Intelligent warehouse shelf return position allocation method based on shelf heat and correlation degree

Abstract

The application discloses a kind of based on intelligent warehouse shelf return position allocation method of shelf heat and correlation degree, the method includes the following steps: one: collate task shelf historical order data, calculate task shelf heat value and correlation degree value;Two: the intelligent warehouse is gridded, extraction task shelf, workstation, mobile robot data;Three: collate shelf distributable position candidate set, according to demand design shelf placement position allocation model;Four: solve optimal shelf placement position allocation scheme;Five: using mobile robot task allocation strategy carries out mobile robot task allocation and identifies whether there is shelf deadlock conflict;Six: using the greedy algorithm of fusion mobile robot task allocation strategy, solve optimal mobile robot carrying scheme.The method considers two factors of shelf heat and shelf correlation degree, solves optimal shelf placement position allocation scheme and mobile robot carrying scheme using intelligent algorithm, can adapt to order demand fluctuation and improve picking efficiency.

Description

Technical Field

[0001] This invention relates to a method for allocating shelf return locations, specifically a smart warehouse shelf return location allocation method based on shelf popularity and relevance, which can allocate shelf placement positions during the picking or replenishment operation return process according to shelf popularity and shelf relevance. Background Technology

[0002] Smart warehouse picking offers advantages such as low labor costs and high accuracy. However, when faced with e-commerce customer orders characterized by small batches and frequent deliveries, as well as fluctuating and cyclical order demands, efficient and accurate picking has become a pressing challenge. Existing methods for allocating shelf return positions in smart warehouses, where picking and replenishment workstations are shared, do not consider the entire task cycle and cannot better adapt to fluctuations in order demand. Therefore, there is an urgent need to develop a new method for allocating shelf return positions in smart warehouses. Summary of the Invention

[0003] To address the issue that existing smart warehouse rack return location allocation methods, which share picking and replenishment workstations, do not consider the entire task cycle and cannot effectively adapt to fluctuations in order demand, this invention provides a smart warehouse rack return location allocation method based on rack popularity and relevance. This method considers both rack popularity and rack relevance, and utilizes intelligent algorithms to solve for the optimal rack placement allocation scheme and mobile robot handling scheme throughout the entire task cycle. This approach can adapt to fluctuations in order demand and improve picking efficiency.

[0004] The objective of this invention is achieved through the following technical solution:

[0005] A smart warehouse rack return location allocation method based on rack popularity and relevance, such as... Figure 1 As shown, it includes the following steps:

[0006] Step 1: Organize historical order data for the task shelf and calculate the task shelf popularity and relevance values. Specific steps are as follows:

[0007] Organize the historical order data of the task shelf, find the historical popularity of the task shelf, order data, and workbench access data, and substitute them into the following formulas for calculating shelf popularity and shelf relevance:

[0008]

[0009]

[0010] in, For the updated number Individual shelf popularity; For the first Historical popularity of each shelf; For the first the number of times of history of being picked and carried by the shelf; the time interval of picking / centralized replenishment for the warehouse; the shelf correlation coefficient; the task shelf , , the total number of task shelves; the workbench , , the total number of picking / replenishment workbenches; the task shelf and the task shelf the number of times of history of being served by the same picking station.

[0011] Step two: grid the smart warehouse, extract task shelf, workbench, mobile robot data. The specific steps are as follows:

[0012] Grid the smart warehouse, extract task shelf set , task shelf position , workbench set , workbench position , mobile robot set , mobile robot speed , mobile robot position , shelf set, picking / replenishment workbench set, mobile robot set, the mobile robot , , the total number of mobile robots.

[0013] Step three: arrange the shelf assignable position candidate set, and design the shelf placement position distribution model according to the demand. The specific steps are as follows:

[0014] Step three one: determine the open shelf position in the smart warehouse, combine the open shelf position with the task shelf position to form the shelf assignable position candidate set , shelf placement position set, total number of shelf placement positions.

