Task allocation method for multiple AGVs of automatic container terminal

Abstract

The invention relates to the technical field of automatic guided vehicle scheduling, and particularly discloses an automatic container terminal multi-AGV task allocation method, which comprises the steps of constructing a multi-target optimization model according to the total number of tasks of a current container and the total number of AGVs; the multi-objective optimization model is solved, a plurality of ordered task subsets of the container are obtained, and the number of the ordered task subsets is the same as the number of the AGVs; constructing a task allocation overhead matrix according to the bidding value of each AGV for the ordered task subset, and determining a matching relationship between the ordered task subset and the AGV according to a solving result of the task allocation overhead matrix; and determining a container task to be executed by each AGV according to a matching relationship between the ordered task subset and the AGV, and sending the corresponding container task to the corresponding AGV. According to the multi-AGV task allocation method for the automatic container terminal, the calculation efficiency and the allocation scheme quality can be balanced.

Description

Technical Field

[0001] This invention relates to the field of automated guided vehicle (AGV) scheduling technology, and in particular to a task allocation method for multiple AGVs in an automated container terminal. Background Technology

[0002] With the development of global economic integration and the continuous growth of international trade, maritime transport plays a vital role. As the core node of maritime transport, the operational efficiency of ports directly impacts the stability and smooth flow of the global supply chain. The operational efficiency of automated container terminals has become one of the key indicators for measuring a country's competitiveness in the logistics sector. Against this backdrop, Automated Guided Vehicles (AGVs), as key unmanned transportation equipment within automated container terminals, play an increasingly important role in improving the automation level and operational efficiency of terminals because they can autonomously complete container loading, unloading, and transfer, achieve precise positioning and navigation, and are easily scheduled through a unified system.

[0003] However, with the increasing size of container ships and the continued surge in maritime transport demand, automated container terminals face increasingly severe logistical processing pressures. These pressures manifest themselves in several ways: a massive and dynamically changing number of tasks, a complex internal road network, and the need for large-scale AGV fleets for efficient collaborative operations. Traditional manual experience-based scheduling and driving methods are ill-suited to handle large-scale, high-intensity operations, and are prone to safety hazards due to factors such as driver fatigue. Furthermore, they have inherent limitations in overall operational optimization and refined management. Although the national government is vigorously promoting the construction of smart ports and applying modern information technology to port operations, the optimization of scheduling in core port operations (especially AGV fleets) remains a key bottleneck restricting overall efficiency improvement.

[0004] In AGV scheduling systems at automated container terminals, task allocation strategies are crucial in determining the overall operational efficiency of the AGV fleet. Currently, existing AGV task allocation methods mainly fall into two categories: centralized allocation and distributed allocation. However, these two methods struggle to achieve an ideal balance between solution quality and computational efficiency, exhibiting the following technical problems: While centralized allocation methods can yield high-quality task allocation schemes, their high computational complexity and excessively long solution times for large-scale problems make it difficult to meet the real-time and robust requirements of terminal fleet scheduling schemes. Conversely, while distributed methods have lower computational overhead and can quickly respond to dynamic changes, they lack a global perspective and unified coordination, easily leading to local optima, uneven load distribution among AGVs, resource waste, and low system operating efficiency.

[0005] Therefore, how to provide a multi-AGV task allocation method that can balance computational efficiency and allocation scheme quality has become a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] This invention provides a task allocation method for multiple AGVs in an automated container terminal, which solves the problem in related technologies that cannot balance computational efficiency and allocation scheme quality when allocating AGV tasks.

[0007] As one aspect of the present invention, a method for task allocation of multiple AGVs in an automated container terminal is provided, comprising:

[0008] A multi-objective optimization model is constructed based on the total number of current container tasks and the total number of AGVs. The optimization objectives of the multi-objective optimization model include at least minimizing the completion time of all tasks, the total travel distance of AGVs, and the task-weighted delay time.

[0009] Solving the multi-objective optimization model yields multiple ordered task subsets for the container, wherein the number of ordered task subsets is the same as the number of AGVs.

[0010] A task allocation cost matrix is ​​constructed based on the bidding value of each AGV for an ordered subset of tasks, and the matching relationship between the ordered subset of tasks and the AGV is determined based on the solution of the task allocation cost matrix.

[0011] The container task to be performed by each AGV is determined based on the matching relationship between the ordered task subset and the AGV, and the corresponding container task is sent to the corresponding AGV.

