"Aspen Airport Volt--VAR Curve Cyberattack and Adversarial Dataset for Robust Smart-Inverter Attack Detection"

"This data article presents three companion datasets for cyberattack detection on smart-inverter Volt-VAR curve breakpoint settings: (i) labeled stealthy attack samples generated by bounded perturbations of the six per-unit voltage breakpoints (v1-v6), (ii) matched labeled non-attack (benign) samples, and (iii) adversarial breakpoint samples crafted to challenge learned detectors. Data are produced using a distribution-system simulation workflow (OpenDSS accessed through Python) on an Aspen Airport feeder model, where breakpoint vectors are generated under strict monotonicity and value-range constraints to emulate realistic command-level setpoint tampering without obviously invalid curves. Each record contains the breakpoint vector and a binary label (attack\/non-attack), enabling reproducible benchmarking of detection models that operate directly on configuration-layer parameters rather than only downstream measurements. In addition to the raw CSV files, we provide reference scripts and notebooks that reproduce the generator, train multiple AI baselines (tree-based and deep models, including a tabular residual network), and export adversarial samples for robustness evaluation. Together, these resources support research on resilient monitoring, adversarial training, and secure DER interoperability workflows."