Iwind Dataset for Intelligent Simulation

The archive includes original simulation inputs and outputs, a complete representative case, processed stress and optimal pitch-angle results, visualization outputs, post-processing scripts, execution records, and metadata supporting data verification and traceability.The archive is distributed using lossless compression and requires approximately 73 GB of disk space after extraction.

In combination with the code, runnable Zwind environment, and reproduction instructions provided in the public GitHub repository, these data support independent reproduction and verification of the core natural-language-driven simulation workflow and the associated engineering analyses reported in this study.Following formal acceptance of the manuscript, the temporary review-period access restriction will be removed, and the deposited dataset will be made publicly accessible through Science Data Bank in accordance with the journal’s data-availability requirements.