A multi-carriage AGV task allocation method based on a clustering algorithm

Abstract

The application provides a multi-load AGV task allocation method based on a clustering algorithm, in a multi-load AGV task allocation scene, the application can better adapt to the dynamic change of tasks in the multi-load AGV storage system and the requirement of multi-objective optimization by constructing a multi-objective function. Meanwhile, in the solving process of the multi-objective function, the application uses a non-dominated sorting genetic algorithm combined with a simulated annealing algorithm to solve the multi-objective task, can consider multiple objective functions at the same time, and can effectively search and optimize the solution space through mechanisms such as non-dominated sorting and crowded distance calculation, further improving the working efficiency and task processing capacity of the whole system.

Description

Technical Field

[0001] The present application belongs to the field of intelligent logistics technology, and particularly relates to a multi-load AGV task allocation method based on a clustering algorithm. Background Art

[0002] With the increasing automation of modern industry, automated guided vehicles (AGVs) are increasingly being used in logistics and manufacturing. AGVs can autonomously navigate complex industrial environments along predetermined paths, completing tasks such as material handling, effectively improving production efficiency and logistics automation. In actual industrial production scenarios, multi-load AGV warehousing systems are becoming increasingly common, and these systems are required to handle a large number of complex tasks.

[0003] In actual production, tasks are generated dynamically, with their type, priority, and timeframes all subject to change. Traditional AGV task allocation methods are often based on simple rules or empirical evidence. These approaches fail to fully consider the multi-position nature of AGVs and the inherent dependencies between tasks. Consequently, some existing allocation methods struggle to adjust allocation strategies effectively and promptly in the face of these dynamic changes. When urgent AGV assignments arise, it can be difficult to quickly reassign tasks to the appropriate AGV, delaying task execution. Summary of the Invention

[0004] In order to solve the above problems, one or more embodiments of this specification describe a multi-load AGV task allocation method based on a clustering algorithm.

[0005] According to a first aspect, a multi-load AGV task allocation method based on a clustering algorithm is provided, the method comprising:

[0006] Acquire target warehouse information, and construct a target warehouse grid map based on the target warehouse information;

[0007] Obtain target AGV information, and establish a first objective function based on the target warehouse grid map and the target AGV information, wherein the first objective function is used to control the total working time of the AGV;

[0008] Obtaining initial order information of target goods, and establishing a second objective function based on the initial order information of the target goods, the target warehouse grid map, and the target AGV information, wherein the second objective function is used to regulate the correlation coefficient of each goods in the task order corresponding to the initial order information of the target goods;

[0009] Establishing an objective constraint function set, the objective constraint function set including at least six constraint functions, the constraint functions being used to constrain the value range of at least one function variable, the types of the function variables including the number of first order lines in each order, the number of second order lines of the task order obtained after clustering the initial orders, the allocation quantity of the task order, the AGV execution status corresponding to each task order, the sorting time of each task order on the AGV corresponding to the task order, and the total number of AGVs in working condition, the number of the first order lines being determined based on the number of goods types in the initial order corresponding to the first order line, and the number of the second order lines being determined based on the number of goods types in the task order corresponding to the second order line;

[0010] The target warehouse information, the target AGV information, and the target cargo initial order information are calculated based on the target solving algorithm and the target constraint function set, and the optimal solutions of the first objective function and the second objective function are obtained respectively. An optimal planning scheme for the AGV allocation task is generated based on the optimal solutions, and the target solving algorithm includes a simulated annealing algorithm and a non-dominated sorting genetic algorithm.

[0011] Preferably, the acquiring target warehouse information and constructing the target warehouse grid map according to the target warehouse information includes:

[0012] Obtain target warehouse information, including warehouse structure information, location information of each shelf in the warehouse, location information of each sorting station in the warehouse, and information on the type of goods corresponding to each shelf;

[0013] Determine a minimum bounding rectangle of the warehouse according to the warehouse structure information, select a vertex of the minimum bounding rectangle of the warehouse as a coordinate origin, and establish a Cartesian coordinate system through the coordinate origin;

[0014] Plotting the positions of the shelves and the sorting stations in the Cartesian coordinate system;

[0015] According to the cargo type information corresponding to each of the shelves, two adjacent shelves with the same cargo type information are set as a grid unit to obtain a target warehouse grid map.

[0016] Preferably, the first objective function is:

[0017]

[0018] in, Represents a task order The corresponding completion time, T represents the task order The corresponding maximum completion time, W C represents the set of task orders obtained after initial order clustering, represents the mth task order in the set of task orders, F1 represents the first objective function, and C represents the order of the matrix.

[0019] Preferably, it is characterized in that the second objective function is:

[0020]

[0021] Among them, α represents the sum of the correlation coefficients of each product after the initial order clustering, Represents a task order The corresponding cargo correlation coefficient matrix, represents the mth task order in the set of task orders, C ij represents the correlation between the i-th item and the j-th item, and C represents the order of the matrix;

[0022] C ij The calculation formula is as follows:

[0023]

[0024] Among them, α represents the sum of the correlation coefficients of each product after the initial order clustering, Represents a task order The corresponding cargo correlation coefficient matrix, represents the mth task order in the set of task orders, C ij represents the correlation between the i-th item and the j-th item, and C represents the order of the matrix.

[0025] Preferably, it is characterized in that the objective constraint function set includes:

[0026]

[0027] Among them, P represents the type of goods, K represents the number of the first order line in the initial order, U represents the number of the second order line of each task order obtained after the initial order is clustered, Q represents the number of goods that the AGV corresponding to each task order can carry, and V w Indicates the total number of AGVs in working state, N indicates the total number of AGVs; y mn Indicates the sorting status of the goods corresponding to the nth second order line of the mth task order in the task order, the y mn is a 0 / 1 variable, if y mn If it is 1, it means y mn The goods corresponding to the second order line of the mth task order are executed by the corresponding AGV. If y mn If it is 0, it means y mnThe goods corresponding to the nth second order line of the corresponding mth task order have not been executed by the corresponding AGV, where M is the total number of task orders; Represents a task order The start time on the corresponding AGV, Represents a task order The end time of the nth second order line on the corresponding AGV, Represents a task order The sorting time of the nth second order line on the corresponding AGV; v represents the AGV driving speed, Represents a task order The corresponding AGV sorting distance, (x0, y0) represents the sorting station position, (x u ,y u ) indicates a task order The corresponding shelf position of the goods currently sorted by the corresponding AGV, (x u-1 ,y u-1 ) indicates a task order The shelf position corresponding to the goods sorted by the corresponding AGV; C represents the order of the matrix, and m and n represent the mth row and nth column of the calculation matrix respectively.