[0015] Step three two: design the following shelf placement position distribution model according to the demand:

[0016] (1)

[0017] Constraints:

[0018] (2)

[0019] (3)

[0020] (4)

[0021] (5)

[0022] wherein, the first term of formula (1) represents the product of the shelf correlation degree and the interval, the second term represents the ratio of the shelf heat and the distance from the assigned position to the workbench, is the weight of the first term proportion; is the interval between two shelves after the shelf position is assigned; represents the distance between the shelf position and the workbench ; is a larger value; is the coordinate of the shelf assigned position ; represents whether the shelf is stored in the shelf assigned position . The first term of formula (1) quantifies the shelf correlation degree factor. High correlation degree shelves are placed in the same storage area, which is beneficial to reduce the travel distance of the mobile robot. However, when the number of mobile robots is large, this area will also become a high-frequency operation area. It is easy to cause mobile robot congestion, increase the number of mobile robot start-stop and turning, and cause operation efficiency to decrease and mobile robot energy consumption to increase. Therefore, the first term of the strategy design means that high correlation degree shelves are grouped to positions far apart in the horizontal direction to avoid forming a high-frequency action area and reduce robot congestion; the second term of formula (1) quantifies the shelf heat factor. High heat shelves are assigned to positions close to the picking station, which can effectively improve the picking efficiency.

[0023] Step three: analyze the demand for intelligent warehouse shelf placement position and the conditions of intelligent warehouse equipment, design the weight value to form a shelf placement position assignment model.

[0024] Step four: design an intelligent genetic algorithm to solve the optimal shelf placement position assignment scheme. The specific steps are as follows:

[0025] Step four one: adopt real number coding method, the gene position on the chromosome represents the shelf number, the length is the total number of shelves, and the gene represents the shelf assigned position.

[0026] The chromosome coding rule is: the shelves in the front row of the heat ranking are randomly matched with the shelves in the front row of the workbench distance ranking, and then the remaining shelves are randomly matched with the remaining shelf placement positions to form a complete chromosome.

[0027] Step four two: the fitness function is designed as:

[0028] (6)

[0029] Step four three: the crossover operation rule is designed using the traditional selection method: to avoid the same shelf placement position from appearing repeatedly in the same chromosome, a chromosome pair is searched before the crossover to form two unique gene groups without the same genes. A gene in the unique gene group of the chromosome pair is randomly selected for crossover each time to generate a new group of chromosomes.

[0030] Step four four: the mutation operation rule is designed: to ensure that the shelf position corresponds to the shelf one by one, a number of special shelf placement positions different from the chromosome genes are generated. A special shelf placement position is randomly selected and replaced with any gene on the chromosome to generate a new offspring. As the number of iterations increases, the differences between chromosomes will become smaller and smaller, and the number of individuals formed by crossover will also become smaller and smaller. To ensure the total number of chromosomes participating in the mutation process, the high fitness value chromosomes in the population will be used for filling.

[0031] Step four five: set the maximum number of iterations, if the iteration execution has reached the maximum number of iterations, stop and output the final result.

[0032] The genetic algorithm is implemented using programming software such as MATLAB and Python. The task shelf, mobile robot, intelligent warehouse open shelf position coordinates and quantity, task shelf historical heat, task shelf historical handling times, task shelf handling times by the same picking station, workbench position coordinates and quantity, and other data are input to solve the shelf placement position allocation scheme that meets the requirements.

[0033] Step five: use the mobile robot task allocation strategy to allocate tasks to the mobile robot and identify whether there is a shelf deadlock conflict; if there is, the transfer shelf position selection strategy in the mobile robot task allocation strategy will provide a transfer position to solve the conflict.

