Military training plan and resource linkage management method and system

Abstract

The invention belongs to the technical field of training management, and discloses a military training plan and resource linkage management method and system. The method comprises the following steps: collecting training resource data and n groups of training task data; establishing a training task network graph model based on the n groups of training task data; processing the training task network graph model by using a community discovery algorithm to obtain a partitioned network graph; constructing a training resource dynamic model based on the training resource data, and simulating the training resource dynamic model to obtain a training resource optimal configuration scheme; constructing a bipartite graph based on the partition network graph and the optimal configuration scheme of the training resources; and solving the bipartite graph by using a Hungary algorithm to obtain an optimal task-resource matching scheme, and sending the optimal task-resource matching scheme to the linkage management terminal for training resource allocation, effectively processing large-scale and high-complexity training plans and resource management problems, and significantly improving the planning and execution capabilities of training.

Description

Technical Field

[0001] The present invention relates to the technical field of training management. More specifically, the present invention relates to a method and system for the linkage management of military training plans and resources. Background Art

[0002] The patent application with the publication number CN115130893A discloses a method, device and system for the linkage management of military training plans and resources, including: determining a target training plan; analyzing the target training plan and generating a target resource requirement instruction according to the analysis result; retrieving corresponding training resources according to the target resource requirement instruction; generating reservation information according to the corresponding training resources and sending the reservation information to a logistics processor; wherein the reservation information includes at least one of venue reservation, equipment scheduling, teaching material preparation and personnel assignment; instructing the logistics processor to generate a logistics support instruction according to the reservation information, and the logistics support instruction is used to instruct logistics personnel to transport the corresponding training resources to a designated location according to the reservation information; which can solve the problem in the prior art that the linkage between military training plans and the resource allocation for military training plans cannot be realized.

[0003] However, in actual training, due to the lack of a systematic analysis of the complex relationships between training tasks, the training plan is often arranged unreasonably and the task connection is not smooth; if the logical relationships between different subjects are not fully considered, the training progress will be chaotic and the effect will be greatly reduced; in terms of resource allocation, traditional methods often adopt static and linear allocation models, which cannot adapt to the dynamic changes of training requirements; the resource allocation is not flexible enough, resulting in an oversupply of resources for some training subjects while other subjects are facing a shortage of resources; in addition, the matching process of training tasks and resources lacks global optimization, and the situation of low resource utilization efficiency often occurs; in the face of emergencies or urgent tasks, the existing methods are difficult to quickly adjust the training plan and resource allocation; resulting in waste of training resources and low training efficiency.

[0004] In view of this, the present invention proposes a method and system for the linkage management of military training plans and resources to solve the above problems. Summary of the Invention

[0005] In order to overcome the above defects of the prior art and to achieve the above object, the present invention provides the following technical solution: A method for the linkage management of military training plans and resources, including: Step 1, collecting training resource data and n groups of training task data;

[0006] Step 2, establishing a training task network graph model based on the n groups of training task data;

[0007] Step 3, processing the training task network graph model by using an improved community discovery algorithm to obtain a partitioned network graph;

[0008] Step 4: Build a dynamic training resource model based on the training resource data, simulate the dynamic training resource model, and obtain an optimal training resource allocation plan;

[0009] Step 5: Based on the partition network diagram and the optimal training resource allocation plan, build a bipartite graph; use the Hungarian algorithm to solve the bipartite graph to obtain an optimal task-resource matching plan, and send the optimal task-resource matching plan to the linkage management terminal.

[0010] Further, the training resource data is information on various training resources; including training venue information, training equipment information, training personnel information, training time information, and training funds information;

[0011] The n groups of training task data include the training project name, training objective, training duration, training priority, training dependency, training resources required, and participants for n training tasks.

[0012] Further, the method for establishing the training task network diagram model includes:

[0013] Initialize a network diagram, define each training task as a node in the network diagram, assign a unique identifier to each node, and include the training task data corresponding to the training task in the node;

[0014] Analyze the relationship types between training tasks based on the training dependencies. The relationship types include sequential relationship, parallel relationship, mutual exclusion relationship, and dependency relationship; according to the relationship types between training tasks, define different types of edges in the network diagram. Different types of edges include directed edges and undirected edges; the relationship types between training tasks connected by directed edges are sequential relationship or dependency relationship; the relationship types between training tasks connected by undirected edges are parallel relationship or mutual exclusion relationship; and assign weights to different types of edges; obtain a preliminary training task network diagram model;

[0015] The formula for the weight is:

[0016]

[0017] Among them, \(W_{i,j}\) is the weight of the edge connecting node \(i\) and node \(j\), and \(\Delta t_{i,j}\) is the time interval between the training tasks corresponding to node \(i\) and node \(j\); \(t_{avg}\) is the average value of the time intervals between all training task pairs, \(t_{std}\) is the standard deviation of the time intervals between all training task pairs, and \(d_{i,j}\) is the degree of dependence between the training tasks corresponding to node \(i\) and node \(j\); \(d_{max}\) is the maximum possible value of the preset degree of dependence; \(A_i\) is the set of training resources required for the training task corresponding to node \(i\), and \(A_j\) is the set of training resources required for the training task corresponding to node \(j\); \(w1\), \(w2\), and \(w3\) are the weight coefficients of each weighted term, and the sum of the three is 1; the preliminary training task network graph model is optimized in isolation to obtain the training task network graph model.

