Automatic stereoscopic warehouse goods allocation distribution method using digital twinning drive

Abstract

The invention provides an automatic stereoscopic warehouse goods allocation distribution method driven by digital twinning. According to the method, a warehouse management system is constructed through five dimensions of a physical layer, a twinning layer, a data layer, a service layer and connection. The physical layer is an actual warehousing operation scene; the twin layer maps the equipment and environment of the physical layer to a virtual space, and constructs a storage digital model; the data layer is responsible for acquiring, processing and storing data of the physical layer and providing accurate input for the service layer and the twin layer; the service layer provides optimization decision making and man-machine interaction functions for user and system management; and the connection layer is responsible for data and control transmission among the layers, so that the collaboration and the real-time performance of the system are ensured. A multi-objective optimization algorithm is supported through a digital twinning technology, intelligence and precision of goods allocation distribution are achieved, and a brand new solution is provided for digital logistics management.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of automated warehouse storage allocation, and particularly relates to an automated warehouse storage allocation method driven by digital twinning. BACKGROUND

[0002] Global manufacturing is undergoing a major transformation driven by integration, intelligence, and digitization. In this process, warehousing, as a key link in enterprise operation processes, is also facing the challenge of transforming from traditional industry to intelligent manufacturing. With the gradual increase of logistics storage costs and the increasingly stringent requirements for warehouse efficiency, automated storage and retrieval systems (AS / RS) gradually replace traditional warehouses with their high space utilization, large storage capacity, and low labor costs, and are widely used in manufacturing, retail, e-commerce, and other fields. In order to improve the operation efficiency and resource utilization of the warehouse, storage allocation optimization has become an important research topic. In a complex warehouse environment, due to the diversity of goods, frequent changes in demand, and complexity of warehouse layout, traditional storage allocation methods have the following shortcomings: 1. Lack of intelligent decision support: Storage allocation needs to balance multiple objectives such as picking efficiency, shelf balancing, and equipment energy consumption. Traditional single-objective or simple multi-objective optimization algorithms are not suitable. At the same time, in traditional methods, even if multiple storage allocation schemes are generated, the decision-making process still heavily relies on the subjective judgment of managers, especially when multiple objectives conflict, this experience-driven decision-making approach may lead to inconsistency and decision bias.

[0003] 2. Poor dynamic response capability: Existing methods are mostly based on static rules or historical data, and are difficult to respond to changes in actual inventory demand, environmental conditions, or system state in real time. When demand changes, traditional methods cannot adjust the storage allocation scheme in a timely manner, resulting in decreased allocation efficiency.

[0004] 3. Lack of systematic digital support: Traditional methods cannot fully utilize real-time data provided by sensors and IoT devices, and lack effective simulation tools to predict the actual effect of storage allocation schemes, making it difficult for managers to adjust allocation strategies to adapt to complex warehouse environments. SUMMARY

[0005] The technical problem to be solved by the present application is to provide an automated warehouse storage allocation method driven by digital twinning to overcome the shortcomings of the prior art. This method combines the physical environment of the actual warehouse with its digital virtual model to optimize storage allocation and improve warehouse operation efficiency.

[0006] The method comprises: constructing a digital twin five-dimensional model, including a physical layer, a twin layer, a data layer, a service layer and a connection; The physical layer is the actual place of the goods access operation, represents the physical entities and environment actually existing in the digital twin warehouse management system, the physical entities include shelves, stackers, in-out warehouse tables, totes and goods; the data of the physical layer is uploaded to the data layer through real-time interaction, while receiving optimization instructions from the service layer; The twin layer is a digital mirror image of the physical space, which provides simulation support for goods location allocation and warehouse operation by constructing a warehouse digital twin model mapped with the physical warehouse; the warehouse digital twin model relies on three-dimensional modeling technology to construct accurate three-dimensional models in Blender software according to the shape, color, material and other characteristics of the physical entities, and ensures that the virtual model is highly consistent with the real warehouse environment through high-quality rendering, and sets up a model scene tree according to the parent-child relationship information between the three-dimensional models, and the model data outputs a three-dimensional model file in the format of.gltf or.glb; the Three.js engine is used to load and parse the three-dimensional model in the format of.gltf or.glb on the Web, combined with the environment data collected by IoT devices, to realize real-time mapping of warehouse layout, device state and environment parameters, and dynamically bind the warehouse data through the Vue3 framework to ensure real-time updating of the number of goods on the shelves, the position and running state of the stacker; all geometric models, dynamic data and rendering scenes are integrated to construct a complete warehouse digital twin system; The data layer serves as a data support layer connecting the physical layer and the twin layer, which obtains dynamic data and static data, cleans and stores the data, and provides input for the service layer, wherein the dynamic data is used to reflect the real-time operation of the automated warehouse, including goods transfer data, device state, environment data, etc., and the static data includes basic data such as warehouse geometric dimensions, device physical constraints, and goods material properties; The service layer provides a human-computer interaction interface and decision support for users, realizes intelligent decision-making and manual decision-making for goods location allocation; The connection includes data connection and control connection, the data connection is used to collect physical layer data to the data layer through IoT (English: Internet of Things, abbreviation: IoT) devices and network devices, the information processed by the data layer is input to the twin layer for modeling and simulation, and to the service layer for generating goods location allocation scheme; the control connection is used to feed back the goods location allocation instructions to the physical layer to guide the actual operation, and to the twin layer for virtual operation.

[0007] The service layer performs the following steps: Step 1, obtaining the goods location state, goods characteristics and device state information from the data layer after processing; Step 2: Use the intelligent improved algorithm SA-NSGA-Ⅱ to generate a set of optional cargo space allocation solutions ,in represents the Ith intelligent cargo location allocation plan; Step 3: Use the decision-making method combining fuzzy analytic hierarchy process (FAHP) and TOPSIS to select the best solution from the set of cargo space allocation solutions. Select the best intelligent cargo space allocation solution ,in Indicates the The coordinates of the cargo location to which the goods are assigned; Step 4: Display the visualization results of the intelligent cargo space allocation plan on the human-computer interaction interface, including the location and cargo space number of each cargo task, and display the utilization rate and equipment load of the cargo space through dynamic charts, so that users can judge the intelligent cargo space allocation plan. Is adjustment needed? If so, the modified solution set is ,in Indicates the adjusted Otherwise, directly output the intelligent allocation plan ; Step 5: According to the final plan , generate the operation instructions of the physical layer equipment, and the corresponding storage equipment will perform the cargo allocation task; Step 6: By real-time monitoring of the status of goods and equipment, the execution results of the physical layer and the twin layer are fed back. If the goods are not placed in the correct location, the alarm is recorded and triggered. After the plan is executed, the data layer information is updated.

