2026-10-14T23:24:07Z https://www.scidb.cn/oai

10.57760/sciencedb.012kv 2026-09-20T10:13:24Z

2026-09-20 Selection information and search opportunities in flexible job-shop scheduling with automated guided vehicles: key computational data doi:10.57760/sciencedb.012kv en This dataset contains selected numerical evidence from a deterministic study of candidate preselection in multi-objective flexible job-shop scheduling with automated guided vehicles. It examines the tradeoff between information used to choose candidates and the number of search opportunities under an evaluation budget. One schedule evaluation denotes one decoded schedule producing a three-objective vector. Energy values are normalized model indices, not measured factory electricity consumption.The release includes run-level metrics, saved objective fronts and reference sets, raw component timings, candidate-pool diagnostics, model sensitivities and generated instances. The three principal blocks contain 4,320, 2,160 and 2,160 saved endpoint rows. These include repeated budgets and are not independent sample counts. All 1,920 original audited candidate pools, 1,440 model-sensitivity summaries and 240 generated instance files are retained. All comparator arms, seeds and transport regimes in the selected blocks are included regardless of result direction.Evidence supporting manuscript 3.7.3 is separated from subsequent 3.8 experiments on pool size, an NSGA-II backbone and equal-time budgets. The latter include 2,160 aggregated base-metric rows and nine primary contrasts. The proposed fourth native-transport family has no admitted formal result and is not represented as completed evidence. Calibration and external-comparator outputs are included, while original third-party benchmark files are excluded and their source is identified in the README.Files use JSON, gzip-compressed JSON, CSV and text documentation. README_PUBLIC.md defines variables, units, provenance and analysis routes. MANIFEST.json inside the ZIP provides SHA256 checksums and documented projections. Retained scientific files are byte-identical to the verified curation source; packaging introduced no new optimization runs. This data-only release supports inspection and selected numerical recalculation, not a one-command rerun of every experiment or figure. Complete execution streams, software and checkpoints remain in a separate private GitHub preservation archive, not public code access. The numerical data, generated instances and study documentation are shared under CC BY 4.0. flexible job-shop scheduling; automated guided vehicles; candidate preselection; multi-objective optimization; evaluation budget; computational experiments Qingyou Wang PUBLIC https://creativecommons.org/licenses/by/4.0/ dataset Science Data Bank