[0028] Preferably, the target warehouse information, the target AGV information, and the target cargo initial order information are calculated based on the target solving algorithm and the target constraint function set to obtain the optimal solutions of the first objective function and the second objective function, respectively, and the optimal planning scheme for the AGV allocation task is generated according to the optimal solutions, including:

[0029] generating a first population based on the target warehouse information for the target goods initial order information and the target AGV information, wherein the target AGV information includes the AGV driving speed, the task allocation corresponding to the AGV, the number of AGVs, the maximum load of the AGV, and the AGV position; and the target goods initial order information includes the order quantity, the type of goods for each order, and the quantity of goods of the same type for each order;

[0030] performing non-dominated sorting and crowding calculation on the first population based on a simulated annealing algorithm to determine a non-dominated rank of each individual in the first population, and dividing the individuals in the first population into N ranks from high to low according to the non-dominated rank;

[0031] Performing a genetic operation on the first population to obtain a target offspring population, clustering the target offspring population and the first population to obtain a second population, performing a non-dominated sorting process and a crowding calculation on the second population based on a simulated annealing algorithm, determining a non-dominated rank of each individual in the second population, and dividing the individuals in the second population into N ranks from high to low according to the non-dominated rank;

[0032] Randomly perturbing the second population to obtain new derived individuals, screening the new derived individuals using a simulated annealing algorithm, and replacing individuals corresponding to the new derived individuals in the second population with the screened new derived individuals to obtain a third population;

[0033] The third population is used as the new first population, and the steps of performing non-dominated sorting processing and congestion calculation on the first population are repeated, while determining the non-dominated level of each individual in the first population. After the number of repetitions reaches a maximum preset number, the optimal solutions of the first objective function and the second objective function are obtained respectively by decoding the third population, and the optimal planning scheme for the AGV allocation task is generated based on the optimal solutions.

[0034] Preferably, generating a first population based on the target warehouse information and the target goods initial order information and the target AGV information includes:

[0035] Clustering the orders according to the target goods initial order information to obtain a target task order set, wherein the target task order set includes multiple task orders;

[0036] For any type of goods in any task order, goods with a quantity greater than a first preset value are divided into at least two groups, and the quantity of goods corresponding to each group is determined based on the number of load positions of the AGV corresponding to the task order;

[0037] Based on any type of ungrouped goods in any of the task orders and the target AGV information, a first individual corresponding to the task order is generated, and a first population is generated based on the first individual.

[0038] Preferably, performing a genetic operation on the first population to obtain a target offspring population, and clustering the target offspring population and the first population to obtain a second population, comprises:

[0039] The first individual in the first population whose non-dominated level is higher than the first level is retained, and two target preset times are randomly selected to compare the crowding degree of the second individuals in the first population whose non-dominated level is not higher than the first level and have not been compared with the crowding degree, and the second individuals with higher crowding degree are retained until all second individuals have completed the crowding degree comparison, thereby obtaining a second population;

[0040] performing gene segment exchange on a second individual having a crossover point in the second population based on a multi-point crossover rule, and storing the second individual after the gene segment exchange as a new individual in the second population;

[0041] Randomly selecting a second preset number of second individuals from the second population, randomly exchanging any two gene positions on each selected second individual to obtain a new individual, and storing the new individual in the second population to obtain a target offspring population.

[0042] Preferably, generating a first individual corresponding to the task order based on any type of ungrouped goods in any task order and the target AGV information, and generating a first population based on the first individual includes:

[0043] Based on the maximum load information of each AGV in the target AGV information, determining whether the quantity of any type of ungrouped goods in any of the task orders is less than the number of loads of the AGV corresponding to the task order;

[0044] When the number of ungrouped goods of any type in the task order is less than the number of AGV carrying capacity corresponding to the task order, the ungrouped goods of any type in the task order are divided into corresponding first bodies according to their types, and the length of the first bodies is determined based on the number of ungrouped goods corresponding to the first bodies;

[0045] The first individuals are aggregated to generate a first population.

[0046] Preferably, the randomly perturbing the second population to obtain new derived individuals, screening the new derived individuals using a simulated annealing algorithm, and replacing corresponding individuals in the second population with the screened new derived individuals to obtain a third population, comprising:

[0047] Set the initial temperature, temperature drop rate and end temperature;

[0048] taking a second individual in the second population whose non-dominated ranking is higher than the second ranking as an original individual, and obtaining a new derived individual by randomly perturbing the original individual;

[0049] Calculating the energy difference of each of the newly derived individuals;

[0050] When the energy difference corresponding to the newly derived individual is not greater than 0, replacing the original individual of the second population corresponding to the newly derived individual with the new derived individual;

[0051] When the energy difference corresponding to the newly derived individual is greater than 0, calculating a target retention probability for the newly derived individual, retaining the newly derived individual according to the target retention probability, and replacing the original individual of the second population corresponding to the newly derived individual with the retained new derived individual;

[0052] The current temperature is updated according to the descent rate. After each update of the current temperature, the steps of taking the second individual in the second population whose non-dominated sorting level is higher than the second level as the original individual and obtaining a new derived individual by randomly perturbing the original individual are repeated until the current temperature is equal to the termination temperature.

[0053] Beneficial Effects: In the multi-load AGV task allocation scenario, the present invention can better adapt to the dynamic changes of tasks and the requirements of multi-objective optimization in the multi-load AGV storage system by constructing a multi-objective function. At the same time, in the process of solving the multi-objective function, the present invention uses a non-dominated sorting genetic algorithm combined with a simulated annealing algorithm to solve the multi-objective task. It can consider multiple objective functions at the same time and can effectively search and optimize the solution space through mechanisms such as non-dominated sorting and congestion distance calculation. It can then perform real-time modulation of emergency AGV allocation tasks, thereby improving the work efficiency and task processing capabilities of the entire system. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0055] Figure 1 A flowchart of a multi-load AGV task allocation method based on a clustering algorithm provided in an embodiment of the present application;

[0056] Figure 2 A schematic diagram of an execution method of a multi-load AGV task allocation method based on a clustering algorithm provided in an embodiment of the present application;

[0057] Figure 3 A schematic diagram of the layout of a target warehouse grid map for a multi-load AGV task allocation method based on a clustering algorithm provided in an embodiment of the present application. DETAILED DESCRIPTION

[0058] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application.

