考虑软时间窗限制和模糊旅途时间的生产配送集成调度优化

Integrated production and distribution scheduling optimization considering soft time windows and fuzzy travel times

中文关键词  生产配送集成调度  软时间窗  模糊旅途时间  模糊加权叠加  自适应变邻域搜索  多目标优化

英文关键词  integrated production and distribution scheduling  soft time windows  fuzzy travel times  fuzzy weighted superposition  adaptive variable neighborhood search  multiobjective optimization

中文摘要

      针对考虑软时间窗限制和模糊旅途时间的生产配送集成调度问题, 本文构建了以生产–库存–配送总成本和提前延期加权惩罚时间为双优化目标的混合整数非线性规划模型. 定义了一种模糊加权叠加操作, 进行模糊加权惩罚时间的计算. 基于优化问题结构设计了三阶段解码规则, 其中涉及巡回环的划分, 通过计算巡回环的最佳配送出发时间获得批次制造顺序, 以及采用后向调整策略进行批次和巡回环的协调. 提出了一种自适应变邻域搜索改进的非支配排序遗传算法(NSGA-II-AVNS)求解该问题. 其中根据问题特征设计了5种具有不同搜索性能的邻域结构, 通过自适应选择机制提高优秀邻域结构的执行次数. 并且为避免迭代后期邻域结构选择固化, 提出了邻域结构分值重置操作. 实验结果表明NSGA-II与AVNS的融合, 较好的平衡了算法的探索和开发能力, 是求解该问题的一种极具竞争力的方法.

英文摘要

      A bi-objective mixed integer nonlinear programming model with total cost of production-inventory-distribution and total weighted early and tardy penalty time is developed for the integrated production and distribution scheduling problem considering soft time windows and fuzzy travel times. A fuzzy weighted superposition operation is defined for the calculation of fuzzy weighted penalty times. A three-stage decoding rule is designed based on the structure of the optimization problem, which involves the division of the tour, obtaining the batch manufacturing sequence by calculating the optimal departure time of the tour, and the coordination of batches and tours by a backward adjustment strategy. An improved non-dominated sorting genetic algorithm II based on adaptive variable neighborhood search (NSGA-II-AVNS) is proposed to solve this problem. Five neighborhood structures with different search properties are designed according to the problem features, and adaptive selection of neighborhood structures to increase the number of executions of excellent neighborhood structures. The neighborhood structure score reset operation is proposed to avoid neighborhood structure selection solidification. The experimental results show that the fusion of NSGA-II and AVNS has well-balanced exploration and exploitation capabilities of the algorithm, and it is a very competitive method to solve this problem.