Distributed task allocation method oriented to border management and control

Abstract

The invention discloses a border control-oriented distributed task allocation method, which comprises the following steps of: S1, dividing N tasks existing in a scene through a density clustering method to form Nc task clusters; s2, task allocation is performed on each task cluster, a task allocation problem is modeled as a mixed integer linear programming problem, and a distributed task allocation algorithm is adopted to solve an optimal task allocation scheme meeting task execution constraint conditions; and S3, introducing timestamps to represent the sequence of information interaction among the agents, and updating information according to a certain action rule to obtain conflict-free task distribution. According to the border management and control-oriented distributed task allocation method disclosed by the invention, a plurality of police forces can cooperatively execute a plurality of border management and control tasks, and each police force is allowed to simultaneously undertake a plurality of tasks, so that the utilization efficiency of police force resources and the border situation disposal capability are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of task allocation, and in particular relates to a distributed task allocation method for border control. Background Art

[0002] The border area is the first line of defense of the national security barrier. Border security, as an important part of national security, is related to the strategic situation of national sovereignty, security, and development interests. Border control task planning is an important way to jointly manage and control the border. It plays an important role in maintaining national border security, ensuring entry and exit order, preventing and controlling border armed conflicts, and developing good neighborly relations. At present, my country's border control still mainly relies on artificial generation schemes, lacks systematic integration, and even less involves intelligent technology. This limitation leads to low control efficiency and difficulty in responding to rapidly changing border security needs. Through systematic research, building a more scientific and efficient control system and integrating modern information technology can significantly improve the management capacity and emergency response speed of border security, thereby effectively maintaining national security and border stability.

[0003] The consensus-based bundle algorithm (CBBA) based on auction rules is a typical distributed task allocation strategy that aims to achieve efficient allocation of specific tasks. Although this method has certain advantages in task allocation, it is usually limited to the resource-task "one-to-one" or "many-to-one" model. When dealing with frequent and sporadic disposal tasks in border areas, due to the wide control range and relatively scarce police resources, the "one-to-one" or "many-to-one" allocation model shows obvious limitations, which seriously affects the efficiency of task disposal.

[0004] Therefore, further improvements are made to the above problems. Summary of the invention

[0005] The main purpose of this invention is to provide a distributed task allocation method for border control, in response to the major needs of "practical, intelligent, and systematic" border control, so as to improve the efficiency of border situation handling and the ability of border control.

[0006] Another object of the present invention is that the method can enable multiple police forces to coordinate in executing multiple border control tasks, and allow each police force to undertake multiple tasks at the same time, thereby improving the utilization efficiency of police resources and the ability to handle border situations.

[0007] To achieve the above objectives, the present invention provides a distributed task allocation method for border control, comprising the following steps:

[0008] Step S1: (Considering the frequent and scattered illegal border crossing incidents) divide the N tasks in the scene by density clustering method to form N c Task clusters;

[0009] Step S2: Assign tasks to each task cluster and model the task assignment problem as a mixed integer linear programming problem (the present invention uses "agent" to represent police resources). For a given task cluster T k and Police Resource Set A k , a distributed task allocation algorithm is adopted, that is, multiple agents independently perform distributed task allocation on the tasks within the cluster to solve the optimal task allocation scheme that meets the task execution constraints;

[0010] Step S3: (Taking into account the limitation of communication resources in border areas, there are different degrees of delay in communication between agents, so there may be conflicts in the task allocation results of each agent; in order to solve the conflict problem of task allocation between agents) introduce timestamps to represent the order of information interaction between agents, and update information according to certain action rules to obtain conflict-free task allocation.

[0011] As a further preferred technical solution of the above technical solution, for step S1:

[0012] The task set is denoted as The target set between task clusters satisfies:

[0013]

[0014] Formula (1) indicates that there is no intersection between any tasks in two task clusters; Formula (2) indicates that the sum of the number of tasks in all task clusters is equal to the total number of tasks in the scene. Assuming that a task cluster T k ={1,2,…,l…,N T} contains N T tasks, task l uses the element group <P l ,n l ,τ l >Description, where P l is the location of the task, n l The police resources required to perform the task l, n l >1 means that task l requires the coordination of multiple police forces to complete; is the time constraint set for executing the task, where and are the earliest and latest execution time nodes of task l, respectively. is the task duration, 0≤λ l <1 is the task time discount factor; by partitioning, the large-scale task allocation problem is reduced to Nc small-scale task allocation problems with no intersection (at this point, we only need to implement effective and conflict-free task allocation solutions under each small-scale task allocation problem to obtain an effective and conflict-free task allocation solution for the entire large-scale task allocation problem).