[0018] Further, the method of performing isolation optimization includes:

[0019] Traverse all nodes in the preliminary training task network graph model, and calculate the degree of each node; mark the nodes with a degree of 0 or 1 as potential isolated nodes; define the other nodes in the preliminary training task network graph model as target positions; preset a connection threshold, and use the nodes corresponding to the target positions with a degree greater than the connection threshold as target nodes;

[0020] For each isolated node, use the A* search algorithm or the Dijkstra algorithm to calculate the shortest path to the nearest target node, push the isolated node along the shortest path, and gradually establish connections with the nodes on the isolated node; preset a similarity threshold, and for each pushed isolated node, calculate the similarity with the target node; if the similarity is greater than or equal to the similarity threshold, establish a new edge between the isolated node and the target node, and assign a weight to the new edge according to the previously defined weight calculation formula; preset a weight threshold, and remove the new edges with weights less than the weight threshold; recalculate the degrees of all nodes and update the connection information of the nodes; repeat until no new isolated nodes are marked to obtain the training task network graph model.

[0021] Further, the method for obtaining the partitioned network graph includes:

[0022] Define the policy space as each task can choose to join different communities; define the utility function \(U(I,C)\);

[0023] U(I, C) = a1 × Sl(I, C) + a2 × RC(I, C) + a3 × SV(I, C) - a4 × Cost(I, C); where Sl(I, C) is the average similarity between training task I and other training tasks in community C; RC(I, C) is the compatibility of resource sharing between training task I and other training tasks in community C; SV(I, C) is the strategic value of adding training task I to community C to the overall training goal; Cost(I, C) is the cost of adding training task I to community C; a1, a2, a3, and a4 are the weights of each item;

[0024] Use the clustering algorithm to generate the initial community partition. Under the initial community partition, for each training task I, calculate the value of the current utility function, denoted as the utility U_I of training task I; calculate the utility U' of adding it to each other community;

[0025] If there exists U' greater than U_I, then move training task I to the community with the maximum utility; repeat until no training task can obtain higher utility by changing the community; at this time, each community contains several training tasks;

[0026] Create a new graph structure, retain all the nodes of the original training task network graph model, add the attributes of the community to each node based on the community to which each training task belongs, and recalculate the weights of the edges between the nodes. On the basis of the original weights, synchronously increase the weights of the edges between the nodes within the same community by a fixed random number α1 within the interval (0, 1); use the force-directed algorithm to optimize the node layout to obtain the partitioned network graph.

[0027] Furthermore, the acquisition method of the optimal training resource allocation scheme includes:

[0028] Represent each type of training resource as a spin variable, and the spin variable takes +1 or -1. +1 means the corresponding training resource is used, and -1 means the corresponding training resource is not used;

[0029] Define the total energy H of the training resource dynamic model;

[0030] where w_u is the weight of training resource u, c_u is the usage cost of training resource u, cmax is the highest usage cost among all training resources; σ_u is the spin variable corresponding to training resource u; J_u,v is the coupling strength between training resource u and training resource v; σ_v is the spin variable of training resource v;

[0031] Among them, \(s_{u,v}\) is the spatial correlation between training resource \(u\) and training resource \(v\); \(s_{max}\) is the maximum value of the spatial correlation between training resources, \(t_{u,v}\) is the temporal correlation between training resource \(u\) and training resource \(v\); \(t_{max}\) is the maximum value of the temporal correlation between training resources, \(e_{u,v}\) is the efficiency correlation between training resource \(u\) and training resource \(v\); \(\beta_1\), \(\beta_2\) and \(\beta_3\) are the weights of each coupling strength item, and the sum of the three is 1;

[0032] Randomly assign +1 or -1 to each spin variable, set the simulation temperature \(T\) and the temperature reduction coefficient \(r\); for each iteration, randomly select a spin variable and calculate the difference \(\Delta H\) in the total energy \(H\) before and after flipping this spin variable; flipping means +1 becomes -1 or -1 becomes +1;

[0033] If \(\Delta H\) is less than 0, accept the flip; if \(\Delta H\) is greater than 0, accept the flip with probability where \(k_1\) is a preset constant; repeat until the change in the total energy is less than the preset energy threshold; during this process, reduce the simulation temperature, that is, multiply the simulation temperature by the temperature reduction coefficient for each iteration as the simulation temperature for the next iteration; record the values of the spin variables when the total energy is the lowest during the simulation process to form the optimal resource allocation plan.

[0034] Furthermore, the method for constructing the bipartite graph includes:

[0035] Create two vertex sets \(V1\) and \(V2\): \(V1\) represents the set composed of training tasks, \(V2\) represents the set composed of training resources, and initialize an empty edge set \(E\); traverse each node in the partition network diagram and map it to a vertex in \(V1\); traverse each resource in the optimal resource allocation plan and map it to a vertex in \(V2\); for each vertex in \(V1\), analyze the resource requirements of this training task and classify the requirements into essential resources and optional resources;

[0036] For each vertex in \(V1\) and each vertex in \(V2\), calculate the mutual compatibility score; for vertex \(v_j\) in \(V1\) and vertex \(v_i\) in \(V2\), if vertex \(v_i\) is an essential resource of vertex \(v_j\), or vertex \(v_i\) is an optional resource of vertex \(v_j\) and the compatibility score is greater than the preset compatibility threshold, then connect an edge between vertex \(v_i\) and vertex \(v_j\) and add it to the edge set \(E\); use the compatibility score as the weight of the edge in the edge set \(E\); obtain the preliminary bipartite graph; use the graph optimization algorithm to optimize the preliminary bipartite graph to obtain the bipartite graph.

