Coupling-Cascaded Neural Operator for Coupling-Aware Multiphysics Surrogate Modeling in Energy Systems: Data, Code, and Reproducibility Package

This repository contains the data, source code, trained model outputs, figure-generation scripts, and reproducibility materials associated with the manuscript “Coupling-Cascaded Neural Operator for Coupling-Aware Multiphysics Surrogate Modeling in Energy Systems”. The study develops a Coupling-Cascaded Neural Operator (CCNO) for multiphysics surrogate modeling in energy systems. CCNO embeds a physical coupling graph into a neural-operator architecture by separating a strongly coupled field pair from a one-way downstream field and uses Spectral Cross-Attention (SCA) for low-frequency cross-field message exchange. The archive includes processed datasets, model checkpoints, training histories, evaluation summaries, physics-consistency diagnostics, cross-fidelity transfer results, PEMFC cross-domain validation outputs, EBR-II SHRT-17 steady-state anchor outputs, Pareto pre-screening results, source code, and scripts used to generate the reported figures. This package is intended to support reproduction of the main numerical results, tables, and figures reported in the manuscript. The manuscript text, submission files, cover letter, and journal-specific materials are not included in this archive. 20:19