[0008] In step 2, the intelligent improved algorithm SA-NSGA-II includes the following steps: Step 2-1, randomly generate a population size of Initialized population , the first Chromosome individuals Middle genes Represents a cargo distribution task, encoding the three-dimensional coordinates of the gene into one-dimensional integers based on the gridding idea. , ; Step 2-2: Check the feasibility of the genes in the chromosome individuals to see if they meet the constraints. The constraints include: Storage location uniqueness constraint: , Shelf load-bearing capacity constraints: , Shelf capacity constraints: , in, 、 、 They are the number of rack rows, columns, and layers respectively; Indicates the The location coordinates of each item on the shelf; Indicates the The occupancy status of the cargo space allocated to each cargo; Indicates the total weight of all goods on a single shelf; Indicates the total weight of all goods on a single row of shelves; 、 Respectively represent the maximum load-bearing capacity of a single row of shelves and the maximum load-bearing capacity of a single layer of shelves; represents the set of positive integers; Step 2-3, set the gene repair operator to perform gene repair on chromosome individuals. Gene values ​​that do not satisfy the constraints Repaired, the repaired gene is ,in, Represents the gene repair operator, which is used to adjust the gene value in the chromosome so that the gene value in the chromosome meets the constraint conditions; Steps 2-4 calculate the fitness of each individual in the population in the cargo space allocation optimization objective function space and establish the following multi-objective cargo space optimization mathematical model: , in, Represents the overall optimization goal, which is to minimize the values ​​of multiple sub-goals; Indicates the balance of the shelf; Indicates the distance between the same type of goods in a single row of shelves; Indicates the time when the stacker is entering or leaving the warehouse; Establish the following criteria for measuring the balance of all shelves in automated warehouses: , in, is the width of the cargo space; Indicates the The quality of the goods; Use Manhattan distance to calculate the distance between the same type of goods in a single row of shelves: , in, is the cargo space height; Indicates the The location coordinates of the goods on the shelf, Goods and The goods are of the same type; Indicates the type of goods; Indicates the total number of goods of the same type; Indicates type The collection of all goods; The following objective function of the stacker's outbound and inbound time is established: , in, is the cargo space length; 、 、 Respectively represent the stacker fork speed, horizontal travel speed, and vertical travel speed; Indicates the Cargo turnover rate; Indicates the coordinates of the loading and unloading platform corresponding to the single row of shelves where the i-th item is located; Steps 2-5, for each chromosome individual in the population ,judge The dominance relationship between the chromosomes and other chromosomes is satisfied if In the sub-goal 、 、 is not inferior to other chromosomes in all aspects, and is strictly superior to other chromosomes in at least one sub-goal, then it is called Dominate other chromosomes, statistically dominate The number of chromosomes , and record Dominated chromosome set , find all chromosomes, The chromosomes are not dominated by any other solution and constitute the first frontier , removed from the population , continue to process the remaining chromosomes, repeating steps 2-5 until all chromosomes are assigned to the corresponding frontier ; Steps 2-6, in sub-goals 、 、 Randomly select any specific target , the non-dominated frontier, that is, the first frontier The chromosomes in the Sort the values ​​of , mark the boundary solutions of the target space as extreme solutions, and assign infinite congestion , for non-boundary solutions, calculate the sum of the normalized distances on all targets, the formula is: ; in, Represents chromosome individual the degree of congestion; f(x) represents the objective function f(x) represents the objective function f(x) represents the objective function f(x) represents the objective function f(x) represents the objective function f(x) represents the objective function f(x) represents the objective function f(x) represents the objective function f(x) represents the objective function f(x) represents the objective function f(x) represents the objective function f(x) represents the objective function Step 2-7, in the binary tournament selection, for each pair of chromosomes in the population , the chromosome with lower non-dominated level is selected preferentially, i.e. the individual in the smallest front in the front Step 2-8, using the order crossover operator, two parents are randomly selected, and the crossover points are determined, wherein , the genes between in the parent are directly copied to the corresponding positions in the child of the parent , the remaining genes in the parent are copied to the empty positions in the child, and the order of the parent genes is maintained, and the generation process of the child is: the genes between in the parent are directly copied to the corresponding positions in the child of the parent , the remaining genes in the parent are copied to the empty positions in the child, and the order of the parent genes is maintained; An adaptive strategy based on crowding degree and iteration number is used to calculate the crossover rate : , wherein , , are the minimum value, the maximum value and the average value of the given crossover rate respectively; , are the minimum value, the maximum value and the average value of the given crossover rate respectively; , are the minimum value, the maximum value and the average value of the given crossover rate respectively; , are the minimum value, the maximum value and the average value of the given crossover rate respectively;​​​​​​​ The crowding degree of each chromosome individual and the average crowding degree of the frontier where they are located; 、 are the current number of iterations and the maximum number of iterations respectively; Steps 2-9, in terms of mutation operations, use a combined mutation method that combines exchange mutation and single-point mutation. In exchange mutation, two gene positions are randomly selected from the current solution for exchange. In single-point mutation, the value of a gene position is randomly selected and replaced with a new random value. An adaptive strategy based on crowding and number of iterations is used to calculate the mutation rate. , realize the adaptive evolution of the population and increase the convergence of the population: , in, 、 、 are the minimum, maximum, and average values ​​of the given mutation rate, respectively; Step 2-10, introduce the simulated annealing SA algorithm into the offspring population. When performing the simulated annealing SA algorithm local search, 、 、 Randomly select any specific target Search as a search direction to generate new individuals: , in, For the new individuals produced; is the current solution; 、 They are The maximum and minimum solutions within the population; for A random number uniformly distributed within Represents a symbolic function, determined by the input value The symbol of the function, Used to control the disturbance direction, when hour is positive, otherwise is negative; is the initial temperature; is the temperature attenuation coefficient; is the current iteration number; for A random number uniformly distributed within By comparing the new and old solutions, the Metropolis criterion is implemented to determine whether the new individual should be retained and the retention probability The calculation formula is: , in, is a natural exponential function; Represents the new individual generated In the The objective function value on the sub-goals; Indicates the current solution In the The objective function value on the sub-goals; Step 2-11, executing steps 2-2 and 2-3 on the offspring population optimized by the simulated annealing SA algorithm; Step 2-12: The parent population selected by steps 2-5 and 2-6 And the offspring population generated by the genetic operations of steps 2-7, 2-8, and 2-9 Merge into a new candidate population ; Step 2-13, introduce congestion-based selection into the elite selection strategy and non-dominated classes Evolutionary ranking index : , according to For candidate populations To sort, select The chromosome individuals form a new generation of elite population , which enables the elite selection criteria to be dynamically adjusted over the evolutionary process; Step 2-14, iterate steps 2-2 to 2-13 until the maximum number of iterations is met , output the optimal Pareto front after iteration and decode the chromosome.