[0012] The multi-objective optimization model is constructed based on the total number of containers and the total number of AGVs, including:

[0013] Construct a task set based on the total number of current containers, and construct an AGV group based on the total number of AGVs;

[0014] The optimization objective is to minimize the completion time of all tasks, the total travel distance of the AGV, and the weighted delay time of the tasks.

[0015] The multi-objective optimization model is constructed based on the optimization objective, the task set, and the AGV group, wherein the expression of the optimization objective function of the multi-objective optimization model is:

[0016] ,

[0017] ,

[0018] ,

[0019] ,

[0020] ,

[0021] Among them, T max D represents the completion time of all tasks. all T represents the total travel distance of all AGVs. wei Indicates the weighted delay time for the task. T represents max The weight parameters, D represents all The weight parameters, T represents wei The weight parameter, r j Z represents the j-th AGV in the AGV group. j Represents a task set. Indicates r j Complete task set Z j Distance traveled Indicates r j From Z j The Middle Task The starting point Drive to the destination The required distance Indicates r j From Z j The Middle Task The destination is reached by traveling to the (k+1)th task. The distance required from the starting point Indicates the speed of the AGV. Indicates the task execution time; n represents the number of AGVs; m represents the number of tasks; m j This represents the number of tasks in the task set, where k represents the index value of the number of tasks in the task set.

[0022] Furthermore, the multi-objective optimization model is solved to obtain multiple ordered subsets of tasks for the container, including:

[0023] Determine the individual coding scheme and construct the chromosome individual, wherein the chromosome individual includes at least the task sequence number of the task set and the breakpoint for dividing the task set;

[0024] Construct a fitness function, which is the reciprocal of the weighted sum of the completion time of all tasks, the total travel distance of the AGV, and the weighted delay time of the tasks;

[0025] The initial population is generated using a greedy algorithm, and a tournament strategy is used for selection. An elite retention strategy is combined to retain the individual with the highest fitness in the next generation of the population.

[0026] Genetic operations are performed on the initial population to generate a new subpopulation, wherein the genetic operations include self-crossing, parental crossing, and single-point mutation;

[0027] The genetic operations described above are iterated according to a preset iteration termination condition until multiple ordered task subsets of the container are obtained, wherein the preset iteration termination condition includes reaching the maximum number of iterations and / or the population fitness tending to stabilize.

[0028] Further, determine the individual coding scheme, including:

[0029] The task numbers in the task set are encoded as positive integers, and the breakpoints in the task set are encoded as 0. The task set is then divided according to the encoded task numbers and breakpoints to obtain task subsets.

[0030] Representing the gene locations on an individual chromosome as the execution order of tasks within a task subset yields an ordered task subset.

[0031] Furthermore, the fitness function is constructed, including:

[0032] Calculate the evaluation vector for each chromosome individual, the evaluation vector including the completion time of all tasks, the total distance traveled by the AGV, and the task-weighted delay time;

[0033] The weighted sum of the evaluation vectors is calculated based on the multi-objective optimization model, and its reciprocal is taken as the fitness value to obtain the fitness function. The expression of the fitness function is as follows:

[0034] ,

[0035] ,

[0036] ,

[0037] ,

[0038] ,

[0039] ,

[0040] ,

[0041] ,

[0042] in, This represents the optimization target for the completion time of all tasks. This represents the optimization objective for the total distance traveled by the AGV. This represents the optimization objective for the task-weighted delay time. This represents the travel path for the j-th part of chromosome individual q to complete its current task. This represents the completion time of the task for the j-th part of chromosome individual q. This represents the weighted delay for the task of the j-th part on chromosome individual q. Indicates task The end point to the mission Distance from the starting point This indicates that the j-th part starts executing from the first task and accumulates the travel distance up to the k-th task. When k=1, , Indicates task The distance from the starting point to the end point. Indicates the index of cumulative tasks.

[0043] Furthermore, an initial population is generated according to a greedy algorithm, including:

[0044] Randomly select several tasks as the starting tasks for each ordered subset of tasks;

[0045] In each ordered task subset, the task that minimizes the fitness function is prioritized as the final ordered task subset, until all tasks have been encoded.

[0046] Insert breakpoints at the boundaries of ordered subsets of tasks to generate an initial population;

[0047] Repeat the above operation until the number of individuals in the initial population meets the preset number of individuals.

