Research On Intelligent Packing Strategy Of E-Commerce Warehousing

Under the policy guidance of the concept of "green development" and the concept of "highquality development",reducing carbon emissions,improving the degree of "unmanned" and "intelligent" in the process of enterprise production and operation,and realizing the goal of "double carbon" have become the consensus of all sectors of society.With the rapid development of e-commerce logistics,the national express package volume has climbed to more than 80 billion pieces per year.Among them,the space utilization rate of many packing boxes is very low,and the standardization of packing box size is low,which is difficult to recycle.After unpacking by express,the packing box is discarded at will.It not only produces huge packaging and transportation costs,but also a large number of intact packaging boxes are discarded,resulting in serious economic waste.The increased express waste has also greatly polluted the living environment of urban residents.Reasonable packing and recycling of express packaging boxes in the field of e-commerce warehousing has increasingly become an important issue closely related to the social life of the majority of residents.Starting from the packing practice in the field of e-commerce warehousing,this paper constructs a "full coverage" strategy system from calculating the order packing scheme to the recycling of packing cases.Aiming at the "green and low-carbon" governance mode of packaging box recycling and reuse,this paper focuses on two key points: improving the loading rate of packaging scheme and optimizing packaging box size.This paper takes the intelligent packing strategy as the research focus,constructs the algorithm system of BSAS,and puts forward the intelligent selection mechanism for different algorithms in the algorithm system.This paper designs a series of heuristic packing strategies and their combination methods,and combines them with the intelligent algorithm framework to design the intelligent packing strategy.A packing box size design method different from the basic module segmentation is proposed,but the packing box size is optimized based on the information characteristics of order items.In order to easily apply the intelligent packing strategy to the actual packing activities in the field of e-commerce warehousing,a fully automatic intelligent packing system is designed in this paper.In the manual packing operation environment,an intelligent packing auxiliary decision-making system is designed to provide three-dimensional visual guidance for packing operators.According to the packing box size optimization method based on historical order information,specific improvement suggestions for optimizing the standardized size of packing box are provided regularly.The main research contents and innovations of this paper include the following aspects:(1)In order to make the intelligent packing strategy play an effective role in the field of ecommerce storage and realize the digital upgrading and transformation of the packing scheme,this paper makes an in-depth research on the packing scheme in saving calculation time and improving the level of order loading rate.On the basis of combing the previous research results,this paper defines the key problems to be studied,focuses on the practical situation of packing in the field of e-commerce warehousing,takes the minimum number of boxes and the highest loading rate as the optimization goal,considers the three-dimensional packing problem in the offline state,and draws lessons from the previous packing strategies,models and algorithms.However,after analysis,it is found that there are many differences in this paper.Under the precondition of multiple box types,the calculation time of packing scheme is minimized and the loading rate is the highest,and the same order items need to be loaded into one box as much as possible.The dynamic variability of packing box size also belongs to the particularity of packing problem in the field of warehouse.(2)According to the specific situation in the field of e-commerce warehousing,this paper constructs a mixed integer programming model for multi box three-dimensional packing problem,and statically describes the optimization objectives and constraints of the packing scheme.When designing the heuristic strategy of packing link,the static model is transformed into a dynamic process,and the heuristic strategies of different packing links are organically combined to form a method system for solving the heuristic strategy combination of packing problem.The calculation results of the corresponding packing scheme in seconds are obtained,which makes the average loading rate of orders on the real data set of the industry reach more than 68%.In order to further optimize the packing scheme and improve the level of order loading rate,this paper takes into account the balanced optimization between calculation time,problem scale and loading rate under the condition of appropriate relaxation of calculation time or small scale of order items.Based on the heuristic strategy combination method system,the Hybrid Genetic Algorithm(HGA),Monte Carlo Tree Search(MCTS)and Deep Reinforcement Learning(DRL)framework are integrated to realize the intelligent optimization of packing strategy,so that the average order loading rate on the real data set is more than 70%.(3)Combining the heuristic strategy combination method with the intelligent algorithm framework to build an intelligent packing strategy system can not only continuously improve the effect of packing scheme,but also compare and analyze the optimization degree between different strategies and algorithms.In terms of fusion HGA,this paper designs a series of improved operators such as "point transformation",which has significantly improved the solution performance and optimization degree compared with the traditional methods.In terms of integrating MCTS,this paper designs a neural network structure suitable for packing problem,and trains the neural network through reinforcement learning framework to improve the "expansion" and "selection" links in the tree search process.The optimization degree of the solution is significantly higher than that of the traditional Monte Carlo tree search algorithm,and the calculation time is 24.84% less than that of the hybrid genetic algorithm.In terms of integrating DRL,this paper designs two methods: item packing order selection and heuristic packing strategy combination selection.As for the selection and design of article packing sequence,it is mainly to convert the article packing process into timing decision-making process.The heuristic packing strategy combination selection is mainly to match the optimal strategy combination according to the successively arriving order information.Through the example calculation in the data set,it is verified that the DRL is feasible to solve the threedimensional packing problem of multiple box types in the field of complex combinatorial optimization,as well as its advantages in solving efficiency.(4)In order to fully tap the potential of the optimization strategy of packing problem in the field of e-commerce warehousing,this paper takes the number of packing box types and the size of each type of packing box as variable conditions.Based on the research on the basic heuristic strategy and intelligent optimization strategy of three-dimensional packing,the order loading rate is improved by optimizing the packing box size.It not only saves the consumption of packaging consumables,reduces the packaging cost and transportation cost,but also puts forward a new method for the standardization of packaging size.In the research of packing box size optimization,aiming at the item space stacking and clustering strategy without fixed size,one-dimensional and two-dimensional item space stacking and clustering strategy with fixed size,the mathematical model of mixed integer programming is constructed,and the intelligent optimization strategy and algorithm combining heuristic strategy combination and hybrid genetic algorithm framework are designed.The calculation experiments on real data sets show that under the condition of the same number of box types,in the heuristic strategy combination method,the two-dimensional fixed size item space stacking and clustering strategy achieves the best performance in the same algorithm,and the average order loading rate reaches more than 74%;Under the intelligent optimization strategy of heuristic strategy combination and hybrid genetic algorithm,the average order loading rate has been greatly improved,reaching more than 78%.Compared with the heuristic strategy combination method under the original packing box size,the average order loading rate is increased by more than 10%.To sum up,the research on the packing problem in the field of e-commerce warehousing shows that the heuristic strategy combination method,the optimization strategy of intelligent algorithm and the research perspective of order packing box size optimization fully consider the actual situation of packing operation.This has an obvious optimization effect on realizing the "green and low-carbon" governance mode of packing problem,saving packaging consumables,reducing the calculation time of packing scheme,improving the level of order loading rate and the standardization method of packing box size.