摘要
为有效解决多周期环境下的易腐品生产-库存-分销集成优化调度难、成本居高不下问题,提出一种改进的无重访遗传算法。该算法利用归一化数据预处理方法进行多维实数编码;在空间二叉分割树数据结构生成规则中,建立多维子空间同时分割机制,形成解的唯一解空间;应用田口实验设计方法调整无重访遗传算法参数,获得近似最佳组合参数。结果表明,该算法可以加快搜索速度,提高求解精度,在不同需求情景下能够有效降低易腐品冷链总成本。
To effectively solve the problem of production-inventory-distribution integration optimization scheduling of perishable products and high cost in the multi-cycle environment,we propose an improved non-revisiting genetic algorithm(NrGA).The normalized data preprocessing method was used to encode multi-dimensional real numbers.In the data structure generation rules of binary space partitioning tree(BSP tree),we established the multi-dimensional subspace simultaneous partitioning mechanism to form the unique solution space.We applied Taguchi experimental design to adjust the parameter of NrGA,and obtained the approximate optimal combination parameters under different demand scenarios.The experimental results show that this algorithm can speed up the search speed,improve the accuracy of the solution,and effectively reduce the total cost of perishable cold chain under different demand scenarios.
作者
刘巍巍
王诗雅
Liu Weiwei;Wang Shiya(School of Mechanical Engineering,Shenyang University of Technology,Shenyang 110870,Liaoning,China)
出处
《计算机应用与软件》
北大核心
2022年第7期67-72,166,共7页
Computer Applications and Software
基金
辽宁省自然科学基金计划重点项目(20170540673)
辽宁省教育厅重点科技计划项目(LZGD2017038)。
关键词
易腐品
冷链
生产-库存-分销
库存控制
遗传算法
归一化
空间二叉分割
田口实验设计
Perishable products
Cold chain
Production-inventory-distribution
Inventory control
Genetic algorithm
Normalization
Binary space partitioning
Taguchi experimental design
