| A new robotic mobile fulfillment system caters to the characteristics of e-commerce orders with multiple and small batches and each order contains many products,so it is sought after by many e-commerce logistics enterprises.However,due to the advantages of traditional fixed shelf systems in terms of space utilization and storage costs,the mobile fulfillment systems cannot completely replace the traditional system.Several e-commerce logistics enterprises choose to transform part of the traditional warehouses in the storage center into mobile fulfillment system warehouses when introducing the new system,resulting in the long-term co-existence of the old and the new storage systems.This leads to a new problem:the two types of systems store all products together.Due to capacity limitations,any single system cannot store all products.And the diversified products contained in e-commerce orders may be stored in two kinds of systems in a decentralized way.As a result,the products of some order needs to be picked in two warehouses first,and then merged together for packing and delivery together.This process spends a lot of time and cost,which seriously violates high-efficiency and low-cost selection requirements for e-commerce.This problem is different from the order splitting problem.Existing researches focus on minimizing the splitting cost of multiple distribution centers,while this paper studies the warehouse management cost in a single distribution center.This paper not only considers the confluence cost of order merging cost in two different warehouses,but also considers the replenishment cost of the same product in both warehouses.The objective of this paper is to minimize the sum of order merging cost and replenishment cost.This paper designs an integer linear programming model to solve this problem,and proves that the problem is an NP-hard problem.To solve large-scale practical problems,this paper proposes a hybrid heuristic algorithm based on large neighborhood search and local search,and designs five local search operators based on the characteristics of the problem.This paper designs an improvement strategy for objective function and product evaluation,and proves the feasibility of this strategy through mathematical theorem.In the experimental analysis part,this paper generates small-scale and large-scale cases according to the actual storage center’s order data for a week.Small-scale cases are used to compare the results of the hybrid heuristic algorithm and Gurobi’s mathematical model.The results show that the hybrid heuristic algorithm designed in this paper can obtain better results than Gurobi running for 4 hours in tens of seconds,when the scale of the calculation example reaches 161 orders in 70 products.Based on large-scale cases,since the existing research does not fully match the heuristic algorithm,this paper selects the FITS algorithm designed by Zhou et al.similar to the research problem and improved with simulated annealing algorithm for comparison.From the solution results,the results of the hybrid heuristic algorithm and the improved FITS algorithm are basically the same,and both are better than the improved simulated annealing algorithm,but the hybrid heuristic algorithm designed in this paper has the shortest running time.This paper selects the unit order merging cost and the proportion of the mobile fulfillment system warehouse to the total warehouse area for sensitivity analysis,and draws two management implications.On the one hand,warehouse management needs to formulate a reasonable order picking wave plan and improve staff picking proficiency,so as to reduce order-merging costs.On the other hand,when e-commerce companies introduce mobile fulfillment systems,they should expand the proportion of mobile fulfillment system warehouses in the overall warehouse area as quickly as possible,and avoid the peak value of the sum of the order merging cost and replenishment cost when the proportion of mobile fulfillment system warehouses proportion reaches 50%. |
PDF Full Text Request