault-tolerant control theories, a hybrid physics-based and data-driven mitigation model is developed for AGC under false data injection attacks (FDIAs). A mathematical model is derived to reveal the causal relation between the FDIA signal and compromised AGC measurement data. Then, data-driven approaches are employed to establish the mapping between the compromised measurement data and power compensation. Finally, the compensation-based mitigation model is developed.