[0059] In the following introduction, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance. The following introduction provides multiple embodiments of the present application. Different embodiments can be replaced or combined, so the present application can also be considered to include all possible combinations of the same and / or different embodiments described. Therefore, if one embodiment includes features A, B, and C, and another embodiment includes features B and D, then the present application should also be considered to include embodiments containing one or more of all other possible combinations of A, B, C, and D, even though the embodiment may not be clearly described in the following text.

[0060] The following description provides examples and does not limit the scope, applicability, or examples set forth in the claims. Changes may be made to the function and arrangement of the elements described without departing from the scope of the present application. Various examples may appropriately omit, replace, or add various processes or components. For example, the described method may be performed in an order different from the order described, and various steps may be added, omitted, or combined. In addition, features described in some examples may be combined in other examples.

[0061] like Figure 1 As shown, Figure 1 : This is a flow chart of a multi-load AGV task allocation method based on a clustering algorithm provided in an embodiment of the present application. The method of this embodiment includes:

[0062] S1. Obtain target warehouse information, and construct a target warehouse grid map based on the target warehouse information.

[0063] The execution subject of this method is the central processing unit of the target warehouse.

[0064] In the embodiments of this specification, Figure 2 As shown in the figure, the central processor is used to obtain and distribute the information of each order, and then the intelligent sorting of the goods in the target warehouse is realized by controlling each AGV, and then manual or machine secondary sorting and packaging are carried out. Figure 3As shown, the warehouse grid map is a map that divides the warehouse environment into several grids and stores occupancy, elevation and other information in each grid. Currently, common warehouse grid maps can be divided into two-dimensional grid maps, three-dimensional grid maps and 2.5-dimensional grid maps according to the differences in the data stored in the grids. For ease of explanation, the target warehouse grid map used in the method of this embodiment is a two-dimensional grid map. The different states of the map are represented by the color of the two-dimensional grid cells. The grid cells have three states: occupied (indicating obstacle areas), idle (indicating open areas without obstacles) and unknown (indicating areas not detected by sensors). For example, in an indoor warehouse scene, the black area can represent the grid cell corresponding to the shelf, white represents the picking channel, and the light color represents the sorting station. Its advantage is that the single data structure and lower data storage requirements are particularly suitable for local trajectory planning with high calculation frequency. Robots in the warehouse (such as AGVs, AMRs, etc.) can understand environmental information based on the grid map, including the location of obstacles, traversable areas, etc., so as to plan the optimal or better path from the starting point to the end point, avoid obstacles, and achieve efficient and safe autonomous navigation and cargo transportation. This embodiment mainly allocates order tasks to multiple multi-load AGVs on the grid map of the target warehouse.

[0065] In one possible implementation, obtaining target warehouse information and constructing a target warehouse grid map based on the target warehouse information includes:

[0066] Obtain target warehouse information, including warehouse structure information, location information of each shelf in the warehouse, location information of each sorting station in the warehouse, and information on the type of goods corresponding to each shelf;

[0067] Determine a minimum bounding rectangle of the warehouse according to the warehouse structure information, select a vertex of the minimum bounding rectangle of the warehouse as a coordinate origin, and establish a Cartesian coordinate system through the coordinate origin;

[0068] Plotting the positions of the shelves and the sorting stations in the Cartesian coordinate system;

[0069] According to the cargo type information corresponding to each of the shelves, two adjacent shelves with the same cargo type information are set as a grid unit to obtain a target warehouse grid map.

[0070] In the embodiments of this specification, Figure 3As shown, establishing a warehouse grid map requires discretizing the warehouse into grids of varying sizes based on its structure. To ensure that each sorting station, shelf, and aisle in the warehouse is fully included in the warehouse grid map, it is necessary to obtain the warehouse's minimum bounding rectangle and, based on actual needs, select a vertex of the warehouse's minimum bounding rectangle as the coordinate origin. This coordinate origin is then used to initialize the Cartesian coordinate system. Furthermore, setting two adjacent shelves with the same cargo type information as a grid unit improves AGV sorting efficiency and shortens the AGV's working path. If there are no adjacent shelves with the same cargo type around the shelf, the shelf is set as an independent grid unit. The target warehouse grid map can be obtained based on the positions of each grid unit, each sorting station, and each sorting aisle.

[0071] S2. Obtain target AGV information, and establish a first objective function based on the target warehouse grid map and the target AGV information. The first objective function is used to regulate the total working time of the AGV.

[0072] In the embodiments of this specification, total AGV operating time is a key objective in the multi-load AGV task allocation process. Reducing the total AGV operating time corresponding to a collection of order tasks can reduce AGV wear and tear during task execution, while also reducing the number of idle AGVs and thereby improving AGV utilization. Therefore, regulating total AGV operating time based on a first objective function is essential. As an example, the first objective function can be a composite function composed of multiple functions, such as a mean error function and a maximum value function.

[0073] In one embodiment, the first objective function is:

[0074]

[0075] in, Represents a task order The corresponding completion time, T represents the task order The corresponding maximum completion time, W C represents the set of task orders obtained after initial order clustering, represents the mth task order in the set of task orders, F1 represents the first objective function, and C represents the order of the matrix.

