| In recent years,total logistics costs and GDP growth have maintained a steady annual growth rate.In terms of the number of express packages,which is quite representative of the prosperity of the logistics industry,my country is developing at an annual growth rate of 30%.Behind the surge in the volume of parcels is that the distribution of commodity inventory in the supply chain is becoming more and more decentralized,and it is getting closer to consumers,which also brings about an increase in the number of warehouses year by year.According to data from the National Bureau of Statistics,Chinese warehouses deliver completed areas at a rate of 25 million to 32 million square meters every year.The driving forces behind the prosperity of these logistics industries all point to the vigorous development of the e-commerce industry.Taking the most representative Alibaba and JD.com in the past 9 years as an example,we choose an indicator GMV(Gross Merchandise Volume,GMV for short)that best represents the prosperity of e-commerce: the former GMV has grown in 9 years 7 times,the latter GMV increased by as much as 26 times in the same time.The sudden rise in industry figures is inseparable from technological progress as the underlying driving force.This article selects the warehouse,the most important node in the supply chain,as a perspective,and observes three generations of transformation of warehouse technology quietly produced in this process: manual warehouse technology,automated warehouse technology,and robot intelligent warehouse technology.The main driving force for the development of warehousing technology is three aspects: first,in terms of manpower,the number of the right-age labor force in my country has declined and labor costs have increased year by year;The strong support of this industry;the third is the intensive investment and financing of major capital trends in the economy,which has helped the development of this industry.In the field of third-generation robot intelligent warehouse technology,this thesis focuses on the key issues of AGV intelligent warehouse with AGV(Automated Guided Vehicle,AGV for short)as the main execution unit and information system as the main decision-making unit.At the same time,this thesis focuses on the key issues of AGV intelligent warehouses,and conducts research on the three levels of planning,planning,and operation,forming three-level methodologies,and verifying the effectiveness of these methods in actual experiments.The strategy layer is mainly concerned with the design and layout of the warehouse.That is,how to design an AGV smart warehouse that has not yet been put into production,or transform an already operating traditional warehouse into an AGV smart warehouse,which involves which SKUs should be stored in the warehouse,how many shelves should be arranged,how many AGVs and how many workstations can meet the needs of future warehouses.Strategic positioning across the supply chain.This thesis abstracts the entire AGV warehouse planning method into six key steps:warehouse location analysis,storage item analysis,flow stock analysis,equipment quantity analysis,layout planning analysis,and road network setting analysis.These six key steps are not isolated,but It is the previous step that always has a prerequisite effect on the subsequent steps.This article demonstrates the effectiveness of the method through an actual AGV smart warehouse planning case in Jingdong: the actual operation data shows that the change of the average cost per unit during the planning period and the operation period is within the controllable range,and the role of the planning methodology in the actual industry obvious.The main concern of the planning layer is the scheduling of AGVs in a day’s production.That is,how to schedule AGVs in a day that has been divided into smaller time intervals(waves): work,charge,and determine the number of orders and charging time that the AGVs working in the current period need to complete,so that at the end of the day,it needs to be completed on the same day Production orders can maximize the timeliness requirements.This thesis formulates the problem as a stochastic dynamic programming problem with exogenous information and shared resources.And note that the number of orders that need to be produced for the day usually fluctuates at a base level.Based on this property and Lagrangian relaxation,this thesis proposes an implicit split flow model.In addition,this thesis proposes a feasible rolling algorithm to solve the AGV scheduling problem.Finally,a numerical experiment was carried out based on the actual data of Jingdong,and the algorithm in this thesis was compared with the simulation algorithm used in actual production,showing that the algorithm in this thesis has 78% less remaining orders than the simulation algorithm,and is more stable.The main concern of the operation layer is the matching of AGVs and tasks,that is,to confirm the shelves corresponding to the outbound tasks(including SKU and number of pieces)of each outbound workstation,and which AGV is required to transport these shelves to the corresponding outbound workstation.Operation,in order to expect the shortest travel distance of AGV.This thesis abstracts this problem into a integer programming model to describe and solve it,in order to make the driving distance of all AGVs involved in task matching the shortest,and to meet the SKU’s inbound and outbound requirements as much as possible.This thesis establishes this problem as a 0-1 integer programming model.The difficulty of solving it directly is that it is impossible to find a feasible solution with good quality in a very short time(3-5 seconds).By observing the constraint characteristics of this model,this thesis adopts the "divide and conquer" approach.The specific method is to penalize a link constraint of the original problem into the objective function of the original problem to form Lagrangian relaxation.In this way,the original problem can be disassembled into two sub-problems.This thesis first solves the second sub-problem,and then substitutes this solution into a fine-tuned version of sub-problem 1(reintroducing the constraints previously placed in the objective function)to obtain the final solution and ensure the validity of the final solution.Numerical experiments show that the quality and speed of the algorithm in this thesis are obviously higher than that of directly calling the Cplex solver.Finally,this article demonstrates the effect of the algorithm through the experimental results carried out in the actual warehouse of JD.com.Finally,this article looks forward to the future development of the AGV intelligent warehouse,from the perspectives of improving the storage density of the warehouse and improving the operational efficiency of the warehouse,and combining an improved AGV intelligent warehouse model in the industry.That is to say,the storage shelf has changed from a movable to a fixed three-dimensional shelf,and the replaced movable unit has become the smallest storage unit: the turnover box.The AGV has changed from the jack-up type to carry one shelf at a time,to a new mode that can obtain multiple turnover boxes at multiple storage points at one time,transport them to the workstation for unified processing,and put them back together.Therefore,this thesis concludes that the evolution direction of AGV intelligent warehouse is: the miniaturization of the mobile storage unit,the storage space and the efficiency of the handling unit operation.In the process of solving the AGV intelligent warehouse problem,the author found that: AGV intelligent warehouse problems that are highly integrated with reality have the characteristics of large constraint space and many solution variables,and due to the real-time nature of warehouse production,the solution of such problems is required need to be completed in a short period of time.However,whether it is at the planning level or at the operational level,this thesis highlights the methods of "turning the big into small" and "divide and conquer" according to the characteristics of these problems.The author believes that this method is a necessary intuition in dealing with optimization problems in the supply chain industry and logistics industry,and hopes to provide a reference for future research in this field. |
PDF Full Text Request