Rapid combat force recombination method based on system architecture model

Abstract

The invention relates to the technical field of combat network analysis, in particular to a rapid combat force recombination method based on a system architecture model, and provides the following scheme: abstracting entities in a combat system into nodes, and constructing a force deployment model comprising investigation nodes, command nodes and force nodes; generating a task demand sequence based on the initial attribute of the target node in a random enemy; establishing a weapon strength distribution optimization model, and generating a weapon strength distribution scheme by taking the task priority satisfaction degree, the resource utilization rate and the load balance as optimization targets and combining a reinforcement learning technology; on the premise that resource, time and path feasibility constraint conditions are met, an optimal path from the military strength node to the target node is planned; according to the method, task scheduling and military strength distribution optimization in a dynamic battlefield environment are realized, the combat response speed and the resource utilization efficiency are improved, and the method is suitable for a multi-task cooperative combat scene.

Description

Technical Field

[0001] The present invention relates to the technical field of combat network analysis, and in particular to a method for rapid reorganization of combat forces based on a system architecture model. Background Art

[0002] In the current complex battlefield environment, the dynamic and diverse nature of combat missions has put forward higher requirements for force deployment and scheduling. In multi-task, multi-node collaborative combat scenarios, the optimization of force allocation not only needs to consider resource constraints, but also needs to meet multi-dimensional constraints such as task priority, time window, and path feasibility. However, the existing force scheduling methods mostly use static allocation models, which cannot adapt to the dynamic changes in the battlefield environment, such as the random appearance of enemy situations, fluctuations in path accessibility, and limited resources. In addition, traditional methods lack effective mechanisms in task conflict detection, dynamic path planning, and load balancing, which can easily lead to resource waste or task failure.

[0003] For example, the patent application with the publication number CN115641235A provides an analysis method for the high-level structure of a heterogeneous combat network based on the spectral clustering of combat models, including taking a subgraph pattern composed of multiple equipment nodes connected in the combat network as a combat model; converting the combat network into a combat model adjacency matrix; dividing the corresponding combat network into a set of interconnected slices according to the combat model adjacency matrix; for each interconnected slice in the set of interconnected slices, performing spectral clustering on its combat model adjacency matrix, and gradually dividing it into clusters whose node scale is not greater than a set threshold; summarizing all clusters of combat models in the interconnected slices to obtain functional clusters of combat models, and then summarizing the functional clusters of all combat models of interest to obtain a set of functional clusters of different combat models in the complete combat network. High-order functional clusters in the combat network can be discovered based on different combat models of interest.

[0004] The above patents have the problems raised by this background technology: it is difficult to balance task priority, resource utilization and load balancing in a dynamic battlefield environment, and there is a lack of effective dynamic feedback mechanism and real-time path planning capabilities, resulting in insufficient flexibility and efficiency in force allocation. To solve the above problems, this application designs a method for rapid reorganization of combat forces based on a system architecture model. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a method for rapid reorganization of combat forces based on an architecture model in view of the deficiencies of the prior art. By abstracting the entities in the combat system into nodes, a force deployment model including reconnaissance nodes, command nodes and force nodes is constructed; under random enemy conditions, a task requirement sequence is generated based on the initial attributes of the target node; a force allocation optimization model is established, with task priority satisfaction, resource utilization and load balancing as optimization goals, and a force allocation plan is generated in combination with reinforcement learning technology; and the optimal path from the force node to the target node is planned under the premise of satisfying resource, time and path feasibility constraints. The method realizes task scheduling and force allocation optimization in a dynamic battlefield environment, improves combat response speed and resource utilization efficiency, and is suitable for multi-task collaborative combat scenarios.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A method for rapid reorganization of combat forces based on a system architecture model, the method comprising:

[0008] Abstract the entities in the combat system into nodes, and build a force deployment model based on the interaction between entities;

[0009] According to the random enemy situation detected by the node, the target node is updated in the force deployment model, and the task requirement sequence is generated according to the initial attributes of the target node;

[0010] According to the task requirement sequence and force deployment model, a force allocation optimization model is established, and according to resource constraints and task priorities, a force allocation plan is generated through the force allocation optimization model;

[0011] According to the force allocation plan, the optimal path for the forces from the force node to the target node is planned.

[0012] The constructing of the force deployment model comprises:

[0013] Divide the entities in the combat system into reconnaissance nodes, command nodes and force nodes according to their functions, and add attribute values ​​to each node according to the entity function, wherein the attribute values ​​include location coordinates, task status, force status, corresponding delay and task priority;

[0014] Establishing an interactive relationship according to the node type, wherein the interactive relationship includes information flow, instruction flow and resource flow;

[0015] Dynamic attributes are added to nodes and interaction relationships to build a force deployment model in the form of a directed graph.