[0076] In the embodiments of this specification, Figure 2 As shown in Figure 1, the initial orders arrive in waves. By processing the initial orders of each wave, the dynamic allocation of multi-position AGV tasks can be achieved. R ), R represents the number of orders in the same wave, and in each order W rThere are multiple first order lines L rk , each first order line L rk Order a product, then W r =(L r1 ,…,L rK ); After obtaining the orders, all orders W will be clustered, and then the initial order W after clustering will be split into multiple task orders W C , M represents the number of task orders after splitting. It also contains multiple second order lines, each second order line Order a product, then The order will then be picked Assigned to N AGVs for handling, because each There are multiple picking requirements, and the AGV will pass through multiple shelves and then arrive at the sorting station. Finally, the picker will re-sort and pack the goods according to the initial order requirements and place them in the outbound area to wait for delivery. The corresponding maximum completion time T is the smallest, then each task order needs to be The corresponding maximum completion time is reduced.

[0077] S3. Obtain the initial order information of the target goods, and establish a second objective function based on the initial order information of the target goods, the target warehouse grid map, and the target AGV information. The second objective function is used to regulate the correlation coefficient of each goods in the task order corresponding to the initial order information of the target goods.

[0078] In the embodiments of this specification, in the actual scenario of warehousing and logistics, the degree of completeness of the initial order goods directly affects the sorting efficiency of the AGV. When the degree of completeness of the initial order goods is low, a large number of goods will be missing or the arrival of the goods will be inconsistent. This situation will cause a large number of AGVs to be in a stagnant state during sorting, rather than bypassing the sorting of the goods corresponding to the second order line of the current task order, thereby reducing the overall operational fluency of the sorting system. In order to improve the sorting efficiency, it is necessary to cluster the initial orders corresponding to the initial order information of the target goods, and at the same time split the clustered initial orders W into multiple task orders. These split orders are the task orders corresponding to the initial order information of the target goods. By taking the sum of the correlation coefficients of each task order as the second target and then adjusting the allocation method of splitting the clustered initial orders into multiple tasks, the time loss caused by the incomplete goods in the AGV sorting process can be reduced. For example, when allocating tasks to multiple AGVs, when the AGV encounters the situation of missing goods or the arrival of the goods, the stagnant AGV can be controlled to perform the sorting work of other categories of goods. As an example, the first objective function may be a composite function composed of multiple functions such as an average error function and a minimum value function.

[0079] In one embodiment, the second objective function is:

[0080]

[0081] Among them, α represents the sum of the correlation coefficients of each product after the initial order clustering, Represents a task order The corresponding cargo correlation coefficient matrix, represents the mth task order in the set of task orders, C ij represents the correlation between the i-th item and the j-th item, and C represents the order of the matrix;

[0082] C ij The calculation formula is as follows:

[0083]

[0084] Where R represents the total number of task orders, a represents the number of orders for goods No. i and No. j ordered at the same time, and K is the number of the first order lines of the initial order before clustering.

[0085] In the embodiment of this specification, when calculating the sum of the correlation coefficients of the goods after the initial orders in the same wave are clustered, the types of the goods after the initial orders are clustered are the same as the types of the goods in the orders before clustering. Therefore, C ijWhen it represents the correlation between the goods No. i and the goods No. j, the corresponding first order line number does not change. The embodiment method of this specification uses the SKU number to control the AGV to identify the corresponding goods. SKU refers to the smallest inventory unit in inventory management, which is used to distinguish the attribute combination of different commodities. Each commodity has a unique SKU number corresponding to its brand, model, configuration, color, size and other attributes. By adjusting the correlation coefficient of each commodity after the initial order clustering in the corresponding task order to the maximum through the second objective function, the correlation of each commodity of each assigned task order can be maximized, thereby optimizing the AGV task allocation effect, thereby reducing the time loss caused by incomplete goods in the AGV sorting process.

[0086] S4. Establish a target constraint function set, which includes at least six constraint functions, and the constraint function is used to constrain the value range of at least one function variable. The types of function variables include the number of first order lines in each initial order, the number of second order lines of task orders obtained after initial order clustering, the allocation quantity of the task orders, the AGV execution status corresponding to each task order, the sorting time of each task order on the AGV corresponding to the task order, and the total number of AGVs in working state. The number of the first order lines is determined based on the number of goods types of the initial order corresponding to the first order line, and the number of the second order lines is determined based on the number of goods types of the task order corresponding to the second order line.

[0087] In the embodiments of this specification, the number of AGVs that can be stored in the target warehouse is fixed, and it is necessary to limit the sorting time of each task order on the corresponding AGV and the total number of AGVs in operation. The types of goods after the initial orders in the same wave are clustered are fixed, and the AGV carrying capacity in the target warehouse is also fixed. The number of second order lines in the clustered task orders will not exceed the number of goods that the multi-carrying AGV can carry. At the same time, in order to improve the utilization rate of AGVs, it is necessary to ensure that AGVs are not idle as much as possible, and all AGVs must be in the execution state as much as possible. In addition, when the initial order after clustering is divided into multiple task orders, it is necessary to limit the execution of each task order by a single AGV to improve the AGV execution efficiency. Therefore, it is necessary to establish a set of objective constraint functions to constrain the number of first order lines in these initial orders, the number of second order lines of the task orders obtained after the initial orders are clustered, the distribution of task orders, the execution status of the AGVs corresponding to each task order, and the sorting time of each task order on the AGV corresponding to the task order and the total number of AGVs in operation. Of course, the function variables that need to be constrained by the constraint function are not limited to the above six factors, such as the aisle temperature of the warehouse, the loss rate of AGV and the power supply of the warehouse. They can also be constrained by expanding new constraint functions in the target constraint function set.

[0088] In one embodiment, the objective constraint function set includes:

[0089]

[0090] Among them, P represents the type of goods, K represents the number of the first order line in the initial order, U represents the number of the second order line of each task order obtained after the initial order is clustered, Q represents the number of goods that the AGV corresponding to each task order can carry, and V w Indicates the total number of AGVs in working state, N indicates the total number of AGVs; y mn Indicates the sorting status of the goods corresponding to the nth second order line of the mth task order in the task order, the y mn is a 0 / 1 variable, if y mn If it is 1, it means y mn The goods corresponding to the second order line of the mth task order are executed by the corresponding AGV. If y mn If it is 0, it means y mn The goods corresponding to the nth second order line of the corresponding mth task order have not been executed by the corresponding AGV, where M is the total number of task orders; Represents a task order The start time on the corresponding AGV, Represents a task order The end time of the nth second order line on the corresponding AGV, Represents a task order The sorting time of the nth second order line on the corresponding AGV; v represents the AGV driving speed, Represents a task order The corresponding AGV sorting distance, (x0, y0) represents the sorting station position, (x u ,y u ) indicates a task order The corresponding shelf position of the goods currently sorted by the corresponding AGV, (x u-1 ,y u-1 ) indicates a task order The shelf position corresponding to the goods sorted by the corresponding AGV; C represents the order of the matrix, and m and n represent the mth row and nth column of the calculation matrix respectively.