[0037] Furthermore, the method for obtaining the optimal task-resource matching plan includes:

[0038] Structurally store the constructed bipartite graph to generate a task-resource association matrix F, find the edge corresponding to the maximum weight in the task-resource association matrix and its weight value mw; in F, find all the edges with weights equal to mw to form an equal-weight line. Starting from the equal-weight line, select an uncovered point on the equal-weight line, and find an augmenting path starting from this point. If an augmenting path is found, augment the non-adjacent set M according to the indication of the augmenting path; if no augmenting path is found, modify the label and continue to find the augmenting path, repeating until all points on the equal-weight line are covered.

[0039] For each edge in M, set the corresponding element in F to 0, indicating that this edge has been occupied; repeat until the weights of all edges are 0; each time an edge corresponding to the maximum weight is found, reconstruct the equal-weight line and continue the next round until all elements in F are 0. Combining the Ms found in each round, the final optimal task-resource matching scheme can be obtained.

[0040] A military training plan and resource linkage management system, which is used to implement the described military training plan and resource linkage management method, includes: a data acquisition module, which is used to acquire training resource data and n groups of training task data;

[0041] A preliminary model construction module, which establishes a training task network graph model based on n groups of training task data;

[0042] A partitioning module, which is used to process the training task network graph model using an improved community discovery algorithm to obtain a partitioned network graph;

[0043] A dynamic simulation module, which constructs a training resource dynamic model based on the training resource data, simulates the training resource dynamic model, and obtains an optimal training resource allocation scheme;

[0044] A scheme fitting module, which constructs a bipartite graph based on the partitioned network graph and the optimal training resource allocation scheme; uses the Hungarian algorithm to solve the bipartite graph to obtain an optimal task-resource matching scheme, and sends the optimal task-resource matching scheme to the linkage management terminal; each module is connected by wired and / or wireless means.

[0045] The technical effects and advantages of the military training plan and resource linkage management method and system of the present invention:

[0046] The present invention realizes the dynamic optimization and precise matching of training plans and resource allocation, greatly improving the overall efficiency and quality of training. It can flexibly respond to complex and changeable training environments, quickly adjust training plans and resource configurations, enhancing the adaptability and flexibility of training. By systematically analyzing and optimizing the relationships between training tasks, the coherence and collaborative effects of training are significantly improved. The resource utilization rate is significantly increased, avoiding resource waste and achieving the maximization of the utilization of limited resources. At the same time, it also provides scientific and intuitive data support for command and decision-making, improving the intelligence and refinement level of training management. In addition, it can effectively handle large-scale and high-complexity training plan and resource management problems, significantly enhancing the training planning and execution capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 It is a schematic diagram of a military training plan and resource linkage management method of the present invention;

[0048] Figure 2 It is a schematic diagram of a military training plan and resource linkage management system of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0049] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0050] Embodiment 1

[0051] Please refer to Figure 1 As shown, a military training plan and resource linkage management method in this embodiment includes:

[0052] Step 1: Collect training resource data and n groups of training task data, where n is an integer greater than 1;

[0053] Step 2: Establish a training task network graph model based on the n groups of training task data;

[0054] Step 3: Use an improved community discovery algorithm to process the training task network graph model to obtain a partitioned network graph;

[0055] Step 4: Build a training resource dynamic model based on the training resource data, simulate the training resource dynamic model, and obtain an optimal training resource allocation plan;

[0056] Step 5: Based on the partitioned network diagram and the optimal training resource allocation plan, construct a bipartite graph; use the Hungarian algorithm to solve the bipartite graph to obtain the optimal task-resource matching plan, and send the optimal task-resource matching plan to the linkage management terminal for training resource allocation.

[0057] The training resource data is information on various training resources, including training venue information, training equipment information, training personnel information, training time information, and training funds information.

[0058] The training venue information includes the type, area, and occupancy capacity of the training venue; the training equipment information includes the quantity of various weapons and equipment, simulators, and training equipment; the training personnel information, such as the number of instructors and trainees; the training time information includes the available training time periods and the duration; the training funds information includes the budgets and costs of various training projects.

[0059] The n groups of training task data include the training project names, training objectives, training durations, training priorities, training dependencies, training required resources, and trainees corresponding to n training tasks.

[0060] The training project names, such as physical training, shooting training, tactical training, etc.; the training objective is the specific objective and requirement of each training project; the training duration is the time required for each training task; the training priority is the ranking of the importance of different training tasks; the training dependency is the sequence or dependency relationship between training tasks; the training required resources are the resource requirements such as venue, equipment, and personnel for completing each training task; the trainees are the training objects and numbers for each training task. Through the analysis and processing of these data, efficient linkage management of the training plan and resources is achieved.

[0061] The methods for establishing the training task network diagram model include:

[0062] Initialize a network diagram, define each training task as a node in the network diagram, assign a unique identifier to each node, such as a task number or name, and include the training task data corresponding to the training task in the node.