[0016] The additional dynamic attributes include:

[0017] Add a blank task sequence, wherein the blank task sequence includes a task time window and a task timing dependency, the task time window is used to add time constraints to tasks and nodes, including the earliest task start time, the latest task completion time and the task duration, and the task timing dependency is used to establish a priority diagram between tasks according to the dependency relationship of multiple tasks;

[0018] Establish a task conflict detection mechanism. When a reconnaissance node or a force node is called by multiple tasks at the same time, adjust the resource usage order through task priority and task time window;

[0019] Through reconnaissance nodes, information on terrain changes is collected, including obstacle generation, path blockage, and changes in traversable width. Based on the battlefield terrain, a dynamic path feasibility map is established between nodes.

[0020] The generating of the task requirement sequence comprises:

[0021] Acquiring initial attributes of the task from the target node through the reconnaissance node, wherein the initial attributes include target location, threat level and task type;

[0022] The task priority is calculated according to the initial attribute, sorted in the task requirement sequence according to the task priority, and updated in the task requirement sequence according to descending order. The calculation formula of the task priority is:

[0023]

[0024] Among them, P represents the task priority, W represents the threat level of the target node, ω1 represents the weight of the threat level, and T s Indicates the earliest start time of the task, T e Indicates the latest completion time of the task, T c represents the current time, ω2 represents the weight of the time constraint, D represents the average distance between the force node and the target node, F represents the mean value of the path accessibility between the force node and the target node, ω3 represents the weight of the path constraint, and R d It represents the difference between the target node resource demand and the current remaining resources, and ω4 represents the weight of resource matching.

[0025] The optimization objectives of the force allocation optimization model include task priority satisfaction, resource utilization and force node load balance.

[0026] The constraints of the force allocation optimization model include:

[0027] The resource requirement of each task is less than or equal to the total resource capacity of the force node;

[0028] The total amount of resources allocated to each force node is less than or equal to its available capacity;

[0029] Resource allocation for each task is completed within its task time window;

[0030] The resource transfer path between the force node and the target node meets the path feasibility requirements.

[0031] The generating of the force allocation plan by the force allocation optimization model includes:

[0032] Input the task requirement sequence and the force deployment model to generate a task-node state graph, wherein the vertices of the task-node state graph represent task nodes and force nodes, the edges represent the resource allocation possibility between the task nodes and the force nodes, and the edge weight is determined by the task priority, path distance, pass probability and resource matching degree;

[0033] A reinforcement learning agent is created, and the state space, action space and reward function of the reinforcement learning agent are initialized, the task-node state diagram is used as an input parameter of the reinforcement learning agent, and a force allocation plan is output.

[0034] The calculation formula of the reward function is:

[0035]

[0036] Among them, Z represents the reward value, i represents a single task in the task requirement sequence, N represents the total number of tasks in the task requirement sequence, M represents the total number of force nodes assigned to task i, and j represents a single force node assigned to task i. represents the weight of the priority satisfaction, represents the weight of resource utilization, Represents the weight of load balancing, P i represents the task priority of task i, η represents the gain coefficient of resource allocation to task priority, R i,j represents the resource demand satisfaction of force node j for task i, u j represents the amount of resources allocated to force node j, U j represents the total resource capacity of force node j, λ j It represents the load standard deviation of force node j, and U represents the average value of resources allocated to all nodes in the force allocation scheme.

[0037] The reinforcement learning agent comprises:

[0038] According to the state space, the allocation probability between the task nodes and the force nodes is calculated, and the allocation probability is reorganized through a random strategy to generate an initial plan;

[0039] Perform constraint check on the initial plan in the action space to determine whether the constraint is satisfied. If not, regenerate the initial plan in the state space and set the reward value of this plan to 0. If satisfied, calculate the reward value of the initial plan according to the reward function.

[0040] Determine whether the reward value is less than the set threshold for three consecutive rounds. If so, output the initial plan corresponding to the maximum reward value in the three rounds as the troop allocation plan. If not, retain the initial plan in the state space and optimize the loss function of the random strategy according to the reward value to regenerate the initial plan.