[0034] In this step, the mobile robot task allocation strategy is designed to solve the shelf position occupation and deadlock conflict that easily occurs when the task shelf returns to the allocated position. The shelf position occupation conflict is that the original task shelf on the newly allocated position of the task shelf has not been picked up / restocked. The shelf position conflict type is shown in Figure 2 The specific steps are as follows:

[0035] Step five one: the mobile robot task allocation strategy refers to the "last in first out" data access idea in computer stack structure, and the first preferred shelf is pushed to the bottom of the stack. Then the shelves are stored in the stack according to the position occupation relationship, forming a task shelf sequence. When the mobile robot task allocation is needed, the shelf position is obtained from the top of the stack. According to this rule, a plurality of task shelf sequences containing all picking / replenishment tasks are finally formed to form a task shelf set. The task shelf sequence in reverse order can solve the common shelf position occupation conflict and form the basic shelf carrying sequence. The random first preferred shelf strategy is used to form the task shelf sequence, and the related factors such as shelf heat and shelf correlation are not considered. In order to realize the diversity of the task shelf set, the shelf high heat priority selection strategy is designed to select the first preferred shelf.

[0036] Step five two: after forming the task shelf sequence, all task shelf sequences in the task shelf set are connected and reversed in the order of forming the task shelf sequence. The first task shelf is allocated according to the principle of scheduling the nearest mobile robot, and the remaining task shelves in the task shelf set are allocated according to the principle of scheduling the idle robot. This strategy can ensure that there is no shelf position occupation conflict in the process of completing the task shelf carrying task of the mobile robot.

[0037] Step five three: the priority selection distance target shelf nearest open position strategy is used as the transfer strategy, that is, in each group of deadlock tasks, the shelf that is executing the return task is temporarily stored in the nearest open shelf position to the allocated position.

[0038] Step six: the greedy algorithm of the fusion mobile robot task allocation strategy is used to solve the optimal mobile robot carrying scheme. The specific steps are as follows:

[0039] Step six one: the mathematical expression of the objective function of the greedy algorithm and the constraint conditions are as follows:

[0040] (7)

[0041] Constraint conditions:

[0042] (8)

[0043] (9)

[0044] (10)

[0045] Among them, the average number of shelves carried by each batch of task robots:

[0046] (11)

[0047] The robot moving distance is calculated using Manhattan distance:

[0048] (12)

[0049] (13)

[0050] (14)

[0051] (15)

[0052] wherein, the first and second terms of formula (7) represent the mobile robot task completion time, and the third term represents the picking / restocking time; the constraint formula (8) constrains each rack to be carried by only one mobile robot at a time; formula (9) constrains the total number of tasks performed by each mobile robot to be no more than the task average; formula (10) constrains each set of deadlock racks to be assigned only one transit location. represents the minimum value of the time taken by the mobile robot to complete the task with the longest duration; is a set of rack transit locations, is the total number of open rack locations, represents the rack picking / restocking transit location, . is the picking / restocking time; is the average number of racks carried by a mobile robot per batch of tasks; represents the mobile robot from the initial location to the rack location; represents the rack from the workstation to the transit location ; represents the rack from the transit location to the rack assignment location ; represents the rack from the workstation to the rack assignment location ; is the coordinate of the rack picking / restocking transit location in the system. represents whether the rack is carried by a mobile robot ; represents whether the rack uses a transit location ; represents whether the rack is moved from the transit location to the rack assignment location by a mobile robot carrying; representing a shelf whether through the workbench directly moved to the shelf allocation position final position.

[0053] Step six two: set the number of task shelf set. According to the shelf position allocation result, the task shelf sequence is generated according to the preferred shelf strategy, and multiple groups of task shelf sets are formed.

[0054] Step six three: as shown in the following greedy algorithm fusion strategy is designed: Figure 4

[0055] Step six three one: search whether there is a deadlock shelf in the task shelf set, if not, skip; if yes, select a transfer position for the deadlock shelf group according to the transfer strategy and add the transfer position to the allocation position carrying task to all tasks.

[0056] Step six three two: if the robot executes the task shelf is the first task shelf in the deadlock shelf group that constitutes a deadlock, after the shelf completes the picking / replenishment, the robot returns the carrying shelf to the transfer position and does not execute the transfer position to the allocation position carrying task; if it is the last task shelf in the deadlock shelf group, the robot will directly return the carrying shelf to the allocation position.

[0057] Step six three three: after the robot completes all tasks, the system will schedule idle robots to complete the transfer position to the allocation position carrying task.

[0058] Step six three four: use the greedy algorithm to obtain the moving robot carrying scheme that minimizes the task completion time.