[0063] Analyze the relationship types between training tasks based on the training dependencies. The relationship types include sequential relationship, parallel relationship, mutually exclusive relationship, and dependency relationship. It should be noted that the sequential relationship is interpreted as task A must be completed before task B, the parallel relationship is interpreted as task C and task D can be carried out simultaneously, the mutually exclusive relationship is interpreted as task E and task F cannot be carried out simultaneously, and the dependency relationship is interpreted as the start of task G depends on the partial or complete completion of task H.

[0064] According to the relationship types between training tasks, different types of edges are defined in the network diagram, including directed edges and undirected edges; the relationship types between training tasks connected by directed edges are sequential relationship or dependency relationship; the relationship types between training tasks connected by undirected edges are parallel relationship or mutual exclusion relationship; and weights are assigned to different types of edges; thus obtaining a preliminary training task network diagram model.

[0065] The formula for the weight is:

[0066]

[0067] Where \(W_{i,j}\) is the weight of the edge connecting nodes \(i\) and \(j\), \(\Delta t_{i,j}\) is the time interval between the training tasks corresponding to nodes \(i\) and \(j\); \(t_{avg}\) is the average value of the time intervals between all training task pairs, \(t_{std}\) is the standard deviation of the time intervals between all training task pairs, \(d_{i,j}\) is the degree of dependence between the training tasks corresponding to nodes \(i\) and \(j\), which is a subjective score (such as from 1 to 10); \(d_{max}\) is the maximum possible value of the preset degree of dependence; \(A_i\) is the set of training resources required for the training task corresponding to node \(i\), \(A_j\) is the set of training resources required for the training task corresponding to node \(j\); \(w1\), \(w2\), and \(w3\) are the weight coefficients of each weighted term, and the sum of the three is 1.

[0068] Isolate and optimize the preliminary training task network diagram model to obtain the training task network diagram model; specifically, traverse all nodes in the preliminary training task network diagram model, and calculate the degree (the number of edges connected to it) of each node; mark the nodes with a degree of 0 or 1 as potential isolated nodes; define the other nodes in the preliminary training task network diagram model as target positions; preset a connection threshold, and take the nodes corresponding to the target positions with a degree greater than the connection threshold as target nodes.

[0069] For each isolated node, use the A* search algorithm or Dijkstra algorithm to calculate the shortest path to the nearest target node, and push the isolated node along the shortest path to gradually establish connections with the nodes on the isolated node; preset a similarity threshold, and calculate the similarity with the target node for each pushed isolated node; specifically, by using natural language processing technology, convert the text information such as the name and target description of the training task into vector representations, and then calculate the cosine similarity between the two vectors as the similarity; if the similarity is greater than or equal to the similarity threshold, establish a new edge between the isolated node and the target node, and assign a weight to the new edge according to the previously defined weight calculation formula.

[0070] Preset a weight threshold, and remove new edges with weights less than the weight threshold; recalculate the degrees of all nodes and update the connection information of the nodes; repeat until no new isolated nodes are marked; by effectively reducing the isolated nodes in the network, improve the connectivity between training tasks while maintaining the original task relationship structure, so as to obtain a more optimized and balanced training task network graph model.

[0071] Use a graph library (such as NetworkX, Graphviz, etc.) to visualize the training task network graph model. Nodes can be represented by different shapes or colors for different types or priorities of tasks; edges are represented by different line types or colors for different types of relationships.

[0072] Convert n groups of training tasks into a structured and visual network graph model, providing a basis for subsequent community discovery algorithms and resource allocation; the model can clearly display the complex relationships between training tasks, helping to optimize the overall training plan and resource allocation.

[0073] The ways to obtain the partitioned network graph include:

[0074] Convert the training task network graph model into a format for game theory analysis, and ensure the integrity of the information of nodes (training tasks) and edges (task relationships); define the strategy space as each task can choose to join different communities; define the utility function U(I,C), which reflects the benefits of a task joining a certain community.

[0075] U(I,C) = a1×Sl(I,C) + a2×RC(I,C) + a3×SV(I,C) - a4×Cost(I,C); where Sl(I,C) is the average similarity between training task I and other training tasks in community C (calculated using, for example, Euclidean distance or cosine similarity); RC(I,C) is the compatibility of resource sharing between training task I and other training tasks in community C; SV(I,C) is the strategic value of adding training task I to community C to the overall training goal; Cost(I,C) is the cost of adding training task I to community C (such as resource competition); a1, a2, a3, and a4 are the weights of each item.

[0076] The calculation method of the compatibility of resource sharing is as follows:

[0077] The defined compatibility scoring metrics include site compatibility score, equipment compatibility score, personnel compatibility score, and time compatibility score; normalize the scoring metrics to the interval [0, 1], and perform weighted summation to obtain the compatibility of resource sharing; the site compatibility score is the matching degree × (1 - time overlap degree); where the matching degree is the matching probability between the site type required by the training task and the main site type used by the community, which is a subjective score (such as 1 to 10 points).

[0078] The device compatibility score is (the number of devices available in the community that match the devices required for the training task) divided by (the total number of devices required for the training task); the personnel compatibility score is the number of suitable personnel available in the community divided by the number of personnel required for the training task; the time compatibility score is the overlap between the available time period for the training task and the main activity time of the community (occupying more than half) divided by the total time required for the training task.

