Enhancing the Resilience of Key Equipment to False Data Injection Attacks in NPPs

摘要

Current cybersecurity efforts may be not adequate for defending the NPPs from cyber-attacks. In addition to network data, process data is used as an indication of compromise. Most cybersecurity research based on process data is I&C system level, which may be not effective in the situation that the attacker compromised some sensor measurements or make it inaccessible since the attack surface is large. To enhance the resilience of the key equipment in NPPs, a localized cybersecurity strategy was proposed to be implemented into the controller itself, which significantly reduces the attack surface minimizing the possible number of compromised sensors. The proposed strategy can detect the fault state caused by false data injection attacks. Once the anomaly is detected, the control variable is inferred by a pre-trained supervised inference model with other state variables to be utilized as a virtual sensor to generate the correct command. This method can aid in the operation of key equipment under some safety events like sensor failure.