[0048] Further, genetic operations are performed on the initial population to generate new subpopulations, including:

[0049] According to the preset crossover rate and preset mutation rate, crossover and mutation operations are performed on each chromosome individual in the initial population to obtain a new subpopulation, wherein the crossover operation includes self-crossover operation and parent crossover operation, and the mutation operation includes single-point mutation operation;

[0050] The self-crossover operation involves selecting two regions on each chromosome individual and exchanging their positions. The parental crossover operation involves selecting the same position regions of two parent chromosome individuals, exchanging them, and repairing illegal solutions. The single-point mutation operation involves mutating one gene in each chromosome individual to another coding value and repairing illegal solutions.

[0051] Further, a task allocation cost matrix is ​​constructed based on the bidding value of each AGV for the ordered task subset, and the matching relationship between the ordered task subset and the AGV is determined based on the solution of the task allocation cost matrix, including:

[0052] The bid value of each AGV for the ordered task subset is calculated based on its own position and the Manhattan distance from the starting point of the first task in the ordered task subset.

[0053] A task allocation cost matrix is ​​constructed based on the bidding value of each AGV for an ordered subset of tasks, wherein the elements of the task allocation cost matrix are the bidding values ​​of each AGV for an ordered subset of tasks.

[0054] The task allocation cost matrix is ​​solved using the Hungarian algorithm to obtain the optimal matching result between the ordered task subset and the AGV.

[0055] Based on the optimal matching result, the ordered task subset is assigned to the corresponding AGV.

[0056] Furthermore, the bidding value of each AGV for the ordered task subset is calculated based on the Manhattan distance between each AGV's own position and the starting point of the first task in the ordered task subset, including:

[0057] Obtain the current position coordinates of each AGV and the starting coordinates of the first task in the ordered task subset;

[0058] The Manhattan distance between the current position of each AGV and the starting position of the first task in the ordered task subset is calculated based on the current position coordinates of each AGV and the starting position of the ordered task subset. The result of the Manhattan distance calculation is used as the bidding value, and the expression for the bidding value is as follows:

[0059] ,

[0060] in, Indicates r j The bid values ​​for an ordered subset of tasks, and represented in the task allocation cost matrix. The middle is located in the i-th row and j-th column, r j This represents the j-th AGV in the AGV group.

[0061] Furthermore, a task allocation cost matrix is ​​constructed based on the bidding value of each AGV for an ordered subset of tasks, including:

[0062] The bidding values ​​are arranged according to the index of the AGV and the index of the ordered task subset, forming a two-dimensional matrix;

[0063] The dimension of the two-dimensional matrix is ​​set to be the same as the number of AGVs and the number of ordered task subsets to obtain the task allocation cost matrix.

[0064] The task allocation method for multiple AGVs in an automated container terminal provided by this invention constructs a multi-objective optimization model based on the total number of tasks for containers and the total number of AGVs. The model is then solved to obtain multiple ordered subsets of tasks for containers. A task allocation cost matrix is ​​constructed based on the bidding value of each AGV for its ordered subset. The matching relationship between the ordered subsets and AGVs is determined based on the solution to the task allocation cost matrix. Finally, the container tasks to be executed by each AGV are determined based on the matching relationship between the ordered subsets and AGVs, and the corresponding container tasks are sent to the corresponding AGVs. This task allocation method for multiple AGVs in an automated container terminal adopts a hybrid scheduling framework of centralized partitioning and distributed allocation, decoupling the complex task allocation problem into two stages. In the centralized stage, a large number of tasks are planned into several high-quality task subsets with a defined execution order from a global perspective. The optimal assignment between task subsets and AGVs is achieved in the distributed phase, which effectively solves the problem of difficulty in balancing solution quality and computational efficiency in existing technologies. While obtaining a high-quality scheduling scheme that is close to the centralized global optimal solution, the solution time is significantly shortened, meeting the real-time requirements of dock operations. Attached Figure Description

[0065] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the following detailed description to explain the invention, but do not constitute a limitation thereof.

[0066] Figure 1 A flowchart of the task allocation method for multiple AGVs in an automated container terminal provided by the present invention.

[0067] Figure 2 The flowchart for constructing a multi-objective optimization model provided by the present invention.

[0068] Figure 3 This is a flowchart for solving a multi-objective optimization model provided by the present invention.

[0069] Figure 4 This is a schematic diagram of individual chromosome encoding provided by the present invention.

[0070] Figure 5 This is a schematic diagram of self-crossing operation provided by the present invention.

[0071] Figure 6 This is a schematic diagram of the parent crossover operation provided by the present invention.

[0072] Figure 7 This is a schematic diagram of a single-point mutation operation provided by the present invention.

[0073] Figure 8This is a schematic diagram illustrating the matching relationship between a subset of ordered tasks and an AGV, as provided by the present invention.