[0059] Compared with the prior art, the present application has the following advantages:

[0060] (1) The shelf placement position allocation model based on shelf heat and shelf correlation degree can effectively give the shelf placement position scheme that maximizes the picking efficiency under the whole task cycle through genetic algorithm solution.

[0061] (2) The shelf position occupation and deadlock conflict resolution method provided by the mobile robot carrying process task allocation strategy is effective in solving other similar problems. BRIEF DESCRIPTION OF DRAWINGS

[0062] Figure 1 The intelligent warehouse shelf return position allocation flowchart;

[0063] Figure 2 The shelf position conflict type introduction diagram;

[0064] Figure 3 ​Example of a task assignment strategy for a mobile robot handling process;

[0065] Figure 4 This is a flowchart of the greedy algorithm. Detailed Implementation

[0066] The technical solution of the present invention will be further described below with reference to the accompanying drawings, but it is not limited thereto. Any modifications or equivalent substitutions to the technical solution of the present invention that do not depart from the spirit and scope of the technical solution of the present invention should be covered within the protection scope of the present invention.

[0067] This invention provides a method for intelligent warehouse shelf return location allocation based on shelf popularity and shelf correlation, the method comprising the following steps:

[0068] Step 1: Extract historical popularity data of the task shelf based on historical order data, count the number of times the task shelf was moved during this execution, and calculate the task shelf popularity value. Similarly, the shelf correlation value for each task is calculated based on the order data. .

[0069] Step 2: Using the smart warehouse floor plan as the origin, rasterize the smart warehouse map and establish a rectangular coordinate system to determine the task shelves. and the coordinates of each task shelf Workbench Position coordinates and mobile robots starting position coordinates Determine picking / replenishment time .

[0070] Step 3: Organize the open locations and task-specific locations of shelves in the smart warehouse to form a set of allocable shelf locations. The location of the task shelves is a significant factor affecting the robot's travel distance, while the relevance of the task shelves is a crucial indicator of robot congestion. Therefore, a 1:1 weighting can be used, and the shelf placement allocation model is as follows:

[0071]

[0072] Constraints:

[0073]

[0074]

[0075] In the shelf placement allocation model, the first term represents the product of the shelf group correlation degree and the spacing. Indicates shelf Whether to store in the designated shelf location The demand for intelligent warehouse shelf placement positions is analyzed, and a shelf placement position allocation model is designed, and the weights of the first and second items are both 0.5.

[0076] Step 4: Design a genetic algorithm to solve the optimal shelf placement position allocation model scheme.

[0077] Step 4.1: The basic steps of the genetic algorithm are as follows: adopt real number coding method, the gene position on the chromosome represents the shelf number, and the length is the total number of shelves; the gene represents the shelf allocation position. The chromosome coding rule is: first, randomly pair the shelves in the front row of the heat ranking with the shelves in the front row of the distance ranking, then randomly match the remaining shelves with the remaining shelf placement positions to form a complete chromosome.

[0078] Step 4.2: Design a fitness function with a weight ratio of 1:1:

[0079]

[0080] Step 4.3: Use traditional selection method, design crossover operation rules: to avoid the same shelf placement position appearing repeatedly in the same chromosome, search for chromosome pairs before crossing to form two unique gene groups without the same genes. Randomly select one gene from the unique gene group of the chromosome pair for crossover to generate a new group of chromosomes.

[0081] Step 4.4: Design mutation operation rules: to ensure that the shelf position corresponds to the shelf one by one, generate a number of special shelf placement positions different from the chromosome genes. Randomly select a special shelf placement position and replace any gene on the chromosome to generate a new offspring. As the number of iterations increases, the differences between chromosomes will become smaller and smaller, and the number of individuals formed by crossing will also become smaller and smaller. To ensure the total number of chromosomes participating in the mutation process, high fitness value chromosomes in the population will be used for filling.

[0082] Step 4.5: Set the maximum number of iterations. Use programming software (such as MATLAB, Python) to implement the genetic algorithm as above. Input the data of task shelves, mobile robots, etc. to solve the shelf placement position allocation scheme that meets the requirements.