[0015] As a further preferred technical solution of the above technical solution, for step S2:

[0016] The task allocation model modeled as a mixed integer linear programming problem is described as:

[0017]

[0018] Among them, p m The order in which agent m executes the assigned tasks, i.e., the task path; S lm (p m ) is the agent m along the task path p m The benefit of executing task l; τ lm (p m ) is the agent m along the task path p m The estimated time to perform task l; x lm ∈{1,0} is the decision variable for task assignment; x lm =1 indicates that agent m performs task l, otherwise x lm =0;

[0019] Task Path Sort by the order of task execution, used to represent the list of tasks to be executed assigned to agent m,|p m | indicates the length of the list, The path list of agent m is empty. Assume that a certain police resource set A k ={1,2,…,m,…,N a}, the number of police resources it contains is N a , police resources m adopts element group express, and L m They represent the location of police resources and the upper limit of the number of tasks to be performed respectively.

[0020] As a further preferred technical solution of the above technical solution, the distributed task allocation algorithm of step S2 is specifically implemented as follows:

[0021] Step S2.1: (Considering that each agent can perform multiple tasks, the agent selects tasks based on the principle of maximizing its own task benefits each time) Let the initial task package of agent m be b m , in the task allocation process, agent m continuously selects the task l∈T with the largest marginal gain k Join b mUntil the mission pack is full or there are no optional missions, Arranged in the order in which the tasks are added to the package, used to represent the set of tasks assigned to agent m, |b m |Represents b m List length, Indicates that the task package of agent m is empty;

[0022] Step S2.2: Based on the new task l * Join m The benefits before and after, the marginal gain is expressed as:

[0023]

[0024] Where n represents l * In p m All possible positions in Indicates that l * Insert into path list p m The nth position in the original path list and the elements after it remain in the original order and move backwards one by one; Represents agent m along task path p m Benefits:

[0025]

[0026] Among them, r lm (p m ) is the task execution benefit. Considering the dynamic timeliness of the task, the time discount is used to calculate the benefit, that is:

[0027]

[0028] in, is the static benefit of task l, ξ lm The cost of travel loss is:

[0029] ξ lm =d lm γ (7)

[0030] Among them, γ is the travel cost coefficient, d lm is the distance between task l and agent m;

[0031] Step S2.3: Consider that the task (may) require multiple agents to collaborate to complete, that is, n l >1, let the number of agents assigned to perform task l be At this time, task l is divided into two cases: the first is that the number of agents assigned to perform task l does not meet the demand, that is, The second type is that the number of agents assigned to perform task l has met the demand, that is, For the second type, the bidding agent m compares its own peripheral gain with the minimum winning bid value of the current task.

[0032]

[0033] When the minimum winning bid value is replaced by the bid value of agent m, at the same time, in its winning bid matrix Y m of Fill in the bid value c lm , and the winning proxy matrix Z m of Set to 1, Z m The position corresponding to the minimum winning bid value is set to 0, and the winning bid value is expressed in the winning bid matrix Y m express, represents the bid value of agent k for task l from the perspective of agent m; the winning agent matrix is ​​represented by Z m Indicates that indicates that agent m believes that agent k is the winner of the bid for task l; otherwise Z m The sum of the non-zero elements in the lth column represents the total number of agents that agent m believes to be performing task l, Y m and Z m All are N a ×N T Dimensional matrix.

[0034] As a further preferred technical solution of the above technical solution, in step S3, the list of timestamps where s mg It represents the moment when agent m obtains the latest information from agent g. The conflict resolution of task allocation between agents includes the following steps:

[0035] Step S3.1: Resolve conflicts by comparing timestamps to ensure that the latest information is adopted. If agent k’s information is newer than agent m’s, update agent m’s winning bid matrix Y m and the winning proxy matrix Z m ;

[0036] Step S3.2: For each task l, check whether there is a conflict in task assignment, that is, check whether the winning bidding matrices of multiple agents for the same task are consistent. If not, replace the smaller bid value with the larger bid value and change the winning agent matrix accordingly.