[0091] In this embodiment, K≤P ensures that the number of first order lines in each order does not exceed the number of goods. Both the initial order and the task order meet this constraint. U≤Q ensures that the number of second order lines in each task order, obtained after clustering the initial orders, does not exceed the number of goods that the AGV corresponding to each task order can carry. This ensures that the goods for each task order can be sorted in a single task, rather than having to complete the task by creating additional task orders.

[0092] Since each picking order will be assigned to a multi-load AGV to complete the picking task, the task allocation result A can be expressed in the form of a two-dimensional matrix, as shown in the following formula:

[0093]

[0094] Among them, y mn is a 0 / 1 variable, y mn Where m and n represent the nth second order line of the mth task order. When the goods corresponding to the nth second order line of the mth task order in the task order are sorted by the corresponding AGV, y mn is 1. To ensure that all task orders can be assigned to each AGV. To ensure that each task order can only be executed by one AGV.

[0095]

[0096] in, Represents the sorting distance of AGV. During the sorting process of AGV, the goods need to be sorted out from the shelves corresponding to the goods, and then transported to the sorting platform for secondary sorting. Because the path conflicts between AGVs are not considered, the Manhattan distance is used as the standard for measuring the distance between two locations in the warehouse, that is, the sum of the two distances of the starting point and the end point in the north-south and east-west directions of the grid map is calculated. Because there are multiple order lines in a sorting order, that is, there are multiple target points, it is necessary to calculate the distance between each target point and the distance from each target point to the sorting platform. Therefore, the calculation process of the sorting distance of AGV needs to consider the position of each sorting platform in the target warehouse grid map and the shelf position corresponding to each goods in each task order. In addition, through Calculating task orders The sorting time on the corresponding AGV is used to coordinate with the first objective function to control the total working time of the AGV, and then to control the allocation of each task order to obtain a better planning solution.

[0097] S5. Calculate the target warehouse information, the target AGV information, and the target cargo initial order information based on the target solving algorithm and the target constraint function set, obtain the optimal solutions of the first objective function and the second objective function respectively, and generate the optimal planning scheme for the AGV allocation task based on the optimal solutions. The target solving algorithm includes a simulated annealing algorithm and a non-dominated sorting genetic algorithm.

[0098] In the embodiments of this specification, the non-dominated sorting genetic algorithm (NSGA) is a multi-objective optimization algorithm that primarily solves optimization problems with multiple conflicting objectives. Its core features include fast non-dominated sorting, crowding distance calculation, and an elitist strategy. Currently, there are two improved versions of the non-dominated sorting genetic algorithm: the non-dominated sorting genetic algorithm with an elitist strategy (NSGA-II) and the non-dominated sorting genetic algorithm III (NSGA-III). In this embodiment, the non-dominated sorting genetic algorithm with an elitist strategy (NSGA-II) is used. For ease of subsequent description, the present invention will be explained using NSGA-II. NSGA-II is a genetic algorithm for multi-objective optimization problems. This algorithm uses fast non-dominated sorting to divide the population into different levels and uses crowding distance to maintain solution diversity. This algorithm introduces an elitist strategy, combining the parent population with its resulting offspring population, which competes to produce the next generation population, avoiding the loss of the best individuals and improving the population level. Finally, the optimal solution set for multiple objectives is found through the concepts of non-dominated sorting and crowding distance. The simulated annealing algorithm is a probability-based optimization algorithm inspired by the principles of solid annealing, which searches for the optimal solution to a problem by simulating the metal annealing process. This method uses a control parameter to simulate the annealing temperature, allowing the algorithm to accept solutions worse than the current solution with a certain probability, helping to escape the local optimal solution and increase the chance of finding the global optimal solution. By combining the simulated annealing algorithm with the non-dominated sorting genetic algorithm, this implementation method can improve the efficiency and accuracy of solving the first and second objective functions, better adapting to the dynamic changes in tasks and the requirements of multi-objective optimization in multi-load AGV warehousing systems.

[0099] In one embodiment, the target warehouse information, the target AGV information, and the target cargo initial order information are calculated based on the target solving algorithm and the target constraint function set to obtain the optimal solutions of the first objective function and the second objective function, respectively, and an optimal planning scheme for the AGV assignment task is generated based on the optimal solutions, including:

[0100] generating a first population based on the target warehouse information for the target goods initial order information and the target AGV information, wherein the target AGV information includes the AGV driving speed, the task allocation corresponding to the AGV, the number of AGVs, the maximum load of the AGV, and the AGV position; and the target goods initial order information includes the order quantity, the type of goods for each order, and the quantity of goods of the same type for each order;

[0101] performing non-dominated sorting and crowding calculation on the first population based on a simulated annealing algorithm to determine a non-dominated rank of each individual in the first population, and dividing the individuals in the first population into N ranks from high to low according to the non-dominated rank;

[0102] Performing a genetic operation on the first population to obtain a target offspring population, clustering the target offspring population and the first population to obtain a second population, performing a non-dominated sorting process and a crowding calculation on the second population based on a simulated annealing algorithm, determining a non-dominated rank of each individual in the second population, and dividing the individuals in the second population into N ranks from high to low according to the non-dominated rank;

[0103] Randomly perturbing the second population to obtain new derived individuals, screening the new derived individuals using a simulated annealing algorithm, and replacing individuals corresponding to the new derived individuals in the second population with the screened new derived individuals to obtain a third population;

[0104] The third population is used as the new first population, and the steps of performing non-dominated sorting processing and congestion calculation on the first population are repeated, while determining the non-dominated level of each individual in the first population. After the number of repetitions reaches a maximum preset number, the optimal solutions of the first objective function and the second objective function are obtained respectively by decoding the third population, and the optimal planning scheme for the AGV allocation task is generated based on the optimal solutions.