[0009] Step 2-3 includes the following steps: Step 2-3-1: Initialize an available set at the beginning of the repair , store all available cargo spaces; Step 2-3-2, check individual chromosomes one by one Each gene ,judge Whether the constraints in step 2-2 are met; Step 2-3-3, set the available collection Classification into two categories, including placeholder sets and the empty set ; Step 2-3-4, through statistics The frequency of genes in the test is used to determine whether there are repeated values. If the gene If the number of occurrences of the gene is greater than 1, it indicates that there is a duplicated gene in the chromosome, otherwise, skip the repair and directly output the current chromosome; Step 2-3-5, for each duplicated gene , find any element in the unoccupied vacancy set to replace the current duplicated gene , and update the occupancy set and the vacancy set after replacement; ; Step 2-3-6, if the repaired chromosome still has infeasible solutions or duplication problems, repeat steps 2-3-4~2-3-5 until the chromosome meets the constraint conditions; Step 2-3-7, after repair, output the final feasible chromosome that satisfies all genes are unique values and are in the available set .

[0010] Step 3 includes the following steps: Step 3-1, obtain the Pareto front solution set obtained by iteration of the SA-NSGA-II algorithm, and construct a multi-index evaluation system combined with , , three optimization objectives; Step 3-2, score each pair of evaluation indexes, and construct a fuzzy judgment matrix according to the judgment basis of pairwise comparison, where is the fuzzy evaluation value of index relative to index , expressed as a triangular fuzzy number ; , , represent the most conservative estimate, the most likely estimate, and the most optimistic estimate of the evaluator on index relative to index ; Step 3-3, calculate the fuzzy weight of index using the geometric mean method; Step 3-4, use the barycenter method to de-fuzzify the fuzzy and uncertain fuzzy weight into a direct actual weight value for decision-making, and obtain the weight of each evaluation index after normalization; Step 3-5, construct a decision matrix , where represents the th The scheme (i.e., chromosome individual) is in the The evaluation value of the indicator is Perform standardization to obtain the normalized decision matrix Among them, due to the evaluation index 、 、 are all cost-type indicators, that is, the smaller the target value, the better. The plan in Normalized evaluation value on the indicators ; 、 Indicators The maximum and minimum values ​​of Step 3-6, using the weights obtained in step 3-4 Normalize the results from steps 3-5 Perform weighting to obtain the weighted normalized decision matrix Among them, The plan in Weighted normalized evaluation value on the indicators ; Steps 3-7, due to the evaluation index 、 、 If they are all cost indicators, then the positive ideal solution is the minimum value among all the options on this indicator, and the negative ideal solution is the maximum value among all the options on this indicator. and negative ideal solutions ,in, Represents a set of cost-type indicators; Steps 3-8, calculate the difference between each solution and the positive ideal solution Euclidean distance , calculate the difference between each solution and the negative ideal solution Euclidean distance ; Step 3-9, calculate the relative closeness of each solution , sort the intelligent cargo location allocation plans according to the relative proximity. The greater the relative proximity, the better the plan, and the optimal cargo location allocation plan is output.

[0011] In steps 3-8, the following formula is used to calculate the difference between each solution and the positive ideal solution. Euclidean distance : , in, Indicates that for the indicator , this indicator in all schemes The optimal value of .

[0012] In steps 3-8, the following formula is used to calculate the difference between each solution and the negative ideal solution. Euclidean distance : , in, Indicates that for the indicator , this indicator in all schemes The worst value of .

[0013] In steps 3-9, the relative closeness of each solution is calculated using the following formula: : .

[0014] The present invention also provides an electronic device, comprising a processor and a memory, wherein the memory stores program code, and when the program code is executed by the processor, the processor executes the steps of the method.

[0015] The present invention also provides a storage medium storing a computer program or instruction, which executes the steps of the method when the computer program or instruction is run on a computer.

[0016] Compared with the existing technology, the advantages of the present invention include: (1) This method can perform real-time dynamic optimization. Through digital twin technology, the optimization of cargo space allocation can respond to changes in demand and environmental conditions in real time, achieving dynamic adjustment and efficient decision-making. Compared with traditional static optimization methods, the flexibility and adaptability of warehouse operations are greatly improved.

[0017] (2) The improved SA-NSGA-II algorithm can optimize multiple objectives simultaneously and output a diverse set of Pareto optimal solutions. At the same time, the FAHP-TOPSIS method is used to quantitatively analyze the pros and cons of each optimization solution, avoiding relying solely on experience or subjective judgment.

[0018] (3) The visualization capability of the digital twin model makes the optimization process of cargo space allocation more intuitive and transparent, providing a solution for the digital and intelligent transformation of warehouse management. At the same time, this method is not only applicable to general automated warehouses, but can also be extended to various scenarios such as multi-temperature warehouses, complex multi-layer warehouses, and unmanned warehouses. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, and the above and / or other advantages of the present invention will become more apparent.

[0020] Figure 1 This is a design diagram of the digital twin framework architecture of the method provided by the present invention.

[0021] Figure 2 This is a technical diagram of the cargo space allocation scheme of the method provided by the present invention.

[0022] Figure 3 This is the flow chart of the improved SA-NSGA-Ⅱ algorithm provided by the present invention.

[0023] Figure 4 Chromosome encoding and decoding diagram of the method provided by the present invention.

