Incorporation of Unmanned Aerial Vehicles (UAVs) with multi-sensor payloads for inspection of nuclear facilities injects vital cyber-security risks, explored inadequately in the literature. In this work, a forensic-ready simulation framework is used to assess the effect of cyber-attacks on the UAV-based structural health monitoring of nuclear containment buildings. We assess the effect of ten different attack strategies under three types of sensors, namely GPS spoofing, LiDAR saturation/injection, and Ground Penetrating Radar (GPR) electromagnetic interference, using a physics-based model to capture the cyber-attack effect that is validated against nuclear industry standards (NUREG-1801, ACI 349.3R). Monte Carlo simulations across 20 seeds show that single-vector attacks degrade detection performance by 8–34%, while coordinated multi-vector attacks cause catastrophic degradation of up to 90%. GPS spoofing exhibits non-linear severity scaling: a 70 cm single-vector drift causes 64% F1 degradation, while a 75 cm drift combined with LiDAR and GPR attacks pushes degradation to over 90%, far beyond a linear additive prediction. The proposed framework generates forensic-ready analysis artifacts such as attack-signature timestamp, sensor fusion, and deviation patterns from the optimal trajectory, providing valuable information after the occurrence of events.