[0041] The optimal path includes:

[0042] Initialize the path map, set the force node in the force allocation plan as the starting point, and set the target node as the end point to be marked in the path map;

[0043] The feasible paths between the force node and the target node are set as the edges of the path graph, and the weight of each edge represents the comprehensive cost of the path, wherein the comprehensive cost includes the path distance and the path resource requirement;

[0044] Select the edge with the smallest cumulative weight as the optimal path for the force node to reach the target node, traverse each force node in the force allocation plan, and output the optimal path plan.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] The present invention abstracts the entities in the combat system into nodes, constructs a dynamic force deployment model, combines reinforcement learning technology to optimize the task allocation plan, and uses real-time path planning to improve resource allocation efficiency and the reliability of task completion; at the same time, it comprehensively considers task priority, resource utilization and load balancing, and dynamically adapts to changes in the battlefield environment, thereby achieving efficient, flexible and intelligent rapid reorganization of forces, and improving combat response speed and task execution effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:

[0048] Figure 1 This is a flow chart of a method for rapid reorganization of combat forces based on a system architecture model according to Embodiment 1 of the present invention;

[0049] Figure 2 This is a schematic diagram of a network mode of a force deployment model according to Embodiment 1 of the present invention;

[0050] Figure 3This is a flow chart of troop allocation according to Embodiment 1 of the present invention. DETAILED DESCRIPTION

[0051] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0052] Example 1

[0053] See also Figure 1 The present invention provides an embodiment: a method for rapid reorganization of combat forces based on a system architecture model, and the specific steps of the method are as follows:

[0054] S1: Abstract the entities in the combat system into nodes, and build a force deployment model based on the interaction between entities;

[0055] In this step, the key entities in the combat system are abstracted into nodes, such as reconnaissance nodes, command nodes, and force nodes, and the nodes are classified according to their functions. The interaction between nodes is represented by information flow, command flow, and resource flow, and a dynamic directed graph model is constructed. Vertices represent nodes, edges represent interaction relationships, and attribute values ​​(such as location coordinates, task status, and resource status) are added to each node to reflect the structure and status of the combat system.

[0056] S2: Update the target node in the force deployment model and generate a sequence of mission requirements;

[0057] In this step, the target nodes in the force deployment model are updated based on the real-time enemy data provided by the reconnaissance node. The initial attributes of each target node include target location, threat level, and mission type. A mission requirement sequence is generated based on the attributes of the target node. The mission requirements include information such as priority, resource requirements, and time window. Priority sorting provides a basis for subsequent allocation;

[0058] S3: Build a force allocation optimization model, and generate a force allocation plan through the force allocation optimization model according to resource constraints and task priorities;

[0059] In this step, a force allocation optimization model is constructed based on the task requirement sequence and force deployment model, and the objective function and constraints are set. The optimization objectives include task priority satisfaction, resource utilization, and load balancing. Reinforcement learning technology is used to generate a force allocation plan, outputting the matching relationship between tasks and force nodes, resource allocation, and task execution priority;

[0060] S4: According to the force allocation plan, plan the optimal path from the force node to the target node;

[0061] In this step, the geographical information and path accessibility of the task nodes and force nodes are extracted according to the force allocation plan to build a path graph model. Through path weight calculation, taking into account factors such as distance, access probability and resource consumption, the graph search algorithm is used to plan the optimal path from the force node to the target node, and it is adjusted in real time in a dynamic battlefield environment;

[0062] Entities in the combat system refer to key elements that play specific functions in the combat process, including reconnaissance equipment, command centers, force units, etc. According to the functions of entities in combat, nodes are divided into the following categories:

[0063] Reconnaissance nodes, responsible for collecting battlefield intelligence, such as drones and radar equipment;

[0064] Command nodes, responsible for analyzing intelligence and making operational decisions, such as command centers;

[0065] Force nodes are responsible for executing combat missions, such as combat troops and firepower units;

[0066] Each node is assigned dynamic attributes to accurately describe its function and status in the battlefield situation. The attributes include:

[0067] Location coordinates, used to indicate the geographic location of a node, expressed in two-dimensional or three-dimensional coordinate form, for path planning and battlefield distribution calculation;

[0068] Task status, which indicates whether the node is executing a task, the task completion progress, and the current load status;

[0069] Force status, including the total amount of resources, remaining resources, resource allocation, and whether the node is overloaded;

[0070] Response latency, which is used to represent the response time of a node to a command, depends on communication delay, task complexity, and resource availability;

[0071] Mission priorities are dynamically calculated based on threat level, resource requirements, and time urgency;

[0072] See also Figure 2, a schematic diagram of the network mode of the force deployment model of the embodiment of the present invention, the reconnaissance part is responsible for collecting battlefield intelligence and reporting the enemy situation, such as the enemy's position, activity range and target characteristics. It is the starting point of the intelligence flow and provides basic information for the entire combat process. The command part is responsible for receiving the information transmitted by the reconnaissance node, analyzing the intelligence and generating decisions, and sending instructions to the combat node to command operations. It is the core node of information analysis and decision-making. The combat part receives the combat instructions from the command node and implements the corresponding combat tasks, including fire strikes, support tasks, etc. The target is the target of the combat operation, such as enemy facilities, important strongholds, etc. The combat node directly acts on the target node through resource allocation. The information flow transmits intelligence data from the reconnaissance node to the command node, such as the target location and enemy activity information, to support decision-making. The command flow is from the command node to the combat node to issue action instructions, such as specific strike targets or combat plans. The resource flow is from the combat node to the target node to deliver resources, such as fire support, material supply, etc., to complete the target task.