[0079] The strategic value is obtained by the weighted sum of the task priority, the relevance of the task to the community theme, the impact of the task on other tasks, and the contribution of the community to the overall training goal; the task priority is obtained from the training task data, and the relevance of the task to the community theme is obtained through expert evaluation; the impact of the task on other tasks is obtained based on the task dependency analysis. Specifically, for each training task, calculate the number of training tasks directly affected by it as the impact; the contribution of the community to the overall training goal is reflected by the resource utilization efficiency; the resource utilization efficiency is the actual resources used divided by the resources allocated to the community.

[0080] Use a simple clustering algorithm (such as K-means) to generate an initial community partition; ensure that the initial partition meets the basic training logic and resource constraints; under the initial community partition, for each training task I, calculate the value of the current utility function, denoted as the utility U_I of training task I; calculate the utility U' of adding to each other community; if there exists U' greater than U_I, then move training task I to the community with the maximum utility, and update the attributes of the affected community and the utilities of other training tasks; repeat until no training task can obtain higher utility by changing the community; at this time, each community contains several training tasks.

[0081] Create a new graph structure, retain all nodes of the original training task network graph model, add the attributes of the community to each node based on the community to which each training task belongs, and recalculate the weights of the edges between nodes. On the basis of the original weights, synchronously increase the weights of the edges between nodes within the same community by a fixed random number α1 within the interval (0, 1); use a force-directed algorithm (such as the Fruchterman-Reingold algorithm) to optimize the node layout, and apply a strong attraction between nodes within the community; apply a weak repulsion between nodes in different communities, and at this time, obtain a partitioned network graph.

[0082] Transform the training task network model into a game theory problem and achieve the optimal clustering of tasks by finding the equilibrium point; considering the individual rationality and overall utility of each training task, it can find a balanced community structure in a complex military training environment, providing a theoretical basis and practical guidance for formulating efficient training plans and resource allocation strategies.

[0083] The ways to obtain the optimal training resource allocation plan include:

[0084] Each training resource (such as venue, equipment, personnel, etc.) is represented as a spin variable, which takes the value of +1 or -1. +1 indicates that the corresponding training resource is being used, and -1 indicates that the corresponding training resource is not being used; define the total energy H of the training resource dynamic model;

[0085] where w_u is the weight of training resource u (which can be determined according to factors such as resource type, scarcity, etc.), c_u is the usage cost or value of training resource u, and cmax is the highest usage cost or value among all training resources; σ_u is the spin variable corresponding to training resource u (+1 or -1); J_u,v is the coupling strength (interaction strength) between training resource u and training resource v; σ_v is the spin variable of training resource v.

[0086] where s_u,v is the spatial correlation between training resource u and training resource v (the proximity of the quantified venue); smax is the maximum value of the spatial correlation between training resources (predetermined), t_u,v is the temporal correlation between training resource u and training resource v (the overlap degree of the quantified usage time); tmax is the maximum value of the temporal correlation between training resources (predetermined), e_u,v is the efficacy correlation between training resource u and training resource v. Specifically, determine the key indicators for measuring the resource synergy effect. For example, training quality improvement, time efficiency, cost savings, training goal achievement, and resource utilization rate; establish a scoring system for each indicator, such as a 1 - 10 point scale, and score each pair of resources on each indicator to obtain the efficacy correlation; β1, β2, and β3 are the weights of each item of the coupling strength, and the sum of the three is 1.

[0087] Randomly assign +1 or -1 to each spin variable, set the simulation temperature T and the cooling coefficient r (0 < r < 1); for each iteration, randomly select a spin variable and calculate the difference ΔH in the total energy H before and after flipping this spin variable; flipping means +1 becomes -1 or -1 becomes +1; if ΔH is less than 0, accept the flip, if ΔH is greater than 0, with a probability Accept the flip, where k1 is a preset constant; repeat until the change in the total energy is less than a preset energy threshold; during this process, reduce the simulation temperature, that is, multiply the simulation temperature by a cooling coefficient for each iteration as the simulation temperature for the next iteration; record the values of the spin variables when the total energy is the lowest during the simulation process to form an optimal resource allocation plan; specifically, mark the training resources with spin variable +1 as used and the training resources with spin variable -1 as unused, and generate a detailed resource allocation table including information such as the usage status, allocation time, and location of each resource; based on considering the complex interactions between resources, obtain a globally optimal training resource allocation plan.

[0088] Create two vertex sets V1 and V2: V1 represents the set composed of training tasks, and V2 represents the set composed of training resources. Initialize an empty edge set E; traverse each node (training task) in the partition network graph and map it to a vertex in V1; traverse each resource in the optimal resource allocation plan and map it to a vertex in V2; for each vertex in V1, analyze the resource requirements of this training task (extracted from the original training task data) and classify the requirements into essential resources and optional resources.

[0089] For each vertex in V1 and each vertex in V2, calculate the mutual compatibility score. The evaluation indicators of the compatibility score include resource type matching degree, whether the resource capacity meets the requirements (0 or 1), whether the resource available time overlaps with the task time (0 or 1), and whether the resource location is suitable for task execution (0 or 1); perform a weighted sum of the evaluation indicators to obtain the compatibility score.