[0024] Figure 5 This is a crossover operator graph of the method provided by the present invention.

[0025] Figure 6 This is a mutation operator graph of the method provided by the present invention.

[0026] Figure 7 This is a flow chart of the gene repair method provided by the present invention.

[0027] Figure 8 The present invention provides a flow chart for combining the fuzzy analytic hierarchy process (FAHP) and the approach-to-ideal solution ranking method (TOPSIS) to determine the optimal cargo location allocation plan.

[0028] Figure 9 A three-dimensional comparison chart of the optimal Pareto frontier of the improved SA-NSGA-Ⅱ and traditional NSGA-Ⅱ and MOPSO algorithms provided by the present invention.

[0029] Figure 10 Schematic diagram of the cargo space allocation results obtained by the MOPSO algorithm.

[0030] Figure 11 This is a schematic diagram of the cargo space allocation results obtained by the NSGA-Ⅱ algorithm.

[0031] Figure 12 It is a schematic diagram of the cargo space allocation result obtained by the improved SA-NSGA-Ⅱ method of the present invention.

[0032] Figure 13 This is an example diagram of the digital twin warehouse implementation of the method provided by the present invention. DETAILED DESCRIPTION

[0033] Reference Figure 1 , an embodiment of the present invention provides a method for allocating cargo space in an automated high-bay warehouse driven by digital twins, including: constructing a five-dimensional digital twin model, including a physical layer, a twin layer, a data layer, a service layer and connections, covering the entire process from physical data collection to virtual twins to decision feedback.

[0034] The physical layer is the actual place of the goods access operation, represents the actual physical entities and environment in the system, including physical entities such as shelves, stackers, and in-out warehouse tables; the data of the physical layer is uploaded to the data layer through real-time interaction, and at the same time receives optimization instructions from the service layer; The twin layer is a digital mirror of the physical space, which provides simulation support for warehouse operation and space allocation by constructing a virtual-real mapping twin model of the physical warehouse; the warehouse digital twin model relies on three-dimensional modeling technology to construct accurate three-dimensional models in Blender software according to the shape, color, material, and other characteristics of physical entities, and ensures that the virtual model is highly consistent with the real warehouse environment through high-quality rendering, and sets up the model scene tree according to the parent-child relationship information between three-dimensional models, and outputs the three-dimensional model file in the format of.gltf or.glb; the Three.js engine is used to load and parse the three-dimensional model in the format of.gltf or.glb on the Web, combined with the environmental data collected by IoT devices, to realize real-time mapping of warehouse layout, device state and environmental parameters, and dynamically bind the warehouse data through the Vue3 framework to ensure real-time updating of the number of goods on the shelves, the position and running state of the stacker; all geometric models, dynamic data and rendering scenes are integrated to build a complete warehouse digital twin system. The data layer, as a data support layer connecting the physical layer and the twin layer, obtains dynamic data and static data, cleans and stores the data, and provides input for the service layer to ensure the real-time and accuracy of the system, wherein the dynamic data is used to reflect the real-time operation of the automated warehouse, including goods transfer data, device state, environmental data, etc., and the static data is the basic data such as warehouse geometric dimensions, device physical constraints, and goods material properties.

[0035] The service layer provides an intuitive operation interface and scientific decision support for users, realizes intelligent decision-making and manual decision-making for space allocation.

[0036] The connection includes data connection and control connection, the data connection is used to collect physical layer data to the data layer through IoT devices (English: Internet of Things, abbreviation: IoT) and network devices, the information processed by the data layer is input to the twin layer for modeling and simulation, and to the service layer for generating space allocation scheme; the control connection is used to feed back the space allocation instructions to the physical layer to guide the actual operation, and to the twin layer for virtual operation.

[0037] Reference Figure 2 The specific steps of the method of combining intelligent decision-making with manual decision-making are as follows: Step 1, obtaining the space state, goods characteristics and device state information processed from the data layer; Step 2, generate a set of optional slot allocation schemes using the improved algorithm SA-NSGA-II ; Step 3, use the decision-making method combining fuzzy analytic hierarchy process (FAHP) and technique for order preference by similarity to ideal solution (TOPSIS) to select the optimal slot allocation scheme from the set of slot allocation schemes ; Step 4, display the visual results of the intelligent slot allocation scheme on the human-computer interaction interface, including the location and slot number of each cargo task, and display the utilization rate of the slot and the load of the equipment through dynamic charts. The user judges whether the intelligent slot allocation scheme needs to be adjusted. If it needs to be adjusted, the modified scheme set is , where represents the slot allocation scheme of the th cargo after adjustment, otherwise, the intelligent allocation scheme is directly output ; ; Step 5, generate operation instructions for physical layer devices according to the final scheme , and execute the slot allocation task by the corresponding warehouse equipment. Step 6, monitor the status of goods and equipment in real time, and feedback the execution results of the physical layer and the twin layer. If the goods are not placed in the correct position, record and trigger an alarm. After the scheme is executed, update the data layer information.