[0073] The specific steps of S1 are as follows:

[0074] S1.1: Divide the entities in the combat system into reconnaissance nodes, command nodes and force nodes according to their functions, ensure that each type of node corresponds to a clear task and attribute in the model, and add attribute values ​​to each node according to the entity function, where the attribute values ​​include location coordinates, task status, force status, corresponding delay and task priority;

[0075] Through node classification and attribute assignment, the model can reflect the dynamic status of each node in the combat system in real time, providing a basis for subsequent task allocation and path planning.

[0076] S1.2: Establishing an interactive relationship according to the node type, wherein the interactive relationship includes information flow, instruction flow and resource flow;

[0077] The information flow is initiated by the reconnaissance node, which transmits real-time intelligence to the command node. The command flow is initiated by the command node, which issues combat orders to the force node. The resource flow is initiated by the force node, which provides fire support, material supply and other resources to the target node. Through the above interactive relationship modeling, the collaboration between nodes can be accurately described, and a multi-level and multi-functional combat network can be built.

[0078] S1.3: Add dynamic attributes to nodes and interaction relationships to build a force deployment model in the form of a directed graph;

[0079] The abstract nodes and interactive relationships are combined to further construct a force deployment model. Vertices represent reconnaissance nodes, command nodes, and force nodes, and edges represent information flow, command flow, and resource flow between nodes. Both vertices and edges are attached with dynamic attributes to reflect changes in battlefield situations in real time. Vertex dynamic attributes include the task status, resource status, and position changes of nodes, and edge dynamic attributes include path accessibility, resource flow, and communication delay. The directed graph structure is dynamically updated based on reconnaissance node feedback and task execution. For example, if the target node location or threat level changes, the model will adjust the path weight and resource allocation plan in real time. If a path is blocked or communication fails, the edge weight will be recalculated to find an alternative path.

[0080] The additional dynamic attributes include:

[0081] Add a blank task sequence, wherein the blank task sequence includes a task time window and a task timing dependency, the task time window is used to add time constraints to tasks and nodes, including the earliest task start time, the latest task completion time and the task duration, and the task timing dependency is used to establish a priority diagram between tasks according to the dependency relationship of multiple tasks;

[0082] A blank task sequence is a preset task container that provides a framework for time constraints and dependencies for node and resource allocation before actual task generation. The task time window includes three key time parameters:

[0083] The earliest start time refers to the earliest time point at which a task can start, which is determined by the completion time of the upstream dependent tasks;

[0084] The latest completion time refers to the last time when the task must be completed, which is determined by the urgency of the task or strategic needs;

[0085] Task duration, which refers to the estimated time required to perform a task, is dynamically adjusted based on the nature of the task and node capabilities;

[0086] Generate a task priority graph based on task dependencies. In the graph, each task node is connected to its dependent task nodes. Use a topological sorting algorithm to sort the task nodes to ensure that the pre-dependencies of task assignment are met. The task priority graph can be dynamically updated according to the task priority.

[0087] Establish a task conflict detection mechanism. When a reconnaissance node or a force node is called by multiple tasks at the same time, adjust the order of resource use by task priority and task time window;

[0088] Through the reconnaissance nodes, terrain change information is collected, including obstacle generation, path blockage, and change in traversable width. Based on the battlefield terrain, a dynamic path feasibility map between nodes is established;

[0089] The specific steps of S2 are as follows:

[0090] S2.1: Obtain the initial attributes of the task from the target node through the reconnaissance node, wherein the initial attributes include the target location, which indicates the geographical coordinates of the target node and is used for path planning and force dispatch; the threat level, which is determined based on indicators such as enemy combat effectiveness and firepower intensity and is used for task priority calculation; the task type, which indicates the nature of the task, such as reconnaissance, strike, defense, etc., and determines the type and quantity of resources required for the task;

[0091] S2.2: Calculate the task priority according to the initial attribute, sort the task in the task requirement sequence according to the task priority, and update the task in the task requirement sequence according to descending order. The calculation formula of the task priority is:

[0092]

[0093] Among them, P represents the task priority, W represents the threat level of the target node, ω1 represents the weight of the threat level, and T s Indicates the earliest start time of the task, T e Indicates the latest completion time of the task, T c represents the current time, ω2 represents the weight of the time constraint, D represents the average distance between the force node and the target node, F represents the mean value of the path accessibility between the force node and the target node, ω3 represents the weight of the path constraint, and R d represents the difference between the target node resource demand and the current remaining resources, and ω4 represents the weight of resource matching;