[0083] Step 5: Use the mobile robot task allocation strategy to allocate tasks to mobile robots and identify whether there is a shelf deadlock conflict; if there is, the transfer shelf position selection strategy in the mobile robot task allocation strategy will give the transfer position required to solve the deadlock conflict.

[0084] Step 5.1: The mobile robot task allocation strategy refers to the "last in, first out" data access idea in computer stack structure. The first preferred shelf is pushed to the bottom of the stack, and then the shelves are stored in the stack in order of position occupation, forming a task shelf sequence. When the mobile robot task allocation is needed, the shelf position is obtained from the top of the stack. According to this rule, a plurality of task shelf sequences containing all picking / restocking tasks are finally formed to form a task shelf set. The task shelf sequence in reverse order can solve the common shelf position occupation conflict and form the basic shelf moving sequence. The random first preferred shelf strategy is used to form the task shelf sequence, without considering the shelf heat, shelf correlation and other related factors. In order to realize the diversity of the task shelf set, the high heat priority selection strategy is designed to select the first preferred shelf.

[0085] Step 5.2: After forming the task shelf sequence, all task shelf sequences in the task shelf set are concatenated and reversed in the order of task shelf sequence formation. The first task shelf is allocated according to the principle of scheduling the nearest mobile robot, and the remaining task shelves in the task shelf set are allocated according to the principle of scheduling the idle robot. This strategy can ensure that there is no shelf position occupation conflict during the process of completing the restocking task of the mobile robot.

[0086] Step 5.3: The priority selection distance target shelf nearest open position strategy is used as the transfer strategy, that is, in each group of deadlock tasks, the shelf that is executing the return task is temporarily stored in the open storage location closest to its assigned location.

[0087] Step 6: The greedy algorithm of the fusion mobile robot task allocation strategy is used to solve the optimal mobile robot moving scheme.

[0088] Step 6.1: The objective function and constraint conditions of the greedy algorithm are as follows:

[0089]

[0090] Wherein the first and second terms represent the mobile robot task completion time, and the third term represents the picking / restocking time.

[0091] Constraint conditions:

[0092] (1) Constrain each shelf to be moved by only one mobile robot at a time:

[0093]

[0094] (2) Constrain the total number of tasks executed by each mobile robot to not exceed the task average:

[0095]

[0096] (3) Constrain each group of deadlock shelves to be assigned only one transfer location:

[0097]

[0098] The average number of shelves handled by each batch of robots is as follows:

[0099]

[0100] The robot's movement distance is calculated using Manhattan distance:

[0101]

[0102]

[0103]

[0104]

[0105]

[0106] in, For the set of transfer locations on the shelf, The total number of open shelf locations, Indicates shelf After picking / replenishment, transit location . For picking / replenishment time; The average number of shelves handled by the mobile robot per batch of tasks; Indicates mobile robot From initial location to shelf Distance between locations; Indicates from the shelf Location to workbench Distance between; Indicates shelf From the cargo counter To the transfer point Distance between; Indicates shelf From transit location To the shelf allocation location Distance between; Indicates shelf From the workbench To the shelf allocation location The distance; For the system of shelves Picking / Replenishment Transit Location The coordinates. Indicates shelf Is it by mobile robot? Transport; Indicates shelf Whether to use a transit location ; Indicate the rack From the transit location Move to the rack assignment location Whether by mobile robot Carry; Indicate the rack Whether via the workbench Move directly to the rack assignment location Final location.

[0107] Step 6.2: Design a greedy algorithm fusion strategy: First, set the number of task rack sets. Then, according to the rack location assignment result, generate a task rack sequence according to the preferred rack strategy to form multiple groups of task rack sets. Check if there are any deadlocked racks in the task rack set. If not, skip; if so, select a transit location for the deadlocked rack group according to the transit strategy and add the transit location to the assignment location carrying task to all tasks. If the robot executes the task rack and it is the first task rack in the deadlocked rack group that forms a deadlock, the robot will return the carrying rack to the transit location after the rack completes the picking / restocking, and will not execute the transit location to the assignment location carrying task; if it is the last task rack in the deadlocked rack group, the robot will return the carrying rack directly to the assignment location. After the robot completes all tasks, the system will schedule an idle robot to complete the transit location to the assignment location carrying task. Finally, use a greedy algorithm to obtain a mobile robot carrying scheme that minimizes task completion time.

