关键词: 色纺企业, 生产调度, 多目标优化, 改进蚁群算法

Abstract: As a branch of textile industry, the colored spun yarn industry has its own characteristics of multi-variety, small batch, significant changes in color demands and short delivery time. The production scheduling of colored spun yarns differs from that of natural yarns in that workshops need to simultaneously produce various colored yarns with different blending ratios. Moreover, each type of colored yarn requires different semi-finished raw materials for each production process. However, traditional manual production scheduling involves a large workload, making it difficult to respond quickly and resulting in a high probability of errors. To address this issue, first, a production scheduling model for colored spinning processes was constructed, by taking into account factors such as machine downtime, order variety, and color sequences. The objective functions include minimum order delivery time, spinning frame changeover times, minimum and maximum completion times, and the total number of overdue orders. Additionally, this paper introduced flexible constraints on yarn variety similarity before and after production order, to manage issues such as fiber staining and order blocking caused by producing different colored yarn simultaneously. This enhances the adaptability of the model in handling urgent order insertions and order blocking issues.
Due to the large solution set range and complex constraints of the model, as well as the high computations and tendency of traditional ant colony algorithms to fall into local optimal solutions in solving large-scale scheduling problems, this paper proposed an improved ant colony algorithm incorporating candidate list strategy, adaptive pheromone updating, and a max-min ant system to solve the model. The algorithm's parameters were selected by using grid search: and pheromone factor α = 2.0, heuristic information factor β =3.0, evaporation factor ρ = 0.75, maximum pheromone concentration τmax = 6.0, and minimum pheromone concentration τmin = 0.15 were obtained, yielding the best optimization results for various scheduling objectives. Simulations were conducted for a workshop with 64 types of orders and 96 spinning machines, comprising 274 pending batch orders. The results demonstrated that while the traditional ant colony algorithm could improve scheduling objectives, its local pheromone updating strategy, which enhances the pheromone concentration of certain paths, could lead to local optimal solutions, causing order delays and exhibiting lower robustness. At the same time, the improved ant colony algorithm could solve the production scheduling problems of different-sized colored spun yarn enterprises with good robustness. The improved ant colony algorithm outperformed other methods in terms of scheduling objectives without order delays. Compared to manual scheduling, the proposed method reduced the order delivery time score fitness value by 83.77%, decreased the downtime for spinning machines waiting for retrofitting by 40.9%, and lowered the maximum completion time by 11.33%.

Key words: colored spun yarn enterprises, production scheduling, multi-objective optimization, improved ant colony algorithm