[0090] For vertex vj in V1 and vertex vi in V2, if vertex vi is an essential resource of vertex vj, or vertex vi is an optional resource of vertex vj and the compatibility score is greater than a preset compatibility threshold, then connect an edge between vertex vi and vertex vj and add it to the edge set E; use the compatibility score as the weight of the edge in the edge set E; obtain a preliminary bipartite graph.

[0091] Optimize the preliminary bipartite graph using graph optimization algorithms (such as the minimum spanning tree algorithm) to remove redundant edges; ensure that each training task vertex is at least connected to its required resources to obtain a bipartite graph; store the constructed bipartite graph structurally, including vertex sets V1, V2, edge set E and its weights, and generate a task-resource association matrix F for subsequent processing by the Hungarian algorithm; find the edge corresponding to the maximum weight in the task-resource association matrix and its weight value mw; in F, find all edges with weights equal to mw to form an equal-weight line, starting from the equal-weight line, select an uncovered point on the equal-weight line, and find an augmenting path starting from this point. If an augmenting path is found, augment the non-adjacent set M (the set of edges found in the bipartite graph, where any two edges are non-adjacent) according to the indication of the augmenting path; that is, adjust the current M according to the found augmenting path; through this adjustment, a larger matching is obtained. If no augmenting path is found, modify the labels and continue to search for an augmenting path, repeating until all points on the equal-weight line are covered; an augmenting path refers to a path in the bipartite graph that starts from an unmatched vertex, passes through alternating unmatched edges and matched edges, and finally reaches another unmatched vertex.

[0092] For each edge in M, set the corresponding element in F to 0, indicating that this edge has been occupied; repeat until the weights of all edges are 0; each time an edge corresponding to the maximum weight is found, reconstruct the equal-weight line and continue the next round until all elements in F are 0. Combining the Ms found in each round, the final optimal task-resource matching scheme is obtained.

[0093] This M is obtained based on the current remaining unmatched tasks and resources. The Ms found in each round are not independent of each other, but are based on the results of the previous round; it is necessary to accumulate the results of each round and merge the Ms found in each round into a total matching result; specifically, if the newly found matching involves unmatched tasks and resources, directly add it to the total matching; if the newly found matching involves already matched tasks or resources, it is necessary to compare the weights of the new and old matches and retain the one with the larger weight; conflicts may occur during the merging process. For example, a resource is assigned to different tasks in different rounds; at this time, it is necessary to determine the final allocation according to the weight size. After obtaining the preliminary total matching, a global optimization can be performed to try to further improve the overall matching weight sum through local adjustments; the final obtained matching scheme includes the complete correspondence between each training task and its corresponding training resource, as well as the weight of each match.

[0094] The Hungarian algorithm is used to find the globally optimal weighted matching in the bipartite graph, thereby obtaining the optimal task-resource matching scheme; taking into account various factors, it can maximize the resource utilization efficiency and meet the training objectives.

[0095] In this embodiment, the dynamic optimization and precise matching of training plans and resource allocation are realized, significantly improving the overall efficiency and quality of training. It can flexibly respond to complex and changeable training environments, quickly adjust training plans and resource configurations, enhancing the adaptability and flexibility of training. Through systematic analysis and optimization of training task relationships, the coherence and collaborative effects of training are significantly improved. The resource utilization rate is significantly increased, avoiding resource waste and achieving the maximization of the utilization of limited resources. At the same time, it also provides scientific and intuitive data support for command and decision-making, improving the intelligence and refinement level of training management. In addition, it can effectively handle large-scale and high-complexity training plan and resource management problems, significantly enhancing the training planning and execution capabilities.

[0096] Embodiment 2

[0097] Please refer to Figure 2 As shown, for the parts not described in detail in this embodiment, refer to the description content of Embodiment 1. A military training plan and resource linkage management system is provided, including:

[0098] A data acquisition module for collecting training resource data and n groups of training task data;

[0099] A preliminary model construction module for establishing a training task network diagram model based on n groups of training task data;

[0100] A partitioning module for using an improved community discovery algorithm to process the training task network diagram model to obtain a partitioned network diagram;

[0101] A dynamic simulation module for constructing a training resource dynamic model based on the training resource data, simulating the training resource dynamic model, and obtaining an optimal training resource allocation plan;

[0102] A scheme fitting module for constructing a bipartite graph based on the partitioned network diagram and the optimal training resource allocation plan; using the Hungarian algorithm to solve the bipartite graph to obtain an optimal task-resource matching plan, and sending the optimal task-resource matching plan to the linkage management terminal; each module is connected by wired and / or wireless means to realize data transmission between modules.

[0103] Embodiment 3

[0104] This embodiment publicly provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it realizes the running mode of the above-provided military training plan and resource linkage management method.

[0105] Since the electronic device introduced in this embodiment is the electronic device adopted for implementing a military training plan and resource linkage management method in the embodiments of the present application, based on the military training plan and resource linkage management method introduced in the embodiments of the present application, those skilled in the art can understand the specific implementation manners and various variations of the electronic device in this embodiment. Therefore, the specific implementation of how this electronic device implements the method in the embodiments of the present application will not be introduced in detail here. As long as those skilled in the art implement the electronic device adopted for the military training plan and resource linkage management method in the embodiments of the present application, it falls within the scope of protection of the present application.

