FiLM-PPO for Requirement-Aware Airfoil Optimization

This dataset contains curated source data and compact processed supporting data for the manuscript "A FiLM-Modulated Conditional PPO Framework for Requirement-Aware Airfoil Optimization". The data support the reported results on training convergence, target-lift tracking, L/D-weight response, FiLM-vs-concatenation ablation, few-shot adaptation to unseen design requirements, airfoil-geometry comparison, and Pareto-front quality analysis. The package includes figure source data, table source data, processed supporting CSV/JSON files, a file manifest, and metadata mapping each file to the corresponding manuscript figure or table. Large intermediate training and search caches were excluded when smaller figure/table source files preserve the values used in the manuscript. The associated code is available at https://github.com/dyll123/FiLM-PPO-airfoil-optimaiztion.本数据集为论文《面向需求感知翼型优化的FiLM调制条件近端策略优化框架》（"A FiLM-Modulated Conditional PPO Framework for Requirement-Aware Airfoil Optimization"）整理并提供了原始源数据与精简处理后的辅助支撑数据。本数据集可支撑论文中报告的多项实验结果，涵盖训练收敛性、目标升力跟踪、升阻比（L/D）权重响应、FiLM与拼接操作的消融对比、针对未知设计需求的少样本适配、翼型几何对比以及帕累托前沿质量分析。该数据集包包含图片源数据、表格源数据、经处理的辅助CSV/JSON文件、文件清单，以及将每个文件映射至对应论文图表的元数据。鉴于体积更小的图表源文件已留存论文中使用的核心数值，本数据集未包含大型训练中间缓存与搜索缓存。相关代码可在 https://github.com/dyll123/FiLM-PPO-airfoil-optimaiztion 获取。