Resource allocation method and device for killing network nodes

Abstract

The embodiment of the invention provides a resource allocation method and device for killing network nodes, and is applied to the technical field of computers. When the method is executed, combat targets are allocated to nodes of the killing network by utilizing real-time battlefield situation data of combat resource node information and combat target information in the killing network so as to obtain a plurality of candidate allocation strategies, and the nodes of the killing network are allocated based on expected execution effects of the candidate allocation strategies. The candidate allocation strategy meeting the expectation is obtained from the multiple candidate allocation strategies to serve as the target allocation strategy, and it can be ensured that the combat target is allocated to the appropriate combat equipment. According to the mode, the reasonability and the accuracy of resource allocation in the nodes are improved, so that the method is adaptive to a complex and changeable battlefield environment, and the response speed and the combat effectiveness of the whole combat task are improved.

Description

Technical Field

[0001] This application relates to the field of computers, and more particularly to a method and apparatus for allocating resources for kill network nodes. Background Technology

[0002] Kill network nodes, also known as friendly nodes in a kill network, refer to various combat resources within the kill network, including different types of weapons, sensors, and communication equipment. One kill network node constitutes one combat resource.

[0003] Resource allocation within a kill network refers to the mobilization of combat resources within these nodes to counter enemy targets in a battlefield environment, and is crucial for ensuring operational effectiveness. Currently, resource allocation for kill network nodes is often based on historical combat experience, combined with pre-defined rules or static scheduling algorithms to specify allocation strategies. However, in complex and ever-changing battlefield environments, this resource allocation strategy often suffers from unreasonable resource allocation, thus affecting the overall response speed and operational effectiveness of the combat mission. Summary of the Invention

[0004] This application provides a method and apparatus for resource allocation of kill network nodes, which is used to rationally allocate node resources in complex and ever-changing battlefield environments, thereby improving the response speed and combat effectiveness of overall combat missions.

[0005] In a first aspect, embodiments of this application provide a resource allocation method for kill network nodes, the method comprising:

[0006] Acquire real-time battlefield situation data; the real-time battlefield situation data includes information on combat resource nodes and combat targets within the kill network.

[0007] Based on the real-time battlefield situation data, multiple candidate allocation strategies are determined; the candidate allocation strategies indicate the candidate allocation strategies for assigning combat targets to nodes within the kill network;

[0008] Determine the expected performance of each candidate allocation strategy;

[0009] Based on the expected execution effect of each candidate allocation strategy, a target allocation strategy is obtained from the multiple candidate allocation strategies, and resource allocation of nodes is performed based on the target allocation strategy.

[0010] Optionally, determining multiple candidate allocation strategies based on the real-time battlefield situation data includes:

[0011] Based on the real-time battlefield situation data, a state space and an action space are determined. The state space consists of quantified combat resource node information and combat target information. The action space indicates the behavioral state of the combat resource nodes within the kill network.

[0012] Based on the state space and the action space, determine the set of candidate combat targets for each combat resource node within the kill network;

[0013] Obtain the matching degree between each combat resource node and its corresponding combat target; based on the matching degree between the combat resource node and its corresponding combat target, assign multiple candidate combat targets to the combat resource node to obtain multiple candidate allocation strategies; the candidate combat targets of the combat resource node indicate combat targets whose matching degree with the combat resource node is greater than or equal to a preset matching degree threshold.

[0014] Optionally, the operational target information includes the target characteristics of the operational target and the environmental constraint information of the operational target, and the overall operational mission includes multiple operational targets;

[0015] The state space includes the quantitative characteristics of nodes in the kill network, the quantitative characteristics of combat targets, and the quantitative characteristics of environmental constraint information. The quantitative characteristics of a node include the remaining available quantity of combat resources, damage effectiveness, and range constraints of the node. The quantitative characteristics of a combat target include the target threat level, target value weight, and vulnerability coefficient of the combat target. The combat target includes multiple combat targets, and each combat target corresponds to a target type of the combat target.