[0037] The present invention has the following beneficial effects:

[0038] The present invention takes the realistic needs of "practical, intelligent and systematic" border control as the starting point, and proposes a distributed task allocation method for border control in view of the current problems of large border control scope, diverse target characteristics and insufficient handling capacity. The method supports three task allocation modes of resource-task "one-to-one", "many-to-one" and "many-to-many", which can effectively improve the utilization efficiency of police resources and the ability to handle border situations. By introducing timestamps to characterize the update of agent information, the conflict of task allocation between different agents can be effectively resolved, ensuring the smooth progress of control tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a framework diagram of the distributed task allocation method for border control of the present invention.

[0040] Figure 2 It is a density clustering diagram of the distributed task allocation method for border control of the present invention.

[0041] Figure 3 It is a flow chart of the distributed task allocation method for border control of the present invention. DETAILED DESCRIPTION

[0042] This patent is based on the national key R&D program "Research and application of border control space-time grid task planning and scheduling technology for precise perception and intelligent information rapid support technology for border control integrated with Skynet", project number: 2023YFC3305903.

[0043] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art can think of other obvious variations. The basic principles of the present invention defined in the following description can be applied to other embodiments, variations, improvements, equivalents, and other technical solutions that do not deviate from the spirit and scope of the present invention.

[0044] In the preferred embodiments of the present invention, those skilled in the art should note that border control and the like involved in the present invention may be regarded as prior art.

[0045] Preferred embodiments.

[0046] The present invention discloses a distributed task allocation method for border control, comprising the following steps:

[0047] Step S1: (Considering the frequent and scattered illegal border crossing incidents) divide the N tasks in the scene by density clustering method to form N c Task clusters;

[0048] Step S2: Assign tasks to each task cluster and model the task assignment problem as a mixed integer linear programming problem (the present invention uses "agent" to represent police resources). For a given task cluster T k and Police Resource Set A k , a distributed task allocation algorithm is adopted, that is, multiple agents independently perform distributed task allocation on the tasks within the cluster to solve the optimal task allocation scheme that meets the task execution constraints;

[0049] Step S3: (Taking into account the limitation of communication resources in border areas, there are different degrees of delay in communication between agents, so there may be conflicts in the task allocation results of each agent; in order to solve the conflict problem of task allocation between agents) introduce timestamps to represent the order of information interaction between agents, and update information according to certain action rules to obtain conflict-free task allocation.

[0050] Specifically, for step S1:

[0051] The task set is denoted as The target set between task clusters satisfies:

[0052]

[0053] Formula (1) indicates that there is no intersection between any tasks in two task clusters; Formula (2) indicates that the sum of the number of tasks in all task clusters is equal to the total number of tasks in the scene. Assuming that a task cluster T k ={1,2,…,l…,N T} contains N T tasks, task l uses the element group <P l ,n l ,τ l >Description, where P l is the location of the task, n l The police resources required to perform the task l, n l >1 means that task l requires the coordination of multiple police forces to complete; is the time constraint set for executing the task, where and are the earliest and latest time nodes for task l to be executed, respectively. is the task duration, 0≤λ l <1 is the task time discount factor; by partitioning, the large-scale task allocation problem is reduced to N c small-scale task allocation problems with no intersection (at this point, we only need to implement effective and conflict-free task allocation solutions under each small-scale task allocation problem to obtain an effective and conflict-free task allocation solution for the entire large-scale task allocation problem).

[0054] More specifically, for step S2:

[0055] The task allocation model modeled as a mixed integer linear programming problem is described as:

[0056]

[0057] Among them, p m The order in which agent m executes the assigned tasks, i.e., the task path; S lm (p m ) is the agent m along the task path p m The benefit of executing task l; τ lm (p m ) is the agent m along the task path p m The estimated time to perform task l; x lm ∈{1,0} is the decision variable for task assignment; x lm =1 indicates that agent m performs task l, otherwise x lm =0;

[0058] Task Path Sort by the order of task execution, used to represent the list of tasks to be executed assigned to agent m,|p m | indicates the length of the list, The path list of agent m is empty. Assume that a certain police resource set A k ={1,2,…,m,…,N a}, the number of police resources it contains is N a , police resources m adopts element group express, and L m They represent the location of police resources and the upper limit of the number of tasks to be performed respectively.