[0094] By comprehensively considering threat levels, time constraints, path costs, and resource matching, we dynamically adjust task priorities to ensure scientific and flexible battlefield resource allocation, prioritize high-threat, high-urgency, and low-cost tasks, and optimize the order of task execution;

[0095] The optimization objectives of the force allocation optimization model include task priority satisfaction, resource utilization and force node load balance;

[0096] Task priority satisfaction reflects the degree of completion of high-priority tasks and is a key factor in optimizing goals. Prioritize high-threat and high-urgency tasks while taking into account path costs and resource matching. In actual applications, task priority satisfaction is achieved through dynamic resource allocation. High-priority tasks receive more allocation ratios under limited resource conditions. The real-time update mechanism of task priority ensures that high-priority tasks can quickly obtain resource support when the battlefield situation changes.

[0097] Resource utilization measures the efficiency of the use of force node resources. The goal is to reduce resource idleness and waste as much as possible through reasonable allocation. In the model, the resource allocation ratio of force nodes is dynamically adjusted according to the needs of the task, so that the resource allocation of each force node is as close to its available capacity as possible. The optimization model will dynamically monitor the resource usage status of force nodes and allocate more resource tasks to low-load nodes;

[0098] The goal of load balancing is to avoid inefficiency of force nodes due to resource overload or uneven resource allocation. Through dynamic regulation, the task load is evenly distributed to multiple nodes to ensure the collaborative combat capability between nodes. In the process of load balancing, task requirements, resource matching and path cost are comprehensively considered, and nodes with lower load are preferentially allocated to avoid affecting the overall combat efficiency due to overload of a single node, improve the collaborative capability between nodes, and ensure the smooth execution of combat missions;

[0099] The constraints of the force allocation optimization model include:

[0100] The resource requirement of each task is less than or equal to the total resource capacity of the force node. During the resource allocation process, ensure that the resource requirement of each task does not exceed the total resource capacity of the force node. Dynamically adjust based on the remaining resources of the node, give priority to nodes with sufficient resources, avoid task allocation failure due to insufficient resources, and improve task completion rate;

[0101] The total amount of resource allocation for each force node is less than or equal to its available capacity. The remaining available resources of the node are updated in real time. Each allocation in the allocation plan must meet the resource capacity constraint. When there is a conflict in task allocation, resources are reallocated through priority adjustment and time window optimization to avoid resource overflow, improve the rationality and feasibility of the allocation plan, and ensure the smooth execution of the task.

[0102] The resource allocation of each task is completed within its task time window. The time window constraint ensures that the resource allocation and execution of each task are completed within the available time range of the task. In actual implementation, by calculating the remaining time window of the task, the allocation priority is dynamically adjusted to ensure that time-critical tasks are completed first. The optimization model will sort the tasks according to the remaining time window and dynamically optimize the allocation order based on the resource allocation situation to ensure the timely completion of time-sensitive tasks and avoid delays that lead to task failures.

[0103] The resource transfer path between the force node and the target node meets the path feasibility requirements. In path planning, the path weight is dynamically updated in combination with the force deployment model, and paths with high passability and low cost are preferentially selected for allocation, thereby improving the efficiency and safety of resource transfer and reducing the risks in the transfer process.

[0104] See also Figure 3, the force allocation flow chart of the embodiment of the present invention, the force allocation plan is generated by the force allocation optimization model, including:

[0105] Input the task requirement sequence and the force deployment model to generate a task-node state graph, wherein the vertices of the task-node state graph represent task nodes and force nodes, the edges represent the resource allocation possibility between the task nodes and the force nodes, and the edge weight is determined by the task priority, path distance, pass probability and resource matching degree;

[0106] The construction of the state diagram is based on the dynamic interaction between task nodes and force nodes, aiming to represent the possibility of task allocation and its comprehensive evaluation factors. Task nodes represent specific task requirements, such as resource requirements, time windows, and task priorities, while force nodes represent the available resources for executing tasks. The weight of the edge comprehensively considers the matching degree between task nodes and force nodes;

[0107] Creating a reinforcement learning agent, initializing the state space, action space and reward function of the reinforcement learning agent, taking the task-node state graph as an input parameter of the reinforcement learning agent, and outputting a force allocation plan;

[0108] The state space takes the task-node state diagram as input, and includes the dynamic properties of the task node (such as task priority, remaining time window), the resource status of the force node (such as remaining resources, load), and the path passability between the task nodes. The state space is automatically updated after each task assignment to ensure that the agent perceives the latest battlefield situation;

[0109] Actions are defined as resource allocation behaviors between task nodes and force nodes. The agent evaluates the feasibility and benefits of each action based on the state space. The action space is dynamically adjusted to exclude infeasible resource allocation schemes, such as allocation actions with insufficient resources or inaccessible paths.