[0105] In the embodiments of this specification, the optimal solutions for the first and second objective functions are calculated by combining the simulated annealing algorithm and the non-dominated sorting genetic algorithm (NSGA-II), rather than integrating the two algorithms separately. Specifically, the simulated annealing algorithm is used to optimize the non-dominated sorting and congestion calculation processes for the first and second populations using the NSGA-II algorithm. Furthermore, during the random perturbation of the second population to obtain newly derived individuals, the simulated annealing algorithm is used to screen these newly derived individuals. This combination of simulated annealing improves the efficiency and accuracy of the NSGA-II algorithm in solving the first and second objective functions, resulting in a more optimal solution. Furthermore, since the third population is a collection of chromosomes generated using the NSGA-II algorithm, decoding the third population is required to obtain the grouping results for the remaining goods and thus obtain the optimal solutions for the first and second objective functions.

[0106] In one embodiment, generating a first population based on the target warehouse information and the target goods initial order information and the target AGV information includes:

[0107] Clustering the orders according to the target goods initial order information to obtain a target task order set, wherein the target task order set includes multiple task orders;

[0108] For any type of goods in any task order, goods with a quantity greater than a first preset value are divided into at least two groups, and the quantity of goods corresponding to each group is determined based on the number of load positions of the AGV corresponding to the task order;

[0109] Based on any type of ungrouped goods in any of the task orders and the target AGV information, a first individual corresponding to the task order is generated, and a first population is generated based on the first individual.

[0110] In the embodiment of this specification, the setting of the first preset value is determined based on the number of AGV loads corresponding to the current task order. In order to reduce the empty rate of the AGV, the first preset value in this embodiment is equal to the number of AGV loads corresponding to the current task order.

[0111] In one embodiment, generating a first individual corresponding to the task order based on any type of ungrouped goods in any task order and the target AGV information, and generating a first population based on the first individual includes:

[0112] Based on the maximum load information of each AGV in the target AGV information, determining whether the quantity of any type of ungrouped goods in any of the task orders is less than the number of loads of the AGV corresponding to the task order;

[0113] When the number of ungrouped goods of any type in the task order is less than the number of AGV carrying capacity corresponding to the task order, the ungrouped goods of any type in the task order are divided into corresponding first bodies according to their types, and the length of the first bodies is determined based on the number of ungrouped goods corresponding to the first bodies;

[0114] The first individuals are aggregated to generate a first population.

[0115] In the embodiments of this specification, the first and second individuals are chromosomes generated based on the corresponding goods of the first and second individuals. The length of each chromosome is determined by the number of ungrouped goods of any type in any of the task orders. For example, if a task order contains three ungrouped goods, a first individual of length 3 is generated corresponding to that task order, and each gene segment in this first individual corresponds to an ungrouped good. In this way, the goods that cannot be fully transported by a multi-load AGV in each task order are recorded and corresponding first individuals are generated. Based on these first individuals, a first population can be generated, facilitating task allocation to these first individuals using a non-dominated sorting genetic algorithm.

[0116] In one embodiment, performing a genetic operation on the first population to obtain a target offspring population, and clustering the target offspring population and the first population to obtain a second population, comprises:

[0117] The first individual in the first population whose non-dominated level is higher than the first level is retained, and two target preset times are randomly selected to compare the crowding degree of the second individuals in the first population whose non-dominated level is not higher than the first level and have not been compared with the crowding degree, and the second individuals with higher crowding degree are retained until all second individuals have completed the crowding degree comparison, thereby obtaining a second population;

[0118] performing gene segment exchange on a second individual having a crossover point in the second population based on a multi-point crossover rule, and storing the second individual after the gene segment exchange as a new individual in the second population;

[0119] Randomly selecting a second preset number of second individuals from the second population, randomly exchanging any two gene positions on each selected second individual to obtain a new individual, and storing the new individual in the second population to obtain a target offspring population.

[0120] In the examples of this specification, the process of genetically manipulating the first population to generate the second population employs, in sequence, an elite selection strategy, a crossover strategy, and a mutation strategy. The elite selection strategy involves retaining the first individual in the first population whose non-dominated rank is higher than the first rank, and then retaining individuals whose non-dominated rank is not higher than the first rank according to a selection method. This selection method can be performed by sorting the individuals by non-dominated rank and performing the aforementioned operation on the first individual whose non-dominated rank is not higher than the first rank, or by directly randomly retaining the first individual in the next rank after the first rank, but is not limited to these two methods. This elite selection strategy preserves a large number of individuals with high non-dominated rank in the first population, thereby reducing the risk of eliminating high-performing individuals that might otherwise be eliminated through random selection.

[0121] The crossover strategy involves exchanging gene segments with second individuals in the second population that have a crossover point. The crossover point in the second population is the position on the chromosome selected to split the chromosome during the generation of the second population from the first. The gene segments are exchanged with second individuals in the second population that have a crossover point. The mutation strategy randomly selects a second preset number of second individuals from the second population and randomly exchanges any two gene positions in each selected second individual to generate new individuals. Using crossover and mutation strategies can increase the diversity of the first population and broaden the scope of exploration for optimal solutions.

[0122] In one embodiment, the randomly perturbing the second population to obtain new derived individuals, screening the new derived individuals using a simulated annealing algorithm, and replacing corresponding individuals in the second population with the screened new derived individuals to obtain a third population comprises:

[0123] Set the initial temperature, temperature drop rate and end temperature;

[0124] taking a second individual in the second population whose non-dominated ranking is higher than the second ranking as an original individual, and obtaining a new derived individual by randomly perturbing the original individual;

[0125] Calculating the energy difference of each of the newly derived individuals;

[0126] When the energy difference corresponding to the newly derived individual is not greater than 0, replacing the original individual of the second population corresponding to the newly derived individual with the new derived individual;

[0127] When the energy difference corresponding to the newly derived individual is greater than 0, calculating a target retention probability for the newly derived individual, retaining the newly derived individual according to the target retention probability, and replacing the original individual of the second population corresponding to the newly derived individual with the retained new derived individual;

[0128] The current temperature is updated according to the descent rate. After each update of the current temperature, the steps of taking the second individual in the second population whose non-dominated sorting level is higher than the second level as the original individual and obtaining a new derived individual by randomly perturbing the original individual are repeated until the current temperature is equal to the termination temperature.