[0106] The above formulas are all dimensionless and take their numerical values for calculation. The formula is obtained by collecting a large amount of data and performing software simulation to obtain a formula that is closest to the actual situation. The preset parameters and threshold selection in the formula are set by those skilled in the art according to the actual situation.

[0107] The above are only the preferred implementation manners of the present invention. The protection scope of the present invention is not limited to the above embodiments. All technical solutions falling within the idea of the present invention belong to the protection scope of the present invention. It should be noted that for ordinary technical users in the technical field, several improvements and refinements made without departing from the principle of the present invention should also be regarded as within the protection scope of the present invention.

Claims

1. A method for the linkage management of military training plans and resources, characterized in that Including: Step 1: Collect training resource data and n groups of training task data; Step 2: Establish a training task network graph model based on the n groups of training task data; Step 3: Process the training task network graph model using an improved community discovery algorithm to obtain a partitioned network graph; Step 4: Construct a training resource dynamic model based on the training resource data, simulate the training resource dynamic model, and obtain an optimal training resource allocation plan; Step 5: Based on the partitioned network graph and the optimal training resource allocation plan, construct a bipartite graph; solve the bipartite graph using the Hungarian algorithm to obtain an optimal task-resource matching plan, and send the optimal task-resource matching plan to the linkage management terminal.

2. A method for the linkage management of military training plans and resources according to claim 1, characterized in that The training resource data is information on various training resources; Including training venue information, training equipment information, training personnel information, training time information, and training funds information; The n groups of training task data include the training project name, training objective, training duration, training priority, training dependency, training resources required, and participants for each of the n training tasks.

3. A method for linked management of military training plans and resources according to claim 2, characterized in that, The method for establishing the training task network graph model includes: Initialize a network graph, define each training task as a node in the network graph, assign a unique identifier to each node, and include the training task data corresponding to the training task in the node; Analyze the relationship types between training tasks based on the training dependencies. The relationship types include sequential relationship, parallel relationship, mutual exclusion relationship, and dependency relationship; according to the relationship types between training tasks, define different types of edges in the network graph. Different types of edges include directed edges and undirected edges; the relationship type between training tasks connected by directed edges is a sequential relationship or a dependency relationship; the relationship type between training tasks connected by undirected edges is a parallel relationship or a mutual exclusion relationship; and assign weights to different types of edges; obtain a preliminary training task network graph model; The formula for the weight is: Where \(W_{i,j}\) is the weight of the edge connecting nodes \(i\) and \(j\), \(\Delta t_{i,j}\) is the time interval between the training tasks corresponding to nodes \(i\) and \(j\); \(t_{avg}\) is the average value of the time intervals between all pairs of training tasks, \(t_{std}\) is the standard deviation of the time intervals between all pairs of training tasks, \(d_{i,j}\) is the degree of dependency between the training tasks corresponding to nodes \(i\) and \(j\); \(d_{max}\) is the maximum possible value of the preset degree of dependency; \(A_i\) is the set of training resources required for the training task corresponding to node \(i\), \(A_j\) is the set of training resources required for the training task corresponding to node \(j\); \(w1\), \(w2\), and \(w3\) are the weight coefficients of each weighted term, and the sum of the three is 1; perform isolated optimization on the preliminary training task network graph model to obtain the training task network graph model.

4. A method for linked management of military training plans and resources according to claim 3, characterized in that, The method for performing isolated optimization includes: Traverse all nodes in the preliminary training task network graph model, calculate the degree of each node; mark the nodes with a degree of 0 or 1 as potential isolated nodes; define the other nodes in the preliminary training task network graph model as target positions; preset a connection threshold, and use the nodes corresponding to the target positions with a degree greater than the connection threshold as target nodes; For each isolated node, use the A* search algorithm or Dijkstra's algorithm to calculate the shortest path to the nearest target node, push the isolated node along the shortest path, and gradually establish connections with the nodes on the isolated node; preset a similarity threshold, and for each pushed isolated node, calculate the similarity with the target node; if the similarity is greater than or equal to the similarity threshold, establish a new edge between the isolated node and the target node, and assign a weight to the new edge according to the previously defined weight calculation formula; preset a weight threshold, and remove the new edges with weights less than the weight threshold; recalculate the degrees of all nodes and update the connection information of the nodes; repeat until no new isolated nodes are marked to obtain the training task network graph model.

5. A method for linked management of military training plans and resources according to claim 4, characterized in that, The acquisition method of the partitioned network graph includes: Define the policy space such that each task can choose to join different communities; define the utility function U(I, C); U(I, C) = a1×Sl(I, C) + a2×RC(I, C) + a3×SV(I, C) - a4×Cost(I, C); where Sl(I, C) is the average similarity between training task I and other training tasks in community C; RC(I, C) is the compatibility of resource sharing between training task I and other training tasks in community C; SV(I, C) is the strategic value of adding training task I to community C to the overall training goal; Cost(I, C) is the cost of adding training task I to community C; a1, a2, a3, and a4 are the weights of each item; Use a clustering algorithm to generate an initial community partition. Under the initial community partition, for each training task I, calculate the value of the current utility function, denoted as the utility U_I of training task I; calculate the utility U' of adding it to each other community; If there exists U' greater than U_I, then move training task I to the community with the maximum utility; repeat until no training task can obtain a higher utility by changing the community; at this time, each community contains several training tasks; Create a new graph structure, retain all the nodes of the original training task network graph model, add the attribute of the community to each node based on the community to which each training task belongs, and recalculate the weights of the edges between the nodes. On the basis of the original weights, synchronously increase the weights of the edges between the nodes within the same community by a fixed random number α1 in the interval (0, 1); use the force-directed algorithm to optimize the node layout to obtain the partitioned network graph.