[0110] The reward function guides the agent to choose the optimal allocation strategy, taking into account the satisfaction of task priorities, resource utilization and load balance. The value of the reward depends on the completion of the task, the efficiency of resource allocation and the balance of the global load;

[0111] Specifically, in the process of generating a force allocation plan through the force allocation optimization model, the task requirement sequence and force deployment model are first input into the system to construct a task-node state graph, in which task nodes and force nodes are used as vertices, and edges represent the possibility of resource allocation between task nodes and force nodes. The weight of the edge is calculated by dynamic attributes such as task priority, path distance, pass probability and resource matching degree. The setting of task priority is dynamically adjusted based on factors such as the threat level and time urgency of the task. The path distance and pass probability reflect the geographical relationship between the force node and the target node and the dynamic changes of the battlefield environment. The resource matching degree is used to evaluate whether the current force node has the resource conditions to perform the task. Based on the construction of the task-node state graph, the reinforcement learning agent is initialized and receives the parameters of the state space, action space and reward function. The state space contains the priority, time window and resource requirements of the task node, as well as the current resource status, availability and path passability of the force node. The action space defines the resource allocation decision options between the task and the node. The reward function measures the pros and cons of the allocation plan through multi-objective optimization. Based on the input task-node state graph, the agent generates resource allocation plans through reinforcement learning strategy iteration, and dynamically updates task priorities and node states to meet the constraints of task time windows and resource requirements. Through reinforcement learning technology, the force allocation model can achieve efficient allocation in multi-task and dynamic battlefield environments, and optimize and adjust the allocation plan based on real-time feedback to ensure rapid response to high-priority tasks, while improving resource utilization and node load balancing, and finally generating a force allocation plan that meets the global optimization goal.

[0112] The calculation formula of the reward function is:

[0113]

[0114] Among them, Z represents the reward value, i represents a single task in the task requirement sequence, N represents the total number of tasks in the task requirement sequence, M represents the total number of force nodes assigned to task i, and j represents a single force node assigned to task i. represents the weight of the priority satisfaction, represents the weight of resource utilization, Represents the weight of load balancing, P i represents the task priority of task i, η represents the gain coefficient of resource allocation to task priority, R i,j represents the resource demand satisfaction of force node j for task i, u j represents the amount of resources allocated to force node j, U j represents the total resource capacity of force node j, λ j represents the load standard deviation of force node j, and U represents the average value of resources allocated to all nodes in the force allocation scheme;

[0115] The reinforcement learning agent comprises:

[0116] According to the state space, the allocation probability between the task nodes and the force nodes is calculated, and the allocation probability is reorganized through a random strategy to generate an initial plan;

[0117] In this step, the reinforcement learning agent first considers the demand attributes of the task node (such as task priority, time window, resource demand, etc.) and the state attributes of the force node (such as resource capacity, geographical location, load condition, etc.) based on the current state space, and dynamically constructs the mapping relationship between the task node and the force node. This mapping relationship is achieved by calculating the allocation probability matrix, where each probability value represents the matching possibility between a task node and a force node. In order to ensure that the allocation scheme is exploratory and diverse, the agent uses a random strategy to rearrange and sample the allocation probability matrix, and introduces a greedy strategy to balance the exploration, thereby generating an initial force allocation scheme.

[0118] Specifically, the calculation of the allocation probability requires a comprehensive evaluation of the path distance between the task and the node, the pass probability, and the resource matching degree of the node, while dynamically adjusting the allocation probability to adapt to changes in the battlefield environment. For example, when the load of a certain force node is too high, its allocation probability will automatically decrease, thereby avoiding the concentration of tasks on a few nodes and causing system performance to deteriorate. The introduction of random strategies further enhances the agent's ability to explore new allocation possibilities and prevents the emergence of local optimal solutions.

[0119] It realizes the reasonable mapping between tasks and force nodes, while retaining the flexibility and adaptability of the allocation scheme, and can dynamically adjust the allocation strategy in a complex battlefield environment to meet the scheduling needs of multiple tasks and multiple nodes. The principle is to capture the dynamic relationship between tasks and force nodes through state space modeling, and perform feasibility evaluation and random adjustment through the allocation probability matrix, so that the intelligent agent can take into account the needs of global optimization and local adjustment when generating allocation schemes.

[0120] The initial plan is checked for constraints in the action space to ensure that the initial plan meets the realistic constraints of task execution and resource allocation, so as to avoid task failure or overload of force nodes due to unreasonable allocation. It is determined whether the constraints are met. If not, the initial plan is regenerated in the state space and the reward value of this plan is set to 0. If it is met, the reward value of the initial plan is calculated according to the reward function.