[0129] In the embodiment of this specification, the second level is set according to actual needs, and the minimum level of the second level is not less than N, where N is the lowest level of individuals in the second population divided from high to low according to the non-dominant level.

[0130] For the original individuals above the second level, their objective function vector is f i =(f i1 , f i2 ,…,f im ), where im represents the objective function value corresponding to the mth gene segment of the original individual i. The objective function vector is composed of the real-time calculation results of the first objective function F1 and the second objective function F2. The real-time calculation results are determined by inputting the first objective function F1 and the second objective function F2 respectively based on the decoding results of the third population. For the new derived individual at the current temperature T, its objective function vector is f' i =(f' i1 , f' i2 ,…,f' im ); through the energy difference formula The energy difference ΔE between the objective function vector of the original individual and the newly derived individual can be calculated.

[0131] If ΔE≤0, that is, the new derived individual is not inferior to the original individual in all objective functions, then the new derived individual is accepted to replace the original individual; if ΔE>0, then the target retention probability is Accept the new derived individuals and replace the original individuals corresponding to the new derived individuals based on the retained new derived individuals. Then update the temperature T = αT according to the temperature drop rate α, and repeat the steps of taking the second individual with a non-dominated ranking higher than the second ranking in the second population as the original individual and obtaining the new derived individuals by randomly perturbing the original individual until the temperature T reaches the termination temperature T min .

[0132] Those skilled in the art will clearly understand that the technical solutions of the embodiments of the present application can be implemented with the help of software and / or hardware. "Controller" and "memory" in this specification refer to software and / or hardware that can independently perform or cooperate with other components to perform specific functions, where the hardware can be, for example, a Field-Programmable Gate Array (FPGA), an integrated circuit (IC), etc.

[0133] The present application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method provided in this embodiment. The computer-readable storage medium may include, but is not limited to, any type of disk, including a floppy disk, an optical disk, a DVD, a CD-ROM, a microdrive, a magneto-optical disk, a ROM, a RAM, an EPROM, an EEPROM, a DRAM, a VRAM, a flash memory device, a magnetic card or an optical card, a nanosystem (including a molecular memory IC), or any type of medium or device suitable for storing instructions and / or data.

[0134] It should be noted that for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that this application is not limited by the order of the actions described, because according to this application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by this application.

[0135] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0136] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a memory and includes several instructions for enabling a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned memory includes: various media that can store program codes, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk.

[0137] Those skilled in the art will appreciate that all or part of the steps in the various methods of the above embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable memory, which may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0138] The above is only an exemplary embodiment of the present disclosure and cannot be used to limit the scope of the present disclosure. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the specification and practicing the disclosure herein, those skilled in the art will easily think of the implementation scheme of the present disclosure. This application is intended to cover any variation, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the art that are not recorded in the present disclosure. The description and examples are to be regarded as exemplary only, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A multi-load AGV task allocation method based on clustering algorithm, characterized in that: The method comprises: Acquire target warehouse information, and construct a target warehouse grid map based on the target warehouse information; Obtain target AGV information, and establish a first objective function based on the target warehouse grid map and the target AGV information, wherein the first objective function is used to control the total working time of the AGV; Obtaining initial order information of target goods, and establishing a second objective function based on the initial order information of the target goods, the target warehouse grid map, and the target AGV information, wherein the second objective function is used to regulate the correlation coefficient of each goods in the task order corresponding to the initial order information of the target goods; Establishing an objective constraint function set, the objective constraint function set including at least six constraint functions, the constraint functions being used to constrain the value range of at least one function variable, the types of the function variables including the number of first order lines in each initial order, the number of second order lines of the task order obtained after clustering the initial orders, the allocation quantity of the task order, the AGV execution status corresponding to each task order, the sorting time of each task order on the AGV corresponding to the task order, and the total number of AGVs in working condition, the number of the first order lines being determined based on the number of goods types in the initial order corresponding to the first order line, and the number of the second order lines being determined based on the number of goods types in the task order corresponding to the second order line; The target warehouse information, the target AGV information, and the target cargo initial order information are calculated based on the target solving algorithm and the target constraint function set, and the optimal solutions of the first objective function and the second objective function are obtained respectively. An optimal planning scheme for the AGV allocation task is generated based on the optimal solutions, and the target solving algorithm includes a simulated annealing algorithm and a non-dominated sorting genetic algorithm.

2. The method according to claim 1, characterized in that The acquiring target warehouse information and constructing a target warehouse grid map according to the target warehouse information includes: Obtain target warehouse information, including warehouse structure information, location information of each shelf in the warehouse, location information of each sorting station in the warehouse, and information on the type of goods corresponding to each shelf; Determine a minimum bounding rectangle of the warehouse according to the warehouse structure information, select a vertex of the minimum bounding rectangle of the warehouse as a coordinate origin, and establish a Cartesian coordinate system through the coordinate origin; Plotting the positions of the shelves and the sorting stations in the Cartesian coordinate system; According to the cargo type information corresponding to each of the shelves, two adjacent shelves with the same cargo type information are set as a grid unit to obtain a target warehouse grid map.

3. The method according to claim 1, characterized in that The first objective function is: in, Represents a task order The corresponding completion time, T represents the task order The corresponding maximum completion time, W C represents the set of task orders obtained after initial order clustering, represents the mth task order in the set of task orders, F1 represents the first objective function, and C represents the order of the matrix.

4. The method according to claim 1, wherein The second objective function is: Among them, α represents the sum of the correlation coefficients of each product after the initial order clustering, Represents a task order The corresponding cargo correlation coefficient matrix, represents the mth task order in the set of task orders, C ij represents the correlation between the i-th item and the j-th item, and C represents the order of the matrix; C ij The calculation formula is as follows: Where R represents the total number of task orders, a represents the number of orders for goods No. i and No. j ordered at the same time, and K is the number of the first order lines of the initial order before clustering.

