Green and on-time scheduling in real-world textile production: a genetic algorithm approach validated by MIP and benchmarked with ACO
提出两种批调度模型，分别最小化碳足迹和完工时间加延迟，在纺织染色数据上遗传算法比蚁群算法更优，碳感知模型可减少55%碳排放和66%延迟，同时提高设备利用率。
This study presents two batch-scheduling models: one minimising carbon footprint and another reducing finishing time plus tardiness. Batch scheduling is frequently used in textile, semiconductor, and chemical industries to group items, optimise throughput, and limit resource consumption. In textile dyeing, large water and energy demands make carbon reduction and on-time delivery essential goals. For MIP, GA, and ACO, the classical dual-objective form–total tardiness + total finishing time–is reformulated as a single carbon-footprint objective and then benchmarked against the two-objective versions. Exact MIP solutions on small instances validate the performance of both GA and ACO; medium-sized data sets enable a direct GA–ACO comparison, and GA's superior results motivate its exclusive use on large-scale scenarios. On the real factory data the carbon-aware GA, relative to its time-based counterpart, cuts total CO2 by 55%, reduces total tardiness by 66%, and raises average utilisation from 63.6% to 75%. A paired t-test over eleven data sets confirms that the CO2 reduction is statistically significant at the 95% confidence level, with no service-level degradation. All large instances are solved within minutes on laptop hardware, demonstrating that the proposed approach delivers sustainability and operational performance simultaneously.