[0121] If the initial solution fails to meet any constraints, the agent will set the reward value of the solution to 0 and fall back to the state space to regenerate the solution, while retaining the known local optimization information in the current allocation to guide the next round of solution generation. Through this dynamic adjustment mechanism, the agent can gradually screen out infeasible allocation actions and optimize the global feasibility and execution efficiency of the solution;

[0122] When the initial solution meets all constraints, the agent will calculate the reward value of the solution according to the set reward function. The reward function combines multi-dimensional indicators such as task priority, resource utilization, load balance and path cost to evaluate the quality of the current allocation plan. The completion of high-priority tasks, efficient use of resources, reduction of path costs and improvement of load balance will significantly increase the reward value;

[0123] Specifically, through constraint testing and reward function evaluation, the agent can efficiently screen out feasible solutions that meet the actual needs of the battlefield, and gradually optimize the force allocation strategy, ultimately achieving coordinated scheduling of multiple tasks and rational use of force nodes. The combination of action space and constraint conditions enables the preliminary evaluation and screening of allocation plans, while the design of the reward function guides the agent toward the global optimization goal. This mechanism ensures the effectiveness of the training direction and strategy optimization in the reinforcement learning process;

[0124] Determine whether the reward value is less than the set threshold for three consecutive rounds. If so, output the initial plan corresponding to the maximum reward value in the three rounds as the force allocation plan. If not, retain the initial plan in the state space and optimize the loss function of the random strategy according to the reward value to regenerate the initial plan.

[0125] Specifically, after each round of plan generation and evaluation, the agent will record the reward value of the current initial plan and compare it with the reward values ​​of the previous two rounds to determine whether the reward values ​​of three consecutive rounds are all less than the set threshold. The set threshold of the reward value is a dynamic adjustment parameter used to measure the overall quality of the allocation plan.

[0126] The new initial plan is more inclined to the plan with better performance in historical data in terms of allocation probability. At the same time, it retains the ability to explore new allocation possibilities through randomness, so that it is more likely to generate high-quality allocation plans in the next round of iterations. The three consecutive rounds of reward value judgment can not only measure the overall trend of the allocation plan, but also ensure the practicality of the output plan through the local optimal plan selection mechanism. Through the dynamic adjustment of random strategies, the reinforcement learning agent gradually approaches the global optimal solution. The core of this mechanism is to balance exploration and utilization, not only using existing high-quality allocation strategies, but also retaining flexibility and innovation space for plan optimization through randomness, so as to continuously improve the adaptability and execution efficiency of force scheduling in a dynamic battlefield environment.

[0127] The specific steps of S4 are as follows:

[0128] S4.1: Initialize the path map, set the force node in the force allocation plan as the starting point, and set the target node as the end point to mark in the path map;

[0129] S4.2: Set the feasible paths between the force node and the target node as the edges of the path graph, and the weight of each edge represents the comprehensive cost of the path, wherein the comprehensive cost includes the path distance and the path resource requirement;

[0130] S4.3: Select the edge with the smallest cumulative weight as the optimal path for the force node to reach the target node, traverse each force node in the force allocation plan, and output the optimal path plan.

[0131] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A method for rapid reorganization of combat forces based on a system architecture model, characterized in that: The method comprises: Abstract the entities in the combat system into nodes, and build a force deployment model based on the interaction between entities; According to the random enemy situation detected by the node, the target node is updated in the force deployment model, and the task requirement sequence is generated according to the initial attributes of the target node; According to the task requirement sequence and force deployment model, a force allocation optimization model is established, and according to resource constraints and task priorities, a force allocation plan is generated through the force allocation optimization model; According to the force allocation plan, the optimal path for the forces from the force node to the target node is planned.

2. A method for rapid reorganization of combat forces based on a system architecture model according to claim 1, characterized in that: The constructing of the force deployment model comprises: Divide the entities in the combat system into reconnaissance nodes, command nodes and force nodes according to their functions, and add attribute values ​​to each node according to the entity function, wherein the attribute values ​​include location coordinates, task status, force status, corresponding delay and task priority; Establishing an interactive relationship according to the node type, wherein the interactive relationship includes information flow, instruction flow and resource flow; Dynamic attributes are added to nodes and interaction relationships to build a force deployment model in the form of a directed graph.