[0038] Referring to Figure 3 , to solve the problems of traditional NSGA-II algorithm in complex practical applications, such as significant increase in computing resource consumption, slow convergence speed, poor maintenance of population diversity, etc., an improved SA-NSGA-II algorithm is proposed to optimize the slot allocation problem. Step 2 includes the following steps: Step 2-1, randomly generate an initial population with a population size of , the th chromosome individual in the population , the th gene in the chromosome individual represents a cargo allocation task. According to the grid idea, the three-dimensional coordinates of the gene are encoded into one-dimensional integers. Step 2-2, check the feasibility of the genes in the chromosome individual, and monitor whether the individual meets the constraint conditions, including: Slot uniqueness constraint: , Rack load capacity constraint: , Rack capacity constraint: , Rack capacity constraint: Step 2-3, set the gene repair operator, repair the chromosome individual gene values that do not meet the constraint conditions Repair the repaired gene ; Step 2-4, calculate the fitness of each individual in the population in the space of the optimization objective function of the storage location, and establish the following multi-objective storage optimization mathematical model: , The following criteria are established to measure the balance of all shelves in the automated warehouse: , The Manhattan distance is used to calculate the distance between the same type of goods in a single row of shelves: , The following objective function is established for the out-of-warehouse and into-warehouse time of the stacker: , Step 2-5, for each chromosome individual in the population , judge the dominance relationship with other chromosomes, if it meets is not worse than other chromosomes in all sub-objective 、 、 , and strictly better than other chromosomes in at least one sub-objective, it is said to dominate other chromosomes, count the number of dominated chromosomes , and record the dominated chromosome set , find all chromosomes , the chromosomes are not dominated by any other solution, which constitutes the first front , remove from the population, continue to process the remaining chromosomes, and repeat the above process until all chromosomes are assigned to the corresponding front ; Step 2-6, randomly select any specific objective in sub-objective 、 、 , , sort the chromosomes in the non-dominated front, i.e. the first front , according to the value of the selected objective function , mark the boundary solution of the objective space as the extreme solution, and give infinite crowding degree , for non-boundary solutions, calculate the sum of the normalized distances in all objectives, the formula is: ; Step 2-7, in the binary tournament selection, for each pair of chromosomes in the population , the chromosome with lower non-dominated rank is selected, i.e. the chromosome in the front is selected The chromosome with the smallest front is selected and if they belong to the same non-dominated rank, the chromosome with larger crowding distance is selected; Step 2-8, using the order crossover operator, two parents , are randomly selected, the crossover points and are determined, wherein , the genes between in the parent are directly copied to the corresponding positions in the child of the parent , the remaining genes are copied from the parent to the empty positions in the child , and the order of the parent genes is maintained, the generation process of the child is similar to , the genes between in the parent are directly copied to the corresponding positions in the child of the parent , the remaining genes are copied from the parent to the empty positions in the child , and the order of the parent genes is maintained.

[0039] An adaptive strategy based on crowding degree and iteration number is used to calculate the crossover rate : , Step 2-9, in terms of mutation operation, a combined mutation method combining exchange mutation and single-point mutation is used, in exchange mutation, two gene positions are randomly selected from the current solution for exchange, in single-point mutation, a gene position value is randomly selected to replace the selected gene position with a new random value, an adaptive strategy based on crowding degree and iteration number is used to calculate the mutation rate , to realize adaptive evolution of the population and increase the convergence of the population: .

[0040] Step 2-10, introduce the simulated annealing SA algorithm in the child population, when performing local search of the simulated annealing SA algorithm, in the sub-target , , Randomly select any specific target Search as a search direction to generate new individuals: , By comparing the new and old solutions, the Metropolis criterion is implemented to determine whether the new individual should be retained and the retention probability for: .

[0041] Step 2-11, performing the feasibility check of step 2-2 and the gene repair of step 2-3 on the offspring population optimized by the simulated annealing SA algorithm; Step 2-12, the parent population selected by the fast non-dominated sorting in step 2-5 and the crowding calculation in step 2-6 And the offspring population generated by the genetic operations of step 2-7 selection, step 2-8 mutation, and step 2-9 crossover Merge into a new candidate population ; Step 2-13, introduce congestion-based selection into the elite selection strategy and non-dominated classes Evolutionary ranking index : , according to For candidate populations To sort, select Individuals form a new generation of elite populations , which enables the elite selection criteria to be dynamically adjusted over the evolutionary process; Step 2-14, iteratively execute steps 2-2 to 2-13 until the maximum number of iterations is met , output the optimal Pareto front after iteration and decode the chromosome.

[0042] Reference Figure 4 In the encoding design, each integer represents the location number of a cargo allocation plan, and the chromosome length is equal to the number of tasks. Row, List, The shelf of the first layer The cargo location code assigned to each cargo is , the three-dimensional coordinates of the allocated cargo location are: , in, is the rounding operator, Modulo operator. Taking a 12-row, 12-column, 10-layer shelf as an example, there are 1440 available storage locations in total, and if there are 8 goods to be allocated, each of the goods is randomly allocated to an idle storage location, and the goods 1 is allocated to the storage location No. 47, after decoding, the three-dimensional coordinates of the storage location to which the goods 1 is allocated are (1, 11, 4), and so on.

[0043] Referring to Figure 5 , in order to retain the excellent genes of the parent individuals and avoid illegal combinations, the order crossover operator is used to cross the task sequence allocated according to the storage location. Taking 8 tasks as an example: Randomly select a pair of chromosomes to determine the crossover point position , : Parent 1: 47, 234, 345, 149, 154, 122, 277, 93; Parent 2: 88, 1, 47, 288, 349, 149, 52, 206; The genes between the parent chromosome 1 are directly copied to the corresponding positions of the offspring chromosome 1, and the remaining genes are copied from the parent chromosome 2 to the vacancies in the offspring chromosome 1, and the order of the parent genes is maintained; the parent chromosome 2 is the same, the genes between the parent chromosome 2 are directly copied to the corresponding positions of the offspring chromosome 2, and the remaining genes are copied from the parent chromosome 1 to the vacancies in the offspring chromosome 2, and the order of the parent genes is maintained: Offspring 1: 88, 1, 345, 149, 154, 47, 288, 349; Offspring 2: 234, 345, 47, 288, 349, 149, 154, 122; Referring to Figure 6 , in order to improve the diversity of the solution and reduce the risk of falling into a local optimal solution, in the mutation operation, a combination mutation method combining exchange mutation and single-point mutation is used. Taking the 8 tasks in the offspring 2 generated by the crossover as an example: First, randomly select a gene position , Exchange its gene: Offspring 2: 88, 1, 345, 149, 154, 47, 288, 349; Offspring 2 after exchange mutation: 88, 47, 345, 149, 154, 1, 288, 349; Randomly select a gene position Replace it with a new random value: Single point mutation offspring 2: 88, 47, 345, 286, 154, 1, 288, 349; Referring to Figure 7 , the optimization algorithm has randomness in the process of performing selection, crossover, mutation and the like, which may cause part of individuals in the generated chromosome to not meet the constraint condition. In order to ensure the feasibility of the offspring, a gene repair operator is set. Step 2-3 includes the following steps: Step 2-3-1, at the beginning of repair, initialize an available set , which stores all available locations; Step 2-3-2, check each gene in the chromosome individual one by one to determine whether it meets the constraint condition in step 2-2; Step 2-3-3, classify the available set into two categories, including an occupied set and an empty set ; Step 2-3-4, determine whether there is a repeated value by counting the frequency of the genes . If the number of occurrences of a gene is greater than 1, it means that there is a repeated gene in the chromosome, otherwise skip the repair and directly output the current chromosome; Step 2-3-5, for each repeated gene , find any element in the unoccupied empty set to replace the current repeated gene , and update the occupied set and the empty set after replacement; Step 2-3-6, if the repaired chromosome still has unfeasible solutions or repetition problems, repeat steps 2-3-4 to 2-3-5 until the chromosome meets the constraint condition; Step 2-3-7, after the repair is completed, output the final feasible chromosome , which meets the condition that all genes are unique values and are in the available set .

