A Validated BIM–Deep Learning Pipeline for Surrogate-Assisted Multi-Objective Optimization of Prefabricated Low-Carbon Wall Panels

This repository contains a package supporting the manuscript: "A Validated BIM–Deep Learning Pipeline for Surrogate-Assisted Multi-Objective Optimization of Prefabricated Low-Carbon Wall Panels." The original BIM-generated dataset and certain project artifacts were unavailable at the time of repository preparation. To support transparency and reproducibility, this repository provides: • A reconstructed synthetic dataset containing 5,200 samples. • BIM workflow documentation derived from the published methodology. • Machine-learning surrogate training scripts. • NSGA-II optimization scripts. • Supplementary tables. • Reconstructed model files. • Reproducibility documentation. All reconstructed artifacts are clearly labeled and documented. Files that were not recoverable from the original study have been reconstructed using the methodology, equations, constraints, and parameters reported in the manuscript.