[0059] Furthermore, the distributed task allocation algorithm of step S2 is specifically implemented as the following steps:

[0060] Step S2.1: (Considering that each agent can perform multiple tasks, the agent selects tasks based on the principle of maximizing its own task benefits each time) Let the initial task package of agent m be b m , in the task allocation process, agent m continuously selects the task l∈T with the largest marginal gain k Join b m Until the mission pack is full or there are no optional missions, Arranged in the order in which the tasks are added to the package, used to represent the set of tasks assigned to agent m, |b m |Represents b m List length, Indicates that the task package of agent m is empty;

[0061] Step S2.2: Based on the new task l * Join m The benefits before and after, the marginal gain is expressed as:

[0062]

[0063] Where n represents l * In p m All possible positions in Indicates that l * Insert into path list p m The nth position in the original path list and the elements after it remain in the original order and move backwards one by one; Represents agent m along task path p m Benefits:

[0064]

[0065] Among them, r lm (p m ) is the task execution benefit. Considering the dynamic timeliness of the task, the time discount is used to calculate the benefit, that is:

[0066]

[0067] in, is the static benefit of task l, ξ lm The cost of travel loss is:

[0068] ξ lm =d lm γ (15)

[0069] Among them, γ is the travel cost coefficient, d lm is the distance between task l and agent m;

[0070] Step S2.3: Consider that the task (may) require multiple agents to collaborate to complete, that is, n l >1, let the number of agents assigned to perform task l be At this time, task l is divided into two cases: the first is that the number of agents assigned to perform task l does not meet the demand, that is, The second type is that the number of agents assigned to perform task l has met the demand, that is, For the second type, the bidding agent m compares its own peripheral gain with the minimum winning bid value of the current task.

[0071]

[0072] When the minimum winning bid value is replaced by the bid value of agent m, at the same time, in its winning bid matrix Y m of Fill in the bid value c lm , and the winning proxy matrix Z m of Set to 1, Z m The position corresponding to the minimum winning bid value is set to 0, and the winning bid value is expressed in the winning bid matrix Y m express, represents the bid value of agent k for task l from the perspective of agent m; the winning agent matrix is ​​represented by Z m Indicates that indicates that agent m believes that agent k is the winner of the bid for task l; otherwise Z m The sum of the non-zero elements in the lth column represents the total number of agents that agent m believes to be performing task l, Y m and Z m All are N a ×N T Dimensional matrix.

[0073] Furthermore, in step S3, the list of timestamps where s mg It represents the moment when agent m obtains the latest information from agent g. The conflict resolution of task allocation between agents includes the following steps:

[0074] Step S3.1: Resolve conflicts by comparing timestamps to ensure that the latest information is adopted. If agent k’s information is newer than agent m’s, update agent m’s winning bid matrix Y m and the winning proxy matrix Z m ;

[0075] Step S3.2: For each task l, check whether there is a conflict in task assignment, that is, check whether the winning bidding matrices of multiple agents for the same task are consistent. If not, replace the smaller bid value with the larger bid value and change the winning agent matrix accordingly.

[0076] It is worth mentioning that the technical features such as border control involved in the patent application of this invention should be regarded as prior art. The specific structure, working principle and possible control method and spatial layout method of these technical features can be selected according to the conventional methods in the field, and should not be regarded as the inventive point of this patent. This patent will not be further elaborated.

[0077] For those skilled in the art, it is still possible to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A distributed task allocation method for border control, characterized in that: The following steps are involved: Step S1: Divide the N tasks in the scene by density clustering to form N c Task clusters; Step S2: Assign tasks to each task cluster and model the task assignment problem as a mixed integer linear programming problem. For a given task cluster T k and Police Resource Set A k , a distributed task allocation algorithm is adopted, that is, multiple agents independently perform distributed task allocation on the tasks within the cluster to solve the optimal task allocation scheme that meets the task execution constraints; Step S3: Introduce timestamps to characterize the order of information interaction between agents, and update information according to certain action rules to obtain conflict-free task allocation.

