A Simulation-Based Metaheuristic Approach to Integrated Scheduling of Seedling Production
A Simulation-Based Metaheuristic Approach to Integrated Scheduling of Seedling Production
复制标题
基于仿真的种苗生产集成调度的元启发式方法
DOI：
10.1109/lra.2021.3140056
复制
发表时间：
2022-04
影响因子：
5.2
通讯作者：
Jingyuan Feng;Xiangpei Hu;N. Kong
中科院分区：
文献类型：
--
作者：
Jingyuan Feng;Xiangpei Hu;N. Kong
苗木生产具有土地利用和增产的经济价值，是现代农业的重要组成部分。苗木生产的三个关键决策是接受新订单、完成积压订单和控制生长速度。第一个决策是在新订单到达时以在线方式做出的，而随后的两个决策是定期做出的。由于三种操作之间的相互作用以及对决策的异构频率的要求，许多分析方法在处理问题时缺乏足够的灵活性，难以在实践中应用。在本文中，我们提出了一种基于模拟的元启发式优化方法来确定上述决策的适当策略，包括对具有代表性的幼苗生产过程进行细粒度模拟模型，对每个决策进行启发式决策规则，以及嵌入最优计算预算分配方案的粒子群优化算法来有效地探索规则组合空间。通过数值实验，我们证明了我们的元启发式算法的可行性；展示了我们的积压订单履行规则比两个基准顺序调度规则的优越性；并论证了定期调节生产速度比保持不变的生产速度更有经济效益。我们的工作提出了模拟优化在智能苗木生产操作管理中的新应用。
Seedling production is important to modern agriculture for its economic value in land utilization and yield promotion. Three crucial decisions for seedling production are new order acceptance, backlogged order fulfillment, and growth rate control. The first decision is made in an on-line fashion whenever a new order arrives, whereas the subsequent two decisions are made periodically. Given the interplay between the three operations and requirements on the heterogeneous frequency of the decisions, many analytical methods lack sufficient flexibility to deal with the problem and difficult to apply in practice. In this letter, we propose a simulation-based metaheuristic optimization approach to identify the appropriate policy for the above decisions, including a fine-grained simulation model for a representative seedling production process, heuristic decision rules on each of the decisions, and a particle swarm optimization algorithm embeds an optimal computing budget allocation scheme to explore the rules combination space efficiently. Through numerical experiments, we justify the viability of our metaheuristic algorithm; show the superiority of our backlog order fulfillment rules over two benchmark sequential dispatching rules; also demonstrate the economic benefit to periodically regulate the production rate rather than keep in invariant rate. Our work presents a novel application of simulation optimization to smart seedling production operations management.