3. A method for rapid reorganization of combat forces based on a system architecture model according to claim 2, characterized in that: The additional dynamic attributes include: Add a blank task sequence, wherein the blank task sequence includes a task time window and a task timing dependency, the task time window is used to add time constraints to tasks and nodes, including the earliest task start time, the latest task completion time and the task duration, and the task timing dependency is used to establish a priority diagram between tasks according to the dependency relationship of multiple tasks; Establish a task conflict detection mechanism. When a reconnaissance node or a force node is called by multiple tasks at the same time, adjust the resource usage order through task priority and task time window; Through reconnaissance nodes, information on terrain changes is collected, including obstacle generation, path blockage, and changes in traversable width. Based on the battlefield terrain, a dynamic path feasibility map is established between nodes.

4. A method for rapid reorganization of combat forces based on a system architecture model according to claim 3, characterized in that: The generating of the task requirement sequence comprises: Acquiring initial attributes of the task from the target node through the reconnaissance node, wherein the initial attributes include target location, threat level and task type; The task priority is calculated according to the initial attribute, sorted in the task requirement sequence according to the task priority, and updated in the task requirement sequence according to descending order. The calculation formula of the task priority is: Among them, P represents the task priority, W represents the threat level of the target node, ω1 represents the weight of the threat level, and T s Indicates the earliest start time of the task, T e Indicates the latest completion time of the task, T c represents the current time, ω2 represents the weight of the time constraint, D represents the average distance between the force node and the target node, F represents the mean path accessibility between the force node and the target node, ω3 represents the weight of the path constraint, and R d It represents the difference between the target node resource demand and the current remaining resources, and ω4 represents the weight of resource matching.

5. A method for rapid reorganization of combat forces based on a system architecture model according to claim 4, characterized in that: The optimization objectives of the force allocation optimization model include task priority satisfaction, resource utilization and force node load balance.

6. A method for rapid reorganization of combat forces based on a system architecture model according to claim 4, characterized in that: The constraints of the force allocation optimization model include: The resource requirement of each task is less than or equal to the total resource capacity of the force node; The total amount of resources allocated to each force node is less than or equal to its available capacity; Resource allocation for each task is completed within its task time window; The resource transfer path between the force node and the target node meets the path feasibility requirements.

7. A method for rapid reorganization of combat forces based on a system architecture model according to claim 6, characterized in that: The generating of the force allocation plan by the force allocation optimization model includes: Input the task requirement sequence and the force deployment model to generate a task-node state graph, wherein the vertices of the task-node state graph represent task nodes and force nodes, the edges represent the resource allocation possibility between the task nodes and the force nodes, and the edge weight is determined by the task priority, path distance, pass probability and resource matching degree; A reinforcement learning agent is created, and the state space, action space and reward function of the reinforcement learning agent are initialized, the task-node state diagram is used as an input parameter of the reinforcement learning agent, and a force allocation plan is output.

8. A method for rapid reorganization of combat forces based on a system architecture model according to claim 7, characterized in that: The calculation formula of the reward function is: Among them, Z represents the reward value, i represents a single task in the task requirement sequence, N represents the total number of tasks in the task requirement sequence, M represents the total number of force nodes assigned to task i, and j represents a single force node assigned to task i. represents the weight of the priority satisfaction, represents the weight of resource utilization, Represents the weight of load balancing, P i represents the task priority of task i, η represents the gain coefficient of resource allocation to task priority, R i,j represents the resource demand satisfaction of force node j for task i, u j represents the amount of resources allocated to force node j, U j represents the total resource capacity of force node j, λ j It represents the load standard deviation of force node j, and U represents the average value of resources allocated to all nodes in the force allocation scheme.

9. A method for rapid reorganization of combat forces based on a system architecture model according to claim 8, characterized in that: The reinforcement learning agent comprises: According to the state space, the allocation probability between the task nodes and the force nodes is calculated, and the allocation probability is reorganized through a random strategy to generate an initial plan; Perform constraint check on the initial plan in the action space to determine whether the constraint is satisfied. If not, regenerate the initial plan in the state space and set the reward value of this plan to 0. If satisfied, calculate the reward value of the initial plan according to the reward function. Determine whether the reward value is less than the set threshold for three consecutive rounds. If so, output the initial plan corresponding to the maximum reward value in the three rounds as the troop allocation plan. If not, retain the initial plan in the state space and optimize the loss function of the random strategy according to the reward value to regenerate the initial plan.

10. The method for rapid reorganization of combat forces based on a system architecture model according to claim 1, characterized in that: The optimal path includes: Initialize the path map, set the force node in the force allocation plan as the starting point, and set the target node as the end point to be marked in the path map; The feasible paths between the force node and the target node are set as the edges of the path graph, and the weight of each edge represents the comprehensive cost of the path, wherein the comprehensive cost includes the path distance and the path resource requirement; Select the edge with the smallest cumulative weight as the optimal path for the force node to reach the target node, traverse each force node in the force allocation plan, and output the optimal path plan.

Images