[0074] Figure 9 This is a schematic diagram illustrating the solution of the task allocation cost matrix provided by the present invention. Detailed Implementation

[0075] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0076] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0077] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of the invention described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0078] This embodiment provides a method for task allocation among multiple AGVs in an automated container terminal. Figure 1 This is a flowchart of a multi-AGV task allocation method for automated container terminals provided by an embodiment of the present invention, such as... Figure 1 As shown, it includes:

[0079] S100. Construct a multi-objective optimization model based on the current total number of container tasks and the total number of AGVs. The optimization objectives of the multi-objective optimization model include at least minimizing the completion time of all tasks, the total travel distance of AGVs, and the task weighted delay time.

[0080] In this embodiment of the invention, a multi-objective optimization model is constructed based on the number of tasks in the container and the number of AGVs. The multi-objective optimization model aims to minimize the total task completion time, the total AGV travel distance, and the task weighted delay time.

[0081] S200. Solve the multi-objective optimization model to obtain multiple ordered task subsets of the container, wherein the number of ordered task subsets is the same as the number of AGVs.

[0082] In this embodiment of the invention, a clustering algorithm can be used to solve the multi-objective optimization model, dividing the task set into an ordered task subset equal to the number of AGVs. The clustering algorithm may specifically include a genetic algorithm.

[0083] S300. Construct a task allocation cost matrix based on the bidding value of each AGV for the ordered task subset, and determine the matching relationship between the ordered task subset and the AGV based on the solution result of the task allocation cost matrix.

[0084] In this embodiment of the invention, a task allocation cost matrix can be constructed from the bidding values ​​of the ordered task subset, and then the task allocation cost matrix can be solved based on the allocation algorithm to obtain the matching relationship between the ordered task subset and the AGV.

[0085] S400. Determine the container task to be performed by each AGV based on the matching relationship between the ordered task subset and the AGV, and send the corresponding container task to the corresponding AGV.

[0086] In this embodiment of the invention, the execution task of each AGV is determined according to the matching result, and the corresponding task is sent to the corresponding AGV. Each AGV performs transportation operations according to the allocated task subset.

[0087] Therefore, the task allocation method for multiple AGVs in an automated container terminal provided by this invention constructs a multi-objective optimization model based on the total number of tasks for the current containers and the total number of AGVs. The multi-objective optimization model is then solved to obtain multiple ordered task subsets for the containers. A task allocation cost matrix is ​​constructed based on the bidding value of each AGV for the ordered task subsets. The matching relationship between the ordered task subsets and AGVs is determined based on the solution result of the task allocation cost matrix. Finally, the container tasks to be executed by each AGV are determined based on the matching relationship between the ordered task subsets and AGVs, and the corresponding container tasks are sent to the corresponding AGVs. This task allocation method for multiple AGVs in an automated container terminal adopts a hybrid scheduling framework of centralized partitioning and distributed allocation, decoupling the complex task allocation problem into two stages. In the centralized stage, a large number of tasks are planned into several high-quality task subsets with a determined execution order from a global perspective. The optimal assignment between task subsets and AGVs is achieved in the distributed phase, which effectively solves the problem of difficulty in balancing solution quality and computational efficiency in existing technologies. While obtaining a high-quality scheduling scheme that is close to the centralized global optimal solution, the solution time is significantly shortened, meeting the real-time requirements of dock operations.

[0088] In this embodiment, a multi-objective optimization model is constructed based on the total number of current container tasks and the total number of AGVs, such as... Figure 2 As shown, it includes:

[0089] S110. Construct a task set based on the total number of tasks in the current containers, and construct an AGV group based on the total number of AGVs;

[0090] In this embodiment of the invention, the number of tasks and the number of AGVs in the scheduling system are determined, and the task set T and AGV group R in the multi-task allocation problem of an automated terminal scenario are modeled. Specifically:

[0091] ,

[0092] ,

[0093] ,

[0094] ,

[0095] ,

[0096] in, Let m represent the i-th task in task set T, and m represent the number of tasks. Indicates task starting point coordinates Indicates task end coordinates Indicates task Generation time, Indicates task Execution priority; r j This represents the j-th AGV in the AGV group. Indicates the number of AGVs; Indicates AGV r j Starting position coordinates Indicates AGV r j Current location The coordinates, Z j Indicates r j The local task set, which includes One task, This indicates the virtual starting task of the AGV's local task set.

[0097] S120. The optimization objective is to minimize the completion time of all tasks, the total travel distance of the AGV, and the weighted delay time of the tasks.