[0044] Referring to Figure 8 , after obtaining a set of optimal Pareto solutions, a decision method combining fuzzy analytic hierarchy process (FAHP) and technique for order preference by similarity to ideal solution (TOPSIS) is used to select a set of optimal location allocation schemes from multiple allocation schemes as the final scheme. Step 3 includes the following steps: Step 3-1, obtain the Pareto frontier solution set obtained by the iteration of SA-NSGA-Ⅱ algorithm, combined with 、 、 Three optimization objectives are used to construct a multi-index evaluation system; Step 3-2: Score the evaluation indicators pairwise and construct a fuzzy judgment matrix based on the pairwise comparison judgment basis. , triangular fuzzy number The scale values ​​1-9 and their corresponding meanings are shown in Table 1.

[0045] Table 1

[0046]

[0047] Step 3-3, calculate the indicator using the geometric mean method Fuzzy weighted ; Step 3-4, use the centroid method to convert the fuzzy and uncertain fuzzy weights Defuzzification into an actual weight value that is directly used for decision making , after normalization, the weight of each evaluation index is obtained ; Steps 3-5: Construct a decision matrix , for the decision matrix Perform standardization to obtain the normalized decision matrix ; Step 3-6, using the weights obtained in step 3-4 Normalize the results from steps 3-5 Perform weighting to obtain the weighted normalized decision matrix ; Steps 3-7, due to the evaluation index 、 、 If they are all cost indicators, then the positive ideal solution is the minimum value among all the options on this indicator, and the negative ideal solution is the maximum value among all the options on this indicator. and negative ideal solutions ; Steps 3-8, calculate the difference between each solution and the positive ideal solution Euclidean distance , calculate the difference between each solution and the negative ideal solution Euclidean distance ; Step 3-9, calculate the relative closeness of each solution , sort the intelligent cargo location allocation plans according to relative proximity. The greater the relative proximity, the better the plan, and output the optimal cargo location allocation plan Pareto.

[0048] Referring to Figure 9 , in order to verify the applicability and effectiveness of the improved SA-NSGA-II algorithm, simulation experiments were carried out on the data of 600 finished cloth warehouse in April 2024 of a textile company. The data parameters of the top 20 turnover box numbers are shown in Table 2.

[0049] Table 2

[0050]

[0051] Among them, the basic parameters of the automated three-dimensional warehouse are shown in Table 3.

[0052] Table 3

[0053]

[0054] Considering the problem size, MOPSO, NSGA-II and improved SA-NSGA-II algorithm were used for optimization, and the parameter settings of the three algorithms are shown in Table 4.

[0055] Table 4

[0056]

[0057] From the population distribution of different dimensions after iteration of MOPSO, NSGA-II and improved SA-NSGA-II optimization method, compared with the initial population, MOPSO, NSGA-II and improved SA-NGSA-II algorithm have significantly improved in population quality, and the values of the three objective functions are significantly smaller; compared with traditional NSGA-II and MOPSO algorithm, improved SA-NSGA-II algorithm performs better in three objective functions, the position of Pareto solution is more competitive and the uniformity of distribution is better, which proves the effectiveness of improved SA-NSGA-II algorithm.

[0058] Referring to Figure 10 , Figure 11 and Figure 12 , in order to make the optimal location allocation scheme, the fuzzy judgment matrix constructed is shown in Table 5.

[0059] Table 5

[0060]

[0061] The objective function , , corresponding to the fuzzy weight is , , , after defuzzification 、 、 , the comprehensive weights of the three objective functions after normalization are 、 、 .

[0062] By approximating the ideal solution ranking method TOPSIS Figure 9 The obtained Pareto solution set is comprehensively sorted, among which the positive ideal solution is , the negative ideal solution is ,The optimal Pareto solution after comprehensive sorting is shown in Table 6.

[0063] Table 6

[0064]

[0065] To more intuitively demonstrate the results of optimized cargo location allocation, the optimal cargo location allocation results were visualized. As shown in the figure, the optimal Pareto solutions of the MOPSO and NSGA-II algorithms have obvious flaws, including unreasonable cargo locations, which have a certain impact on the efficiency of stacker cranes in and out of the warehouse. The simulation results of the SA-NSGA-II algorithm are stable, once again demonstrating its superiority in cargo location optimization.

[0066] Reference Figure 13 Based on the method, a simulation platform with virtual-reality synchronization is constructed, which enables real-time communication and action synchronization between physical equipment and twin models. The final allocation results will be presented in the three-dimensional visual interactive interface as shown in the figure. The system will intuitively display multiple key indicators within the warehouse to operators through multi-dimensional visual charts, which can effectively rely on data application capabilities to improve warehouse operation efficiency and transparency.

[0067] This invention provides a method for allocating cargo space in an automated warehouse using digital twins. While there are numerous methods and approaches for implementing this technical solution, the foregoing is merely a preferred embodiment of the invention. It should be noted that those skilled in the art may make improvements and modifications without departing from the principles of the invention, and such improvements and modifications are considered within the scope of protection of this invention. Any components not specified in this embodiment may be implemented using existing technologies.