Claims

1. A method for assigning a return position of an intelligent warehouse shelf based on shelf heat and relevance, characterized in that The method comprises the following steps: Step one: organize task shelf historical order data, calculate task shelf heat value and correlation value; Step two: grid the intelligent warehouse, extract task shelf, workbench, and mobile robot data; Step three: organize shelf dispensable position candidate set, design shelf placement position dispensing model according to demand, the specific steps are as follows: Step three one: determine open shelf positions in the smart warehouse, combine the open shelf positions with the task shelf positions to form a shelf dispensable position candidate set , is a shelf dispensable position candidate set, is a total number of shelf dispensable positions; Step three two: design the following shelf placement position dispensing model according to demand: （1） Constraint condition: （2） （3） （4） （5） Wherein, the first item of formula (1) represents the product of the shelf correlation degree and the interval, the second item represents the ratio of the shelf heat and the distance from the assigned position to the workbench, is the weight of the first item; is the interval between two shelves after the shelf position is assigned; represents the distance between the shelf position and the workbench ; is a larger value; is the coordinate of the assigned position of the shelf ; represents whether the shelf is stored in the assigned position of the shelf ; ; Step three: analyze the smart warehouse shelf placement position demand and smart warehouse equipment conditions, design weight value , form a shelf placement position allocation model; Step four: design intelligent genetic algorithm to solve optimal shelf placement position dispensing scheme; Step five: use mobile robot task dispensing strategy to dispense mobile robot tasks and identify whether there is shelf deadlock conflict; if there is, the transfer shelf position selection strategy in the mobile robot task dispensing strategy will provide a transfer position to solve the conflict, the specific steps are as follows: Step five one: the mobile robot task dispensing strategy refers to the "last in first out" data access idea in the computer stack structure, the shelf is first pressed into the bottom of the stack, and then the shelves are stored in the stack in order of position occupation, forming a task shelf sequence, when mobile robot task dispensing is needed, the shelf position is obtained from the top of the stack, according to this rule, finally a plurality of task shelf sequences containing all picking / replenishment tasks are formed to form a task shelf set; Step five two: after forming the task shelf sequence, all task shelf sequences in the task shelf set are concatenated and reversed in order according to the task shelf sequence formation order, the first task shelf is dispensed according to the principle of dispatching the nearest mobile robot, and the remaining task shelves in the task shelf set are dispensed according to the principle of dispatching idle robots; this strategy can ensure that there is no shelf position occupation conflict during the mobile robot task dispensing process; Step five three: use the strategy of preferentially selecting the nearest open position to the target shelf as the transfer strategy, that is, in each group of deadlock tasks, the shelf that is executing the return task is temporarily stored in the nearest open shelf position to its dispensing position; Step six: use the greedy algorithm that fuses the mobile robot task dispensing strategy to solve the optimal mobile robot dispensing scheme. 2.The method of claim 1, wherein The specific steps of step one are as follows: Organize task shelf historical order data, find task shelf historical heat, order data, and workbench access data, and substitute them into the following shelf heat and shelf correlation calculation formulas respectively: wherein, is the updated heat of the i-th shelf; is the heat of the i-th shelf; is the number of times the i-th shelf has been picked in history; is the time interval for the warehouse to pick / centralize replenishment; is the correlation coefficient of the i-th shelf; represents the i-th task shelf, , is the total number of task shelves; represents the i-th workstation in the system, , is the total number of picking / replenishment workstations; represents the number of times the i-th task shelf and the i-th task shelf have been served by the same picking station in history.​​​​​​ 3.The method of claim 1, wherein The specific steps of step two are as follows: The intelligent warehouse is rasterized to extract a task shelf set , a task shelf position , a workbench set , a workbench position , a mobile robot set , a mobile robot speed , a mobile robot position , is a shelf set, is a total number of task shelves, is a picking / replenishment workbench set, is a total number of picking / replenishment workbenches, is a mobile robot set, represents the th mobile robot in the system, , is a total number of mobile robots. 