[0098] S130. Construct the multi-objective optimization model based on the optimization objective, the task set, and the AGV group, wherein the expression for the optimization objective function of the multi-objective optimization model is:

[0099] ,

[0100] ,

[0101] ,

[0102] ,

[0103] ,

[0104] Among them, T max D represents the completion time of all tasks. all T represents the total travel distance of all AGVs. wei Indicates the weighted delay time for the task. T represents max The weight parameters, D represents all The weight parameters, T represents wei The weight parameter, r j Z represents the j-th AGV in the AGV group. j Represents a task set. Indicates r j Complete task set Z j Distance traveled Indicates r j From Z j The Middle Task The starting point Drive to the destination The required distance Indicates r j From Z j The Middle Task The destination is reached by traveling to the (k+1)th task. The distance required from the starting point Indicates the speed of the AGV. Indicates the task execution time; n represents the number of AGVs; m represents the number of tasks; m j This represents the number of tasks in the task set, where k represents the index value of the number of tasks in the task set.

[0105] In embodiments of the present invention, such as Figure 3 As shown, solving the multi-objective optimization model yields multiple ordered subsets of tasks for the container, including:

[0106] S210. Determine the individual coding scheme and construct the chromosome individual, wherein the chromosome individual includes at least the task sequence number of the task set and the breakpoint for dividing the task set;

[0107] In this embodiment of the invention, the specific process of solving a multi-objective optimization model is illustrated using a genetic algorithm as an example.

[0108] Specifically, determine the individual coding scheme, such as Figure 4 As shown, it includes:

[0109] 1) Encode the task numbers in the task set as positive integers and the breakpoints in the task set as 0. Divide the task set according to the encoded task numbers and breakpoints to obtain task subsets.

[0110] 2) Represent the gene positions on individual chromosomes as the execution order of tasks in a task subset to obtain an ordered task subset.

[0111] S220. Construct a fitness function, which is the reciprocal of the weighted sum of the completion time of all tasks, the total travel distance of the AGV, and the weighted delay time of the tasks.

[0112] In this embodiment of the invention, constructing the fitness function includes:

[0113] 1) Calculate the evaluation vector for each chromosome individual, wherein the evaluation vector includes the completion time of all tasks, the total distance traveled by the AGV, and the task-weighted delay time;

[0114] 2) Calculate the weighted sum of the evaluation vectors according to the multi-objective optimization model, and take its reciprocal as the fitness value to obtain the fitness function. The expression of the fitness function is:

[0115] ,

[0116] ,

[0117] ,

[0118] ,

[0119] ,

[0120] ,

[0121] ,

[0122] ,

[0123] in, This represents the optimization target for the completion time of all tasks. This represents the optimization objective for the total distance traveled by the AGV. This represents the optimization objective for the task-weighted delay time. This represents the travel path for the j-th part of chromosome individual q to complete its current task. This represents the completion time of the task for the j-th part of chromosome individual q. This represents the weighted delay for the task of the j-th part on chromosome individual q. Indicates task The end point to the mission Distance from the starting point This indicates that the j-th part starts executing from the first task and accumulates the travel distance up to the k-th task. When k=1, , Indicates task The distance from the starting point to the end point. Indicates the index of cumulative tasks.

[0124] It should be understood that when hour, This means that it is assumed that the AGV will immediately begin to perform the first task.

[0125] S230. Generate an initial population using a greedy algorithm, and use a tournament strategy for selection, combined with an elite retention strategy to retain the individual with the highest fitness in the next generation of the population.

[0126] In this embodiment of the invention, generating an initial population according to a greedy algorithm includes:

[0127] 1) Randomly select several tasks as the starting tasks for each ordered subset of tasks;

[0128] 2) In each ordered task subset, the task that minimizes the fitness function is selected as the final ordered task subset, until all tasks are encoded.

[0129] 3) Insert breakpoints at the boundaries of ordered task subsets to generate the initial population;

[0130] 4) Repeat the above operation until the number of individuals in the initial population meets the preset number of individuals.

[0131] S240. Perform genetic operations on the initial population to generate a new subpopulation, wherein the genetic operations include self-crossing, parental crossing, and single-point mutation.