Claims

1. A method for allocating cargo space in an automated warehouse driven by digital twins, characterized in that: include: Build a five-dimensional digital twin model, including the physical layer, twin layer, data layer, service layer, and connection; The physical layer is the actual location where goods are stored and retrieved, representing the actual physical entities and environment. These physical entities include shelves, stackers, access platforms, turnover boxes, and goods. Data from the physical layer is uploaded to the data layer through real-time interaction, while also receiving optimization instructions from the service layer. The twin layer is a digital mirror of the physical space. By building a warehouse digital twin model that maps the virtual and real space with the physical warehouse, it provides simulation support for cargo space allocation and warehouse operations. The data layer serves as the data support layer connecting the physical layer and the twin layer. It obtains dynamic data and static data, cleans and stores the data, and provides input for the service layer. Dynamic data is used to reflect the real-time operation of the automated warehouse, including cargo flow data, equipment status, and environmental data. Static data includes warehouse geometry, equipment physical constraints, and cargo material properties. The service layer provides users with a human-computer interaction interface and decision support, realizing the combination of intelligent decision-making and manual decision-making for cargo space allocation; The connection includes a data connection and a control connection. The data connection is used to collect physical layer data to the data layer through IoT devices and network devices. The information processed by the data layer is input into the twin layer for modeling and simulation, and input into the service layer for generating a cargo space allocation plan; the control connection is used to feed back cargo space allocation instructions to the physical layer to guide actual operations, and to the twin layer for virtual operations.

2. The method according to claim 1, characterized in that The service layer performs the following steps: Step 1: Obtain the cargo location status, cargo characteristics, and equipment status information after processing from the data layer; Step 2: Use the intelligent improved algorithm SA-NSGA-Ⅱ to generate a set of optional cargo space allocation solutions ,in represents the Ith intelligent cargo location allocation plan; Step 3: Use the decision-making method combining fuzzy analytic hierarchy process (FAHP) and TOPSIS to select the best solution from the set of cargo space allocation solutions. Select the best intelligent cargo space allocation solution ,in Indicates the The coordinates of the cargo location to which the goods are assigned; Step 4: Display the visualization results of the intelligent cargo space allocation plan on the human-computer interaction interface, including the location and cargo space number of each cargo task, and display the utilization rate and equipment load of the cargo space through dynamic charts, so that users can judge the intelligent cargo space allocation plan. Is adjustment needed? If so, the modified solution set is ,in Indicates the adjusted Otherwise, directly output the intelligent allocation plan ; Step 5: According to the final plan , generating the operation instructions of the physical layer equipment, and the storage equipment performs the cargo allocation task; Step 6: By real-time monitoring of the status of goods and equipment, the execution results of the physical layer and the twin layer are fed back. If the goods are not placed in the correct location, the alarm is recorded and triggered. After the plan is executed, the data layer information is updated.

3. The method according to claim 2, characterized in that In step 2, the intelligent improved algorithm SA-NSGA-II includes the following steps: Step 2-1, randomly generate a population size of Initialized population , the first Chromosome individuals Middle genes Represents a cargo distribution task, encoding the three-dimensional coordinates of the gene into one-dimensional integers based on the gridding idea. , ; Step 2-2: Check the feasibility of the genes in the chromosome individuals to see if they meet the constraints. The constraints include: Storage location uniqueness constraint: , Shelf load-bearing capacity constraints: , Shelf capacity constraints: , in, 、 、 They are the number of rack rows, columns, and layers respectively; Indicates the The location coordinates of each item on the shelf; Indicates the The occupancy status of the cargo space allocated to each cargo; Indicates the total weight of all goods on a single shelf; Indicates the total weight of all goods on a single row of shelves; 、 Respectively represent the maximum load-bearing capacity of a single row of shelves and the maximum load-bearing capacity of a single layer of shelves; represents the set of positive integers; Step 2-3, set the gene repair operator to perform gene repair on chromosome individuals. Gene values ​​that do not satisfy the constraints Repaired, the repaired gene is ,in, represents the gene repair operator; Steps 2-4 calculate the fitness of each individual in the population in the cargo space allocation optimization objective function space and establish the following multi-objective cargo space optimization mathematical model: ， in, Represents the overall goal of optimization; Indicates the balance of the shelf; Indicates the distance between the same type of goods in a single row of shelves; Indicates the time when the stacker is entering or leaving the warehouse; Establish the following criteria for measuring the balance of all shelves in automated warehouses: ， in, is the width of the cargo space; Indicates the The quality of the goods; Use Manhattan distance to calculate the distance between the same type of goods in a single row of shelves: ， in, is the cargo space height; Indicates the The location coordinates of the goods on the shelf, Goods and The goods are of the same type; Indicates the type of goods; Indicates the total number of goods of the same type; Indicates type The collection of all goods; The following objective function of the stacker's outbound and inbound time is established: ， in, is the cargo space length; 、 、 Respectively represent the stacker fork speed, horizontal travel speed, and vertical travel speed; Indicates the Cargo turnover rate; Indicates the coordinates of the loading and unloading platform corresponding to the single row of shelves where the i-th item is located; Steps 2-5, for each chromosome individual in the population ,judge The dominance relationship between the chromosomes and other chromosomes is satisfied if In the sub-goal 、 、 is not inferior to other chromosomes in all aspects, and is strictly superior to other chromosomes in at least one sub-goal, then it is called Dominate other chromosomes, statistically dominate The number of chromosomes , and record Dominated chromosome set , find all chromosomes, The chromosomes are not dominated by any other solution and constitute the first frontier , removed from the population , continue to process the remaining chromosomes, repeating steps 2-5 until all chromosomes are assigned to the corresponding frontier ; Steps 2-6, in sub-goals 、 、 Randomly select any specific target , the non-dominated frontier, that is, the first frontier The chromosomes in the Sort the values ​​of , mark the boundary solutions of the target space as extreme solutions, and assign infinite congestion , for non-boundary solutions, calculate the sum of the normalized distances on all targets, the formula is: ； in, Represents chromosome individual the degree of congestion; Represents the objective function In the individual The next adjacent individual The value of Represents the objective function In the individual The previous adjacent individual The value of 、 Represent the objective function At the current non-dominated frontier The maximum and minimum values ​​in ; Steps 2-7, in binary tournament selection, for each pair of chromosomes in the population , give priority to chromosomes with low non-dominated levels, that is, give priority to the frontier middle Minimal frontier If the individual and If they belong to the same non-dominated class, the individual with the larger crowding degree is selected; Step 2-8, use the sequential crossover operator to randomly select two parents 、 , determine the intersection and ,in , the parent Located in the middle The genes between them are directly copied to the parent Offspring of The corresponding position within the parent Replicate the remaining genes to offspring The vacancies in the parent gene sequence are kept, and the offspring The generation process is: the parent Located in the middle The genes between them are directly copied to the parent Offspring of The corresponding position within the parent Replicate the remaining genes to offspring The vacancies within the parent gene are filled and the order of the parent gene is maintained; Adopting an adaptive strategy based on crowding and number of iterations to calculate the crossover rate : ， in, 、 、 are the minimum, maximum, and average values ​​of the given crossover rate, respectively; 、 Respectively The crowding degree of each chromosome individual and the average crowding degree of the frontier; 、 are the current number of iterations and the maximum number of iterations respectively; Steps 2-9, in terms of mutation operations, use a combined mutation method that combines exchange mutation and single-point mutation. In exchange mutation, two gene positions are randomly selected from the current solution for exchange. In single-point mutation, the value of a gene position is randomly selected and replaced with a new random value. An adaptive strategy based on crowding and number of iterations is used to calculate the mutation rate. , realize the adaptive evolution of the population and increase the convergence of the population: ， in, 、 、 are the minimum, maximum, and average values ​​of the given mutation rate, respectively; Step 2-10, introduce the simulated annealing SA algorithm into the offspring population. When performing the simulated annealing SA algorithm local search, 、 、 Randomly select any specific target Search as a search direction to generate new individuals: ， in, For the new individuals produced; is the current solution; 、 They are The maximum and minimum solutions within the population; for A random number uniformly distributed within Represents a symbolic function, determined by the input value The symbol of the function, Used to control the disturbance direction, when hour is positive, otherwise is negative; is the initial temperature; is the temperature attenuation coefficient; is the current iteration number; for A random number uniformly distributed within By comparing the new and old solutions, the Metropolis criterion is implemented to determine whether the new individual should be retained and the retention probability The calculation formula is: ， in, is a natural exponential function; Represents the new individual generated In the The objective function value on the sub-goals; Indicates the current solution In the The objective function value on the sub-goals; Step 2-11, executing steps 2-2 and 2-3 on the offspring population optimized by the simulated annealing SA algorithm; Step 2-12: The parent population selected by steps 2-5 and 2-6 And the offspring population generated by the genetic operations of steps 2-7, 2-8, and 2-9 Merge into a new candidate population ; Step 2-13, introduce congestion-based selection into the elite selection strategy and non-dominated classes Evolutionary ranking index : ， according to For candidate populations To sort, select Chromosome individuals form a new generation of elite population , which enables the elite selection criteria to be dynamically adjusted over the evolutionary process; Step 2-14, iterate steps 2-2 to 2-13 until the maximum number of iterations is met , output the optimal Pareto front after iteration and decode the chromosome.

