Research On Integrated Optimization Of Order Allocation And AGV Scheduling In Robotic Mobile Fulfillment System

The accelerated development of e-commerce has brought about a surge in orders,which has put forward higher requirements for the order processing efficiency of distribution centers.With the rise of Artificial Intelligence and the development of Internet of Things technology,automated storge equipment has gradually entered the warehouse,which greatly shortens the time of operation in the warehouse and improves the safety and accuracy.In recent years,a new "Parts-to-picker" order picking system,Robotic Mobile Fulfillment System(RMFS),has begun to be used in distribution centers.In this system,Automated Guided Vehicles(AGV)orderly transports the racks to visit the picking stations,and the picker stays on the picking station and continuously picks items from the racks until all the order requirements are fulfilled.The main research work of this paper is as follows:(1)Based on the order picking technology of RMFS,in the picking mode that order cannot be split and packed independently,this paper studies picking operations with multiple racks visiting multiple picking stations,taking into account the queuing of AGV at the picking station.Then the Integrated Order Allocation and Robot Scheduling Problem(IOARSP)is proposed.The problem combines order allocation decisions with AGV scheduling decisions,studies how orders are allocated to the picking stations and how to schedule the AGV transport racks to meet the needs of the picking stations,so that the makespan for finishing picking a batch of orders is the shortest,improving the efficiency of the system.(2)The mixed integer programming model of the IOARSP is established,and the objective function is to minimize the makespan of the system.For the characteristics of the problem,this paper designs a variable neighborhood search algorithm(VNSII),including two disturbance neighborhoods and five local search neighborhoods.Neighborhood variation takes variable neighborhood descent(VND)as the framework,which can continuously performs perturbation and local optimization on the current solution,so as to achieve an efficient search of the solution space.In addition,since the IOARSP involves AGV queuing to visit the picking station,this paper proposes an algorithm to obtain the objective function by analyzing the solution.(3)In this paper,42 examples of different sizes are constructed to prove the feasibility and effectiveness of the VNSII algorithm and seven neighborhoods.Explore the effects of different picking station layouts and parameters on picking efficiency,and explore the impact of different order allocation on AGV scheduling optimization.The experimental results show that: a)In the small example experiments,the average relative gap between the results solved by the VNSII algorithm and the CPLEX is 0.1%,which proves the feasibility and effectiveness of the VNSII algorithm.b)During the algorithm iteration process,the five local search neighborhoods show different degrees of optimization ability,and the perturbation process shows the ability to jump out of local optima.c)Compared with the L-shaped and U-shaped layouts of picking stations,the picking time under the I-shaped layout is reduced by 11.4% and 16.1% respectively,indicating that the I-shaped layout is more suitable for the research scenario of this paper.d)The number of picking stations and the number of AGV have a significant impact on the picking time.e)In all case experiments,compared with only optimizing AGV scheduling,the picking time under the integrated optimization of order allocation and AGV scheduling is shortened by an average of 20%.And with the increase of orders,the advantages of integrated optimization are more obvious.