2. A distributed task allocation method for border control according to claim 1, characterized in that: For step S1: The task set is denoted as The target set between task clusters satisfies: Formula (3) indicates that there is no intersection between any tasks in two task clusters; Formula (4) indicates that the sum of the number of tasks in all task clusters is equal to the total number of tasks in the scene. Assuming that a task cluster T k ={1,2,…,l…,N T } contains N T tasks, task l uses the element group <P l ,n l ,τ l >Description, where P l is the location of the task, n l The police resources required to perform the task l, n l >1 means that task l requires the coordination of multiple police forces to complete; is the time constraint set for executing the task, where and are the earliest and latest time nodes for task l to be executed, respectively. is the task duration, 0≤λ l <1 is the task time discount factor; by partitioning, the large-scale task allocation problem is reduced to N c A non-overlapping small-scale task allocation problem.

3. A distributed task allocation method for border control according to claim 2, characterized in that: For step S2: The task allocation model modeled as a mixed integer linear programming problem is described as: Among them, p m The order in which agent m executes the assigned tasks, i.e., the task path; S lm (p m ) is the agent m along the task path p m The benefit of executing task l; τ lm (p m ) is the agent m along the task path p m The estimated time to perform task l; x lm ∈{1,0} is the decision variable for task assignment; x lm =1 indicates that agent m performs task l, otherwise x lm =0; Task Path Sort by the order of task execution, used to represent the list of tasks to be executed assigned to agent m,|p m | indicates the length of the list, The path list of agent m is empty. Assume that a certain police resource set A k ={1,2,…,m,…,N a }, the number of police resources it contains is N a , police resources m adopts element group express, and L m They represent the location of police resources and the upper limit of the number of tasks to be performed respectively.

4. A distributed task allocation method for border control according to claim 3, characterized in that: The distributed task allocation algorithm of step S2 is specifically implemented as follows: Step S2.1: Assume that the initial task package of agent m is b m , in the task allocation process, agent m continuously selects the task l∈T with the largest marginal gain k Join b m Until the mission pack is full or there are no optional missions, Arranged in the order in which the tasks are added to the package, used to represent the set of tasks assigned to agent m, |b m |Represents b m List length, Indicates that the task package of agent m is empty; Step S2.2: Based on the new task l * Join b m The benefits before and after, the marginal gain is expressed as: Where n represents l * In p m All possible positions in Indicates that l * Insert into path list p m The nth position in the original path list and the elements after it remain in the original order and move backwards one by one; Represents agent m along task path p m Benefits: Among them, r lm (p m ) is the task execution benefit. Considering the dynamic timeliness of the task, the time discount is used to calculate the benefit, that is: in, is the static benefit of task l, ξ lm The cost of travel loss is: x lm =d lm c (9) Among them, γ is the travel cost coefficient, d lm is the distance between task l and agent m; Step S2.3: Consider that the task requires multiple agents to collaborate to complete, that is, n l >1, let the number of agents assigned to perform task l be At this time, task l is divided into two cases: the first is that the number of agents assigned to perform task l does not meet the demand, that is, The second type is that the number of agents assigned to perform task l has met the demand, that is, For the second type, the bidding agent m compares its own peripheral gain with the minimum winning bid value of the current task. When the minimum winning bid value is replaced by the bid value of agent m, at the same time, in its winning bid matrix Y m of Fill in the bid value c lm , and the winning proxy matrix Z m of Set to 1, Z m The position corresponding to the minimum winning bid value is set to 0, and the winning bid value is expressed in the winning bid matrix Y m express, represents the bid value of agent k for task l from the perspective of agent m; the winning agent matrix is ​​represented by Z m Indicates that indicates that agent m believes that agent k is the winner of the bid for task l; otherwise Z m The sum of the non-zero elements in the lth column represents the total number of agents that agent m believes to be performing task l, Y m and Z m All N a ×N T Dimensional matrix.

5. A distributed task allocation method for border control according to claim 4, characterized in that: In step S3, the list of timestamps where s mg It represents the moment when agent m obtains the latest information from agent g. The conflict resolution of task allocation between agents includes the following steps: Step S3.1: Resolve conflicts by comparing timestamps to ensure that the latest information is adopted. If agent k’s information is newer than agent m’s, update agent m’s winning bid matrix Y m and the winning proxy matrix Z m ; Step S3.2: For each task l, check whether there is a conflict in task assignment, that is, check whether the winning bidding matrices of multiple agents for the same task are consistent. If not, replace the smaller bid value with the larger bid value, and change the winning agent matrix accordingly.

Images