4. The method according to claim 3, characterized in that Step 2-3 includes the following steps: Step 2-3-1: Initialize an available set at the beginning of the repair , store all available cargo spaces; Step 2-3-2, check individual chromosomes one by one Each gene ,judge Whether the constraints in step 2-2 are met; Step 2-3-3, set the available Classification into two categories, including placeholder sets and the empty set ; Step 2-3-4, through statistics The frequency of genes in the test is used to determine whether there are repeated values. If the gene If the number of occurrences is greater than 1, it means that there are duplicate genes in the chromosome. Otherwise, the repair is skipped and the current chromosome is output directly. Steps 2-3-5, for each repeated gene , find the set of unoccupied slots Any element in , used to replace the current duplicate gene , update the placeholder collection after the replacement is completed and the empty set ; Step 2-3-6: If the repaired chromosome still has infeasible solutions or duplication problems, repeat steps 2-3-4 to 2-3-5 until the chromosome meets the constraints; Step 2-3-7: After the repair is completed, the final feasible chromosome is output , satisfying all genes All are unique values ​​and are in the available set middle.

5. The method according to claim 4, characterized in that Step 3 includes the following steps: Step 3-1, obtain the Pareto frontier solution set obtained by the iteration of SA-NSGA-Ⅱ algorithm, combined with 、 、 Three optimization objectives are used to construct a multi-index evaluation system; Step 3-2: Score the evaluation indicators pairwise and construct a fuzzy judgment matrix based on the pairwise comparison judgment basis. ,in, is an indicator Relative to the indicator The fuzzy evaluation value is expressed as a triangular fuzzy number ; 、 、 Represents the evaluator's evaluation of the indicators Relative to the indicator The most conservative estimate, the most likely estimate, and the most optimistic estimate; Step 3-3, calculate the indicator using the geometric mean method Fuzzy weight ; Step 3-4, use the centroid method to convert the fuzzy and uncertain fuzzy weights Defuzzification into an actual weight value that is directly used for decision making , after normalization, the weight of each evaluation index is obtained ; Steps 3-5: Construct a decision matrix ,in Indicates the The plan in The evaluation value of the indicator is Perform standardization to obtain the normalized decision matrix Among them, The plan in Normalized evaluation value on the indicators ; 、 Indicators The maximum and minimum values ​​of Step 3-6, using the weights obtained in step 3-4 Normalize the results from steps 3-5 Perform weighting to obtain the weighted normalized decision matrix Among them, The plan in Weighted normalized evaluation value on the indicators ; Steps 3-7, determine the positive ideal solution and negative ideal solutions ,in, Represents a set of cost-type indicators; Steps 3-8, calculate the difference between each solution and the positive ideal solution Euclidean distance , calculate the difference between each solution and the negative ideal solution Euclidean distance ; Step 3-9, calculate the relative closeness of each solution , sort the intelligent cargo location allocation plans according to the relative proximity. The greater the relative proximity, the better the plan, and the optimal cargo location allocation plan is output.

6. The method according to claim 5, characterized in that In steps 3-8, the following formula is used to calculate the difference between each solution and the ideal solution. Euclidean distance : ， in, Indicates that for the indicator , this indicator in all schemes The optimal value of .

7. The method according to claim 6, characterized in that In steps 3-8, the following formula is used to calculate the difference between each solution and the negative ideal solution. Euclidean distance : ， in, Indicates that for the indicator , this indicator in all schemes The worst value of .

8. The method according to claim 7, characterized in that In steps 3-9, the relative closeness of each solution is calculated using the following formula: : .

9. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores program codes, and when the program codes are executed by the processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 8.

10. A storage medium, characterized in that: A computer program or instruction is stored, and when the computer program or instruction is run on a computer, the steps of the method according to any one of claims 1 to 8 are executed.

Images