Quantile-boosted Attention Network Model for Dynamic Flexible Flow Shop Scheduling with Safety Control

Description

This dataset contains 32 test instances for the Dynamic Flexible Flow Shop Scheduling with Maximum Safety Risk Index (DFFSS-MSRI) problem, as described in the accompanying paper. File naming convention: {scale}_s{stages}_{index}.txt - scale ∈ {tiny, small, medium, large} (10, 20, 40, 100 jobs respectively) - stages ∈ {3,4,5,6} (number of processing stages per job) - index ∈ {1,2} (two instances per combination) Each file is a plain text file formatted as follows: - Lines starting with '#' are comments and provide metadata (number of jobs, stages, machines per stage, parameter ranges). - Data lines (after the header) follow the format: JobID, ReleaseTime, Stage1_Machine1_ProcTime, Stage1_Machine1_Risk, Stage1_Machine2_ProcTime, Stage1_Machine2_Risk, ..., StageK_MachineM_ProcTime, StageK_MachineM_Risk The number of machines per stage is given in the comment line (e.g., "# 各阶段机器数: [3, 3, 2]" for small_s3_1.txt). Processing times and risk values are integers. All instances are generated randomly according to the rules specified in Section 5 (Table 4) of the associated paper. The generation code is available from the corresponding author upon reasonable request. This dataset supports the experimental results reported in the paper. For details on the scheduling model and algorithms, please refer to the publication.

Steps to reproduce

The instances are generated by the following procedure (detailed in the paper's Section 5): 1. Number of jobs N ∈ {10,20,40,100} 2. Number of stages K ∈ {3,4,5,6} 3. Number of machines per stage m_k randomly selected from [2,4] 4. Processing time p_{ikl} ~ Uniform(3,10) 5. Safety risk risk_{ikl} ~ Uniform(2,8) 6. Release times r_i generated with increasing trend to simulate dynamic job arrivals The exact generation code is available upon request.

Institutions

• University of Science and Technology Beijing

  Beijing, Beijing

Categories

Licence