[0016] The process of obtaining the matching degree between each combat resource node and its corresponding combat target includes:

[0017] According to formula (1), the matching degree between the combat resource node and the corresponding combat target is obtained:

[0018] (1)

[0019] in, Let be the matching degree between the i-th node and the j-th combat target, where i and j are both positive integers. The damage performance of the i-th node is... Vulnerability of the j-th combat target Let be the distance between the i-th node and the j-th combat target. Let the target value weight be the j-th operational target. Let the target threat level be the j-th operational target. Let be the remaining amount of combat resources in the i-th node. The total combat resources in the i-th node. The first weighting coefficient, These are the second weighting coefficients, all of which are greater than 0.

[0020] Optionally, determining the expected execution effect of each of the candidate allocation strategies includes:

[0021] Based on the pre-constructed reward function, the reward function value of each candidate allocation strategy is determined;

[0022] The reward function value of each candidate allocation strategy is positively correlated with at least one of the following: completion time, success rate, resource utilization rate, and / or target achievement degree of the overall combat mission corresponding to the candidate allocation strategy; the reward function value of each candidate allocation strategy indicates the execution effect of the candidate allocation strategy.

[0023] The step of obtaining the target allocation strategy from a plurality of candidate allocation strategies based on the expected execution effect of each candidate allocation strategy includes:

[0024] Based on the reward function value of each candidate allocation strategy, the allocation strategy with the largest reward function value among the multiple candidate allocation strategies is selected as the target allocation strategy.

[0025] Optionally, the reward function is used to maximize the target damage outcome, minimize resource consumption, and minimize the risk of the kill net.

[0026] Optionally, the reward function is shown in formula (2):

[0027] (2)

[0028] R is the reward function. For damage reward function, For risk penalty function, A time-based reward function for the completion of the overall combat mission. This is a resource consumption penalty function. For the third weighting coefficient, It is the fourth weighting coefficient. The fifth weighting coefficient is greater than 0;

[0029] Using formula (3), obtain :

[0030] (3)

[0031] Where n is the number of the combat targets, This represents the hit probability of the i-th node against the j-th combat target. The damage performance of the i-th node is... Vulnerability of the j-th combat target Let the target value weight be the j-th operational target. For 0-1 variables, This indicates that the i-th node is assigned to the j-th combat target. This indicates that the i-th node was not assigned to the j-th combat target;

[0032] Obtain using formula (4) :

[0033] (4)

[0034] in, This represents the probability that the j-th operational target has not been allocated resources; This represents a 0-1 function where the target value is below a threshold. The time is 1; This represents the penalty coefficient for low-value targets; Let m be the target threat level of the j-th operational target, and m be the number of operational targets in the entire operational system.

[0035] Obtain using formula (5)

[0036] (5)

[0037] in, This represents the unit cost of the i-th node; Indicates the remaining proportion of resources

[0038] Obtain using formula (6) :

[0039] (6)

[0040] in, Indicates the total validity period of node allocation; This indicates the time taken to complete the resource allocation for the node; This represents the time-based reward coefficient.

[0041] Optionally, the step of determining multiple candidate allocation strategies based on the real-time battlefield situation data; determining the expected execution effect of each candidate allocation strategy; and obtaining a target allocation strategy from the multiple candidate allocation strategies based on the expected execution effect of each candidate allocation strategy includes:

[0042] Using reinforcement learning algorithms, multiple candidate allocation strategies are determined for the real-time battlefield situation data; the expected execution effect of each candidate allocation strategy is determined; and a target allocation strategy is obtained from the multiple candidate allocation strategies based on the expected execution effect of each candidate allocation strategy.

[0043] The reinforcement learning algorithm is used to obtain a state space and an action space based on the real-time battlefield situation data; and to process multiple candidate allocation strategies based on the state space, the action space, and a preset reward function to obtain the target allocation strategy; the state space consists of quantified combat resource node information and combat target information; the action space indicates the candidate combat targets of the combat resource nodes within the kill network; the reward function is positively correlated with at least one of the following: overall combat mission completion time, success rate, resource utilization rate, and / or target achievement degree.

