Dataset of Standard Re-entrant Hybrid Flow Shop Scheduling for New-energy Battery Substrate Manufacturing

This benchmark dataset is curated standardized re-entrant hybrid flow shop scheduling instances reconstructed from desensitized real engineering data from a leading new-energy battery substrate manufacturer in China. Existing public re-entrant scheduling benchmarks mainly focus on semiconductor wafer production, while standardized test datasets targeting long-cycle batch manufacturing of battery substrates are rarely available, which this dataset fills. The dataset covers a complete seven-operation production line, with the 4th annealing stage defined as the bottleneck workstation. The parallel machine configuration for each operation is set as (2, 2, 2, 3, 1, 2, 1), as listed in Table A.1. Fixed batch setup time and fixed pure processing time for each non-annealing operation are provided; processing time and setup time equal zero for missing operations in reentrant routes. Four product families F1–F4 with unique annealing durations (7.6 h, 8.0 h, 8.4 h, 8.8 h) are defined in Table A.2, together with 12 standardized reentrant paths covering 0 to 3 rolling-annealing circulation layers, represented by (j,l) operation pairs. A sequence-dependent family changeover time matrix exclusive to annealing machines is supplied in Table A.3. Five multi-scale instance groups G1–G5 are constructed, corresponding to 10, 20, 30, 40 and 50 batches respectively. Each batch stands for one coil group containing three raw coils. Table A.4 records full detailed job parameters of the smallest instance G1, including job ID, product family, path, due date and sequence of non-zero processing durations. Table A.5 lists the quantity of four families and uniform random due date intervals for each group, and Table A.6 provides the allocation count of 12 paths for G2 to G5 to support reproducible expansion of large-scale instances. All tabular data are stored in editable Word format for direct simulation coding and intelligent scheduling algorithm verification. This dataset incorporates multiple practical industrial constraints including fixed batch capacity, multi-layer reentry and family-based annealing changeover, which can be used to evaluate hybrid flow shop models and metaheuristic optimization algorithms. This dataset is co-submitted alongside the related research article published in Journal of Industrial Information Integration. By releasing the standardized instances publicly, this work provides shared test resources for research on new-energy material production scheduling and industrial information integration systems.本基准数据集由中国头部新能源电池基板制造商脱敏后的真实工程数据重构而来，是经过标准化整理的可重入混合流水车间调度实例集。现有公开的可重入调度基准数据集主要聚焦于半导体晶圆生产，而针对电池基板长周期批量制造的标准化测试数据集则较为稀缺，本数据集填补了这一研究空白。 该数据集覆盖完整的七工位生产线，其中第4道退火工序被定义为瓶颈工作站。各工序的并行机配置为(2, 2, 2, 3, 1, 2, 1)，详见表A.1。非退火工序均设置固定的批量准备时间与固定纯加工时间；可重入路径中缺失的工序，其加工时间与准备时间均设为0。表A.2中定义了4个产品族F1–F4，各自对应的退火时长分别为7.6 h、8.0 h、8.4 h、8.8 h；同时涵盖12条标准化可重入路径，覆盖0至3层循环退火工序，以(j,l)工序对形式表示。表A.3提供了专属于退火机的、依赖序列的产品族切换时间矩阵。 本次研究构建了5个多尺度实例组G1–G5，分别对应10、20、30、40和50个批量。每个批量代表包含3个原卷的卷组。表A.4记录了最小实例G1的完整作业参数细节，包括作业ID、产品族、路径、交货期以及非零加工时长的工序顺序。表A.5列出了各实例组的4个产品族数量分布与统一随机生成的交货期区间；表A.6则给出了G2至G5的12条路径的分配计数，以支持大规模实例的可复现扩展。 所有表格数据均以可编辑的Word格式存储，可直接用于仿真编程与智能调度算法验证。本数据集整合了多项实际工业约束，包括固定批量容量、多层可重入流程以及基于产品族的退火工序切换，可用于评估混合流水车间模型与元启发式优化算法。 本数据集与发表于《Journal of Industrial Information Integration》的相关研究论文一同提交。通过公开发布该标准化实例集，本工作为新能源材料生产调度与工业信息集成系统的研究提供了共享测试资源。