[0132] In this embodiment of the invention, genetic operations such as self-crossing, parental crossing, and single-point mutation are performed to generate new subpopulations. Specifically, performing genetic operations on the initial population to generate new subpopulations includes:

[0133] According to the preset crossover rate and preset mutation rate, crossover and mutation operations are performed on each chromosome individual in the initial population to obtain a new subpopulation, wherein the crossover operation includes self-crossover operation and parent crossover operation, and the mutation operation includes single-point mutation operation;

[0134] The self-crossover operation involves selecting two regions on each chromosome individual and exchanging their positions. The parental crossover operation involves selecting the same position regions of two parent chromosome individuals, exchanging them, and repairing illegal solutions. The single-point mutation operation involves mutating one gene in each chromosome individual to another coding value and repairing illegal solutions.

[0135] Based on the set crossover rate mutation rate Each chromosome in an individual has a chance of crossover and mutation; crossover operations are divided into self-crossover operations (such as...). Figure 5 (as shown) and parent crossover operations (such as) Figure 6 As shown), the mutation operation is a single-point mutation operation (such as...). Figure 7 (As shown); the self-crossover operation is to select two regions on an individual's chromosome and exchange their positions; the parental crossover operation is to select the same position regions of two parent individuals and exchange them, and repair any illegal solutions that may be generated; the single-point mutation operation is to mutate one gene in an individual to another coding value and repair illegal solutions.

[0136] S250. The genetic operation is iterated according to the preset iteration termination condition until multiple ordered task subsets of the container are obtained, wherein the preset iteration termination condition includes reaching the maximum number of iterations and / or the population fitness tends to stabilize.

[0137] In this embodiment of the invention, a task allocation cost matrix is ​​constructed based on the bidding value of each AGV for an ordered subset of tasks, and the matching relationship between the ordered subset of tasks and the AGV is determined based on the solution result of the task allocation cost matrix. Figure 8 As shown, it includes:

[0138] S310. Calculate the bid value of each AGV for the ordered task subset based on the Manhattan distance between each AGV's own position and the starting point of the first task in the ordered task subset.

[0139] In this embodiment of the invention, for each ordered task subset, the bidding value needs to be calculated based on the Manhattan distance between the position of each AGV and the starting point of the ordered task subset.

[0140] Specifically, the bidding value of each AGV for the ordered task subset is calculated based on its own position and the Manhattan distance from the starting point of the first task in the ordered task subset, including:

[0141] 1) Obtain the current position coordinates of each AGV and the starting coordinates of the first task in the ordered task subset;

[0142] 2) Calculate the Manhattan distance between the current position of the AGV and the starting position of the first task in the ordered task subset based on the current position coordinates of each AGV and the starting coordinates of the first task in the ordered task subset. Use the calculated Manhattan distance as the bidding value. The expression for the bidding value is:

[0143] ,

[0144] in, Indicates r j The bid values ​​for an ordered subset of tasks, and represented in the task allocation cost matrix. The middle is located in the i-th row and j-th column, r j This represents the j-th AGV in the AGV group.

[0145] S320. Construct a task allocation cost matrix based on the bidding value of each AGV for an ordered subset of tasks, wherein the elements of the task allocation cost matrix are the bidding values ​​of each AGV for an ordered subset of tasks.

[0146] Specifically, a task allocation cost matrix is ​​constructed based on the bidding value of each AGV for an ordered subset of tasks, including:

[0147] 1) Arrange the bidding values ​​according to the index of the AGV and the index of the ordered task subset to form a two-dimensional matrix;

[0148] 2) Set the dimension of the two-dimensional matrix to be the same as the number of AGVs and the number of ordered task subsets to obtain the task allocation cost matrix.

[0149] It should be understood that the dimension of the task allocation cost matrix in the embodiments of the present invention is the same as the number of AGVs and the number of ordered task subsets.

[0150] S330. Solve the task allocation cost matrix according to the Hungarian algorithm to obtain the optimal matching result between the ordered task subset and the AGV;

[0151] In this embodiment of the invention, the optimal allocation scheme of the task allocation cost matrix is ​​solved with the goal of minimizing the total cost; the allocation result is output, ensuring that each ordered subset of tasks is assigned to a unique AGV, such as... Figure 9 The diagram shown illustrates the solution to the task allocation cost matrix.

[0152] S340. Based on the optimal matching result, the ordered task subset is assigned to the corresponding AGV.

[0153] In this embodiment of the invention, when the AGV receives separate ordered task subsets, the AGV performs transportation operations according to the assigned ordered task subsets. Specifically, the AGV receives the assigned ordered task subsets and their execution order; the AGV moves from its current position to the first task starting point and completes the transportation operations in sequence according to the order in the ordered task subsets.

