基于学习效果的蔬菜电商成组拣货排序方法

按单拣货是电子商务背景下物流配送中心最关键、最复杂的作业环节.它直接衔接着物流配送和客户,且大部分是劳动密集型作业,因此按单拣货也是出错率最多,耗费时间最长的环节.在拣货之前对订单考虑配送因素和相似性成组,能够大大降低拣货成本.蔬菜电商拣货系统是基于人的柔性作业系统,拣货人员的学习效果导致作业效率随时间变化,从而订单的拣货时间不确定,对按单拣货作业有着不可忽视的影响.本文针对蔬菜B2C电子商务直销背景下拣货环节的订单成组作业优化问题进行研究,基于拣货人员的学习效果,建立最小化订单拣货完成时间之和的拣货序列优化模型.针对该问题多阶段,多层次决策特点,基于序贯决策思想,提出两阶段的求解方案:第一阶段订单成组,提出同时考虑订单配送距离,打包材料相似性和订单相似性三种指标融合的订单成组准则,并给出启发式算法对订单进行成组;第二阶段成组订单作业调度,提出基于修订式非递减的订单规模排序方法对组内订单进行排序,组间排序是基于具有降低搜索范围,提高搜索能力的改进模拟退火-遗传算法.通过数值实验和算法比较,验证了本文算法的有效性和实用性.研究结果表明,本文得到的方法能大大缩减拣货时间成本,为蔬果类商品网上直销企业生成拣货作业计划提供理论指导.

Batch picking and sorting approach for fresh vegetables e-commerce considering learning effect

Order picking is the most critical and complicated operation in the logistics distribution center under the background of e-commerce. It links logistics distribution and customers directly and is labor-intensive, which therefore makes it a time-consuming process with higher error rate. Batching orders according to their similarities before order picking is effective way to reduce the operation cost. Since the order picking system is a human-based flexible operation system, the learning effect of workers could have great influence on operation efficiency over time, thus would have non-negligible impact on actual operation time. Aiming at order operation for fresh vegetables e-commerce, batch picking and sorting is studied in this research considering workers' learning effect along with time to minimize the sum of the total completion time. A sequential two phase solution approach is then proposed based on the hierarchical characteristics of this problem. In the first phase (order batching), new similarity metric is put forward which combines three indicators including orders' distribution distance, packing materials and order similarity, and a heuristic algorithm is given to batch the orders. At second phase (batch scheduling), orders within the same batch are scheduled based on revised non-decreasing order size and then an improved SA-genetic algorithm is presented by reducing searching scope and enhancing searching ability to schedule different order batches. The validity and practicability of proposed approach are verified by numerical comparison experiments. Results show that our approach could greatly reduce total time of order picking operation, which could provide theoretical guidance for direct online sales companies of fresh fruits and vegetables.