6. A method for linked management of military training plans and resources according to claim 5, characterized in that The acquisition method of the optimal training resource allocation plan includes: Represent each type of training resource as a spin variable, and the spin variable takes +1 or -1. +1 means the corresponding training resource is used, and -1 means the corresponding training resource is not used; Define the total energy H of the training resource dynamic model; Among them, \(w_u\) is the weight of training resource \(u\), \(c_u\) is the usage cost of training resource \(u\), \(c_{max}\) is the highest usage cost among all training resources; \(\sigma_u\) is the spin variable corresponding to training resource \(u\); \(J_{u,v}\) is the coupling strength between training resource \(u\) and training resource \(v\); \(\sigma_v\) is the spin variable of training resource \(v\); Among them, s_u,v is the spatial correlation between training resources u and training resource v; smax is the maximum value of the spatial correlation between training resources, t_u,v is the temporal correlation between training resources u and training resource v; tmax is the maximum value of the temporal correlation between training resources, e_u,v is the efficiency correlation between training resources u and training resource v; β1, β2, and β3 are the weights of the coupling strength terms, and the sum of the three is 1; Randomly assign +1 or -1 to each spin variable, set the simulation temperature T and the cooling coefficient r; for each iteration, randomly select a spin variable and calculate the difference ΔH in the total energy H before and after flipping the spin variable; flipping means +1 becomes -1 or -1 becomes +1; If ΔH is less than 0, accept the flip. If ΔH is greater than 0, accept the flip with probability , where k1 is a preset constant; repeat until the change in the total energy is less than a preset energy threshold; during this process, reduce the simulation temperature, that is, multiply the simulation temperature by a temperature reduction coefficient for each iteration as the simulation temperature for the next iteration; record the values of the spin variables when the total energy is the lowest during the simulation process to form an optimal resource allocation plan.

7. A method for the linkage management of military training plans and resources according to claim 6, characterized in that, The method of constructing the bipartite graph includes: Create two vertex sets V1 and V2: V1 represents the set composed of training tasks, and V2 represents the set composed of training resources. Initialize an empty edge set E. Traverse each node in the partitioned network graph and map it to a vertex in V1. Traverse each resource in the optimal resource allocation scheme and map it to a vertex in V2. For each vertex in V1, analyze the resource requirements of the training task and classify the requirements into essential resources and optional resources. For each vertex in V1 and each vertex in V2, calculate the mutual compatibility score. For vertex vj in V1 and vertex vi in V2, if vertex vi is an essential resource of vertex vj, or vertex vi is an optional resource of vertex vj and the compatibility score is greater than the preset compatibility threshold, then connect an edge between vertex vi and vertex vj and add it to the edge set E. Use the compatibility score as the weight of the edge in the edge set E. Obtain a preliminary bipartite graph. Use a graph optimization algorithm to optimize the preliminary bipartite graph to obtain a bipartite graph.

8. A method for the linked management of military training plans and resources according to claim 7, characterized in that, The obtaining method of the optimal task-resource matching scheme includes: Structurally store the constructed bipartite graph to generate a task-resource association matrix F. Find the edge corresponding to the maximum weight in the task-resource association matrix and its weight value mw. In F, find all edges with weights equal to mw to form an equal-weight line. Starting from the equal-weight line, select an uncovered point on the equal-weight line, and find an augmenting path starting from this point. If an augmenting path is found, augment the non-adjacent set M according to the indication of the augmenting path. If no augmenting path is found, modify the labels and continue to search for an augmenting path, repeating until all points on the equal-weight line are covered. For each edge in M, set the corresponding element in F to 0, indicating that this edge has been occupied. Repeat until the weights of all edges are 0. Each time an edge corresponding to the maximum weight is found, reconstruct the equal-weight line and continue the next round until all elements in F are 0. Combine the Ms found in each round to obtain the final optimal task-resource matching scheme.

9. A military training plan and resource linkage management system, which is used to implement the military training plan and resource linkage management method according to any one of claims 1 to 8, characterized in that, It includes: A data acquisition module for acquiring training resource data and n groups of training task data. A preliminary model construction module for establishing a training task network graph model based on n groups of training task data. A partitioning module for using an improved community discovery algorithm to process the training task network graph model to obtain a partitioned network graph. A dynamic simulation module for constructing a training resource dynamic model based on the training resource data and simulating the training resource dynamic model to obtain an optimal training resource allocation scheme. A scheme fitting module for constructing a bipartite graph based on the partitioned network graph and the optimal training resource allocation scheme. Use the Hungarian algorithm to solve the bipartite graph to obtain an optimal task-resource matching scheme, and send the optimal task-resource matching scheme to the linkage management terminal. Each module is connected by wired and / or wireless means.

Images