[0154] In summary, the task allocation method for multiple AGVs in automated container terminals provided by this invention effectively solves the problem of balancing solution quality and computational efficiency in existing technologies through an innovative hybrid framework of centralized partitioning and distributed allocation. The embodiments of this invention achieve a high-quality scheduling scheme approaching the centralized global optimal solution while significantly shortening the solution time, meeting the real-time requirements of terminal operations. Its multi-objective optimization model comprehensively considers completion time, travel distance, and task delay, resulting in more comprehensive scheduling results, effectively improving the overall system operating efficiency and reducing AGV energy consumption and losses. This method exhibits stable and superior performance under different operating conditions and strong robustness. By periodically executing the algorithm, it can effectively handle newly arriving tasks and large-scale tasks. Furthermore, by periodically running the scheduling algorithm, it can efficiently process new tasks continuously generated by the dynamic arrival of ships in automated terminals, thereby meeting the continuous scheduling requirements in real-world operational scenarios. Therefore, this invention provides an efficient, flexible, and reliable technical solution for multi-AGV scheduling in automated ports, and has significant practical value for improving port operation efficiency and resource utilization.

[0155] It is understood that the above embodiments are merely exemplary implementations used to illustrate the principles of the present invention, and the present invention is not limited thereto. For those skilled in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also considered to be within the scope of protection of the present invention.

Claims

1. A method for task allocation among multiple AGVs in an automated container terminal, characterized in that, include: A multi-objective optimization model is constructed based on the total number of current container tasks and the total number of AGVs. The optimization objectives of the multi-objective optimization model include at least minimizing the completion time of all tasks, the total travel distance of AGVs, and the task-weighted delay time. Solving the multi-objective optimization model yields multiple ordered task subsets for the container, wherein the number of ordered task subsets is the same as the number of AGVs. A task allocation cost matrix is ​​constructed based on the bidding value of each AGV for an ordered task subset, and the matching relationship between the ordered task subset and the AGV is determined based on the solution of the task allocation cost matrix. Based on the matching relationship between the ordered task subset and the AGV, the container task to be performed by each AGV is determined, and the corresponding container task is sent to the corresponding AGV. The multi-objective optimization model is constructed based on the total number of containers and the total number of AGVs, including: Construct a task set based on the total number of current containers, and construct an AGV group based on the total number of AGVs; The optimization objective is to minimize the completion time of all tasks, the total travel distance of the AGV, and the weighted delay time of the tasks. The multi-objective optimization model is constructed based on the optimization objective, the task set, and the AGV group, wherein the expression of the optimization objective function of the multi-objective optimization model is: ， ， ， ， ， Among them, T max D represents the completion time of all tasks. all T represents the total travel distance of all AGVs. wei Indicates the weighted delay time for the task. T represents max The weight parameters, D represents all The weight parameters, T represents wei The weight parameter, r j Z represents the j-th AGV in the AGV group. j Represents a task set. Indicates r j Complete task set Z j Distance traveled Indicates r j From Z j The Middle Task The starting point Drive to the destination The required distance Indicates r j From Z j The Middle Task The destination is reached by traveling to the (k+1)th task. The distance required from the starting point Indicates the speed of the AGV. Indicates the task execution time; n represents the number of AGVs; m represents the number of tasks; m j This represents the number of tasks in the task set, where k represents the index value of the number of tasks in the task set.

2. The task allocation method for multiple AGVs in an automated container terminal according to claim 1, characterized in that, Solving the multi-objective optimization model yields multiple ordered subsets of tasks for the container, including: Determine the individual coding scheme and construct the chromosome individual, wherein the chromosome individual includes at least the task sequence number of the task set and the breakpoint for dividing the task set; Construct a fitness function, which is the reciprocal of the weighted sum of the completion time of all tasks, the total travel distance of the AGV, and the weighted delay time of the tasks; The initial population is generated using a greedy algorithm, and a tournament strategy is used for selection. An elite retention strategy is combined to retain the individual with the highest fitness in the next generation of the population. Genetic operations are performed on the initial population to generate a new subpopulation, wherein the genetic operations include self-crossing, parental crossing, and single-point mutation; The genetic operations described above are iterated according to a preset iteration termination condition until multiple ordered task subsets of the container are obtained, wherein the preset iteration termination condition includes reaching the maximum number of iterations and / or the population fitness tending to stabilize.

3. The task allocation method for multiple AGVs in an automated container terminal according to claim 2, characterized in that, Determine the individual coding scheme, including: The task numbers in the task set are encoded as positive integers, and the breakpoints in the task set are encoded as 0. The task set is then divided according to the encoded task numbers and breakpoints to obtain task subsets. Representing the gene locations on an individual chromosome as the execution order of tasks within a task subset yields an ordered task subset.