5. The method according to claim 1, wherein The constraint functions of the objective constraint function set include: Among them, P represents the type of goods, K represents the number of the first order line in the initial order, U represents the number of the second order line of each task order obtained after the initial order is clustered, Q represents the number of goods that the AGV corresponding to each task order can carry, and V w Indicates the total number of AGVs in working state, N indicates the total number of AGVs; y mn Indicates the sorting status of the goods corresponding to the nth second order line of the mth task order in the task order, the y mn is a 0 / 1 variable, if y mn If it is 1, it means y mn The goods corresponding to the second order line of the mth task order are executed by the corresponding AGV. If y mn If it is 0, it means y mn The goods corresponding to the nth second order line of the corresponding mth task order have not been executed by the corresponding AGV, where M is the total number of task orders; Represents a task order The start time on the corresponding AGV, Represents a task order The end time of the nth second order line on the corresponding AGV, Represents a task order The sorting time of the nth second order line on the corresponding AGV; v represents the AGV driving speed, Represents a task order The corresponding AGV sorting distance, (x0, y0) represents the sorting station position, (x u ,y u ) indicates a task order The corresponding shelf position of the goods currently sorted by the corresponding AGV, (x u-1 ,y u-1 ) indicates a task order The shelf position corresponding to the goods sorted by the corresponding AGV; C represents the order of the matrix, and m and n represent the mth row and nth column of the calculation matrix respectively.

6. The method according to claim 1, characterized in that The target warehouse information, the target AGV information, and the target cargo initial order information are calculated based on the target solving algorithm and the target constraint function set to obtain the optimal solutions of the first objective function and the second objective function, respectively, and an optimal planning scheme for the AGV allocation task is generated according to the optimal solutions, including: generating a first population based on the target warehouse information for the target goods initial order information and the target AGV information, wherein the target AGV information includes the AGV driving speed, the task allocation corresponding to the AGV, the number of AGVs, the maximum load of the AGV, and the AGV position; and the target goods initial order information includes the order quantity, the type of goods for each order, and the quantity of goods of the same type for each order; performing non-dominated sorting and crowding calculation on the first population based on a simulated annealing algorithm to determine a non-dominated rank of each individual in the first population, and dividing the individuals in the first population into N ranks from high to low according to the non-dominated rank; Performing a genetic operation on the first population to obtain a target offspring population, clustering the target offspring population and the first population to obtain a second population, performing a non-dominated sorting process and a crowding calculation on the second population based on a simulated annealing algorithm, determining a non-dominated rank of each individual in the second population, and dividing the individuals in the second population into N ranks from high to low according to the non-dominated rank; Randomly perturbing the second population to obtain new derived individuals, screening the new derived individuals using a simulated annealing algorithm, and replacing individuals corresponding to the new derived individuals in the second population with the screened new derived individuals to obtain a third population; The third population is used as the new first population, and the steps of performing non-dominated sorting processing and congestion calculation on the first population are repeated, while determining the non-dominated level of each individual in the first population. After the number of repetitions reaches a maximum preset number, the optimal solutions of the first objective function and the second objective function are obtained respectively by decoding the third population, and the optimal planning scheme for the AGV allocation task is generated based on the optimal solutions.

7. The method according to claim 6, characterized in that The generating a first population based on the target warehouse information and the target goods initial order information and the target AGV information includes: Clustering the orders according to the target goods initial order information to obtain a target task order set, wherein the target task order set includes multiple task orders; For any type of goods in any task order, goods with a quantity greater than a first preset value are divided into at least two groups, and the quantity of goods corresponding to each group is determined based on the number of load positions of the AGV corresponding to the task order; Based on any type of ungrouped goods in any of the task orders and the target AGV information, a first individual corresponding to the task order is generated, and a first population is generated based on the first individual.

8. The method according to claim 6, characterized in that The performing of a genetic operation on the first population to obtain a target offspring population, and clustering the target offspring population and the first population to obtain a second population, comprises: The first individual in the first population whose non-dominated level is higher than the first level is retained, and two target preset times are randomly selected to compare the crowding degree of the second individuals in the first population whose non-dominated level is not higher than the first level and have not been compared with the crowding degree, and the second individuals with higher crowding degree are retained until all second individuals have completed the crowding degree comparison, thereby obtaining a second population; performing gene segment exchange on a second individual having a crossover point in the second population based on a multi-point crossover rule, and storing the second individual after the gene segment exchange as a new individual in the second population; Randomly selecting a second preset number of second individuals from the second population, randomly exchanging any two gene positions on each selected second individual to obtain a new individual, and storing the new individual in the second population to obtain a target offspring population.

9. The method according to claim 7, characterized in that The step of generating a first individual corresponding to the task order based on any type of ungrouped goods in any task order and the target AGV information, and generating a first population based on the first individual, includes: Based on the maximum load information of each AGV in the target AGV information, determining whether the quantity of any type of ungrouped goods in any of the task orders is less than the number of loads of the AGV corresponding to the task order; When the number of ungrouped goods of any type in the task order is less than the number of AGV carrying capacity corresponding to the task order, the ungrouped goods of any type in the task order are divided into corresponding first bodies according to their types, and the length of the first bodies is determined based on the number of ungrouped goods corresponding to the first bodies; The first individuals are aggregated to generate a first population.

10. The method according to claim 6, characterized in that The method of randomly perturbing the second population to obtain new derived individuals, screening the new derived individuals using a simulated annealing algorithm, and replacing corresponding individuals in the second population with the screened new derived individuals to obtain a third population comprises: Set the initial temperature, temperature drop rate and end temperature; taking a second individual in the second population whose non-dominated ranking is higher than the second ranking as an original individual, and obtaining a new derived individual by randomly perturbing the original individual; Calculating the energy difference of each of the newly derived individuals; When the energy difference corresponding to the newly derived individual is not greater than 0, replacing the original individual of the second population corresponding to the newly derived individual with the new derived individual; When the energy difference corresponding to the newly derived individual is greater than 0, calculating a target retention probability for the newly derived individual, retaining the newly derived individual according to the target retention probability, and replacing the original individual of the second population corresponding to the newly derived individual with the retained new derived individual; The current temperature is updated according to the descent rate. After each update of the current temperature, the steps of taking the second individual in the second population whose non-dominated sorting level is higher than the second level as the original individual and obtaining a new derived individual by randomly perturbing the original individual are repeated until the current temperature is equal to the termination temperature.

Images