4.The method of claim 1, wherein The specific steps of step four are as follows: Step four one: adopt real number coding mode, the gene position on the chromosome represents the shelf number, the length is the total number of shelves, and the gene represents the shelf dispensing position; Step four two: design the fitness function as follows: （6） Step four three: design the crossover operation rule: to avoid the same shelf placement position from appearing repeatedly in the same chromosome, search for a pair of chromosomes before crossover to form two unique gene groups without the same gene, randomly select a gene from the unique gene group of a pair of chromosomes for crossover each time to generate a new group of chromosomes; Step four four: design variation operation rules: in order to ensure that the shelf position corresponds to the shelf, a number of special shelf positions different from the chromosome gene are generated; Randomly select a special shelf position and replace any gene on the chromosome to generate a new offspring; With the increase of iteration times, the difference between chromosomes will be smaller and smaller, and the number of individuals formed by crossing will also be less and less. In order to ensure the total number of chromosomes participating in the variation link, high fitness value chromosomes in the population will be used to fill in; Step four five: set the maximum iteration number, if the iteration execution has reached the maximum iteration number, stop and output the final result. 5.The method of claim 4, wherein In step four one, the chromosome coding rule is: first, randomly pair the shelves with high heat ranking and the shelves with high workbench distance ranking, then randomly match the remaining shelves with the remaining shelf positions to form a complete chromosome. 6.The method of claim 1, wherein The specific steps of step six are as follows: Step six one: the objective function of the greedy algorithm is mathematically represented and the constraint conditions are as follows: （7） Constraint conditions: （8） （9） （10） Wherein, the average number of shelves carried by the robot for each batch of tasks: （11） The robot moving distance is calculated using Manhattan distance: （12） （13） （14） （15） （16） wherein, the first and second terms of formula (7) represent the mobile robot task completion time, and the third term represents the picking / restocking time; the constraint formula (8) constrains each shelf to be carried by only one mobile robot at a time; formula (9) constrains the total number of tasks performed by each mobile robot to be no more than the task average; formula (10) constrains each set of deadlocked shelves to be assigned only one transit location; represents the minimum value of the time taken by the mobile robot to complete the task with the longest duration; is a set of shelf transit locations, is the total number of open shelf locations, represents a shelf after picking / restocking, ; is the picking / restocking time; is the average number of shelves carried by a mobile robot per batch of tasks; represents a mobile robot from the initial location to the shelf location; represents a shelf from the workstation to the transit location ; represents a shelf from the transit location to the shelf assignment location ; represents a shelf from the workstation to the shelf assignment location ; is the coordinate of the shelf picking / restocking transit location in the system; represents whether a shelf is carried by a mobile robot ; represents whether a shelf uses a transit location ; represents whether a shelf is carried by a mobile robot from the transit location to the shelf assignment location ; represents whether a shelf is moved directly from the workstation to the shelf assignment location final location; Step six two: set the number of task shelf sets, generate task shelf sequences according to the preferred shelf strategy based on the shelf position allocation result, and form multiple groups of task shelf sets; Step six three: design the fusion strategy of greedy algorithm: Step six three one: search whether there is a deadlock shelf in the task shelf set, if not, skip; If there is, select a transfer position for the deadlock shelf group according to the transfer strategy and add the transfer position to the allocation position to the transport task after the completion of all tasks; Step six three two: if the robot executes the task shelf is the first task shelf that constitutes a deadlock in the deadlock shelf group, after the completion of the picking / replenishment of the shelf, the robot will return the transported shelf to the transfer position and not execute the transport task from the transfer position to the allocation position; If it is the last task shelf in the deadlock shelf group, the robot will directly return the transported shelf to the allocation position; Step six three three: after the robot completes all tasks, the system will dispatch idle robots to complete the transport task from the transfer position to the allocation position; Step six three four: use the greedy algorithm to obtain the moving robot transport scheme that minimizes the task completion time.

Images