4. The task allocation method for multiple AGVs in an automated container terminal according to claim 2, characterized in that, Constructing the fitness function includes: Calculate the evaluation vector for each chromosome individual, which includes the completion time of all tasks, the total distance traveled by the AGV, and the task-weighted delay time; The weighted sum of the evaluation vectors is calculated based on the multi-objective optimization model, and its reciprocal is taken as the fitness value to obtain the fitness function. The expression of the fitness function is as follows: ， ， ， ， ， ， ， ， in, This represents the optimization target for the completion time of all tasks. This represents the optimization objective for the total distance traveled by the AGV. This represents the optimization objective for the weighted delay time of the task. This represents the travel path for the j-th part of chromosome individual q to complete its current task. This represents the completion time of the task for the j-th part of chromosome individual q. This represents the weighted delay for the task of the j-th part on chromosome individual q. Indicates task The end point to the mission Distance from the starting point This indicates that the j-th part starts executing from the first task and accumulates the travel distance up to the k-th task. When k=1, , Indicates task The distance from the starting point to the end point. Indicates the index of cumulative tasks.

5. The task allocation method for multiple AGVs in an automated container terminal according to claim 2, characterized in that, The initial population is generated using a greedy algorithm, including: Randomly select several tasks as the starting tasks for each ordered subset of tasks; In each ordered task subset, the task that minimizes the fitness function is prioritized as the final ordered task subset, until all tasks have been encoded. Insert breakpoints at the boundaries of ordered subsets of tasks to generate an initial population; Repeat the above operation until the number of individuals in the initial population meets the preset number of individuals.

6. The task allocation method for multiple AGVs in an automated container terminal according to claim 2, characterized in that, Genetic operations are performed on the initial population to generate new subpopulations, including: According to the preset crossover rate and preset mutation rate, crossover and mutation operations are performed on each chromosome individual in the initial population to obtain a new subpopulation, wherein the crossover operation includes self-crossover operation and parent crossover operation, and the mutation operation includes single-point mutation operation; The self-crossover operation involves selecting two regions on each chromosome individual and exchanging their positions. The parental crossover operation involves selecting the same position regions of two parent chromosome individuals, exchanging them, and repairing illegal solutions. The single-point mutation operation involves mutating one gene in each chromosome individual to another coding value and repairing illegal solutions.

7. The task allocation method for multiple AGVs in an automated container terminal according to claim 1, characterized in that, A task allocation cost matrix is ​​constructed based on the bidding value of each AGV for an ordered subset of tasks. The matching relationship between the ordered subset of tasks and the AGVs is determined based on the solution to the task allocation cost matrix, including: The bid value of each AGV for the ordered task subset is calculated based on its own position and the Manhattan distance from the starting point of the first task in the ordered task subset. A task allocation cost matrix is ​​constructed based on the bidding value of each AGV for an ordered subset of tasks, wherein the elements of the task allocation cost matrix are the bidding values ​​of each AGV for an ordered subset of tasks. The task allocation cost matrix is ​​solved using the Hungarian algorithm to obtain the optimal matching result between the ordered task subset and the AGV. Based on the optimal matching result, the ordered task subset is assigned to the corresponding AGV.

8. The task allocation method for multiple AGVs in an automated container terminal according to claim 7, characterized in that, The bid value of each AGV for the ordered task subset is calculated based on its own position and the Manhattan distance from the starting point of the first task in the ordered task subset, including: Obtain the current position coordinates of each AGV and the starting coordinates of the first task in the ordered task subset; The Manhattan distance between the current position of each AGV and the starting position of the first task in the ordered task subset is calculated based on the current position coordinates of each AGV and the starting position of the ordered task subset. The result of the Manhattan distance calculation is used as the bidding value, and the expression for the bidding value is as follows: ， in, Indicates r j The bid values ​​for an ordered subset of tasks, and represented in the task allocation cost matrix. The middle is located in the i-th row and j-th column, r j This represents the j-th AGV in the AGV group.

9. The task allocation method for multiple AGVs in an automated container terminal according to claim 7, characterized in that, Construct a task allocation cost matrix based on the bidding value of each AGV for an ordered subset of tasks, including: The bidding values ​​are arranged according to the index of the AGV and the index of the ordered task subset, forming a two-dimensional matrix; The dimension of the two-dimensional matrix is ​​set to be the same as the number of AGVs and the number of ordered task subsets to obtain the task allocation cost matrix.

Images