Supplementary information for:
Digital Twin and Artificial Intelligence Enabled Industrial Internet of Things for Real-Time Predictive Control: A Simulation Benchmark Study

Contents
- simulation_timeseries.csv: archived 500-step closed-loop outputs and recorded AI-MPC execution times.
- performance_metrics.csv: reported RMSE, IAE, control total variation, and actuator saturation results.
- training_data.npz: 3,600 training samples (X) and corresponding one-step residual targets (Y) used for the residual-model stage.
- verify_results.py: deterministic verification script that recomputes the reported closed-loop metrics from the archived outputs.
- requirements.txt: Python packages needed for verification.

The study is simulation-only. No human subjects, confidential industrial data, or undisclosed physical measurements were used.

Important reproducibility note: wall-clock execution times are environment-dependent. The archived timing trace documents the benchmark run reported in the manuscript; timing should not be treated as hardware-independent latency.