[0044] Optionally, the reinforcement learning algorithm includes a deep Q-network.

[0045] Optionally, the method further includes:

[0046] Based on the real-time battlefield situation data, the overall situation value of the current battlefield is determined according to formula (7):

[0047] (7)

[0048] in, This indicates that the i-th node is assigned to the j-th combat target. This indicates that the i-th node was not assigned to the j-th combat target; Indicates time-based reward items. It is a positive number. The overall situational awareness value; m is the number of combat targets included in the overall combat mission, where m is a positive integer; n is the number of nodes in the kill network, where n is a positive integer;

[0049] Let the matching degree between the i-th node and the j-th combat target be:

[0050] (8)

[0051] Where i represents the i-th node, which is the target node, and j represents the j-th combat target, where i and j are both positive integers. The damage performance of the i-th node is... Vulnerability of the j-th combat target Let be the distance between the i-th node and the j-th combat target. Let the target value weight be the j-th operational target. Let the target threat level be the j-th operational target. Let be the remaining amount of combat resources in the i-th node. The total combat resources in the i-th node. The first weighting coefficient, These are the second weighting coefficients, all of which are greater than 0;

[0052] The parameters of the reinforcement learning algorithm are adjusted based on the overall situation of the current battlefield.

[0053] Secondly, embodiments of this application provide a resource allocation device for kill network nodes, the device comprising:

[0054] An information acquisition unit is used to acquire real-time battlefield situation data; the real-time battlefield situation data includes information on combat resource nodes and combat targets within the kill network.

[0055] The strategy determination unit is configured to determine multiple candidate allocation strategies based on the real-time battlefield situation data; the candidate allocation strategies indicate candidate allocation strategies for allocating combat targets to nodes within the kill network; determine the expected execution effect of each candidate allocation strategy; and obtain a target allocation strategy from the multiple candidate allocation strategies based on the expected execution effect of each candidate allocation strategy.

[0056] The strategy allocation unit is used to allocate resources to nodes based on the target allocation strategy.

[0057] Thirdly, embodiments of this application provide a computer program product, the computer program product comprising: a computer program (also referred to as code or instructions), which, when the computer program is run, causes a computer to perform a method in any of the possible implementations of any of the above aspects.

[0058] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program (also referred to as code or instructions) that, when run on a computer, causes the computer to perform the method in any of the possible implementations of any of the above aspects.

[0059] Fifthly, embodiments of this application provide a chip system including one or more processors for calling and executing instructions stored in memory, causing the methods in any of the above aspects or possible implementations to be executed. The chip system may be composed of chips or may include chips and other discrete devices.

[0060] This application provides a method and apparatus for resource allocation of kill network nodes. The method includes: acquiring real-time battlefield situation data; the real-time battlefield situation data includes information on combat resource nodes and combat targets within the kill network; determining multiple candidate allocation strategies based on the real-time battlefield situation data; the candidate allocation strategies indicating candidate allocation strategies for allocating combat targets to nodes within the kill network; determining the expected execution effect of each candidate allocation strategy; and, based on the expected execution effect of each candidate allocation strategy, obtaining a target allocation strategy from the multiple candidate allocation strategies, and performing node resource allocation based on the target allocation strategy.

[0061] By utilizing real-time battlefield situational data on combat resource nodes and targets within a kill network, and based on this data, combat targets are assigned to nodes within the kill network to generate multiple candidate allocation strategies. Based on the expected performance of these strategies, the optimal allocation strategy is selected as the target allocation strategy, ensuring that combat targets are assigned to the most suitable combat equipment. This approach offers two advantages: First, it allows for the rapid acquisition of node allocation strategies synchronized with the current battlefield situation, avoiding the information lag inherent in relying on historical experience to determine node resource allocation. Second, by constructing a correlation between combat targets and node information, this method selects the strategy with the best performance from multiple candidate strategies, improving the rationality and accuracy of resource allocation within nodes. This adapts to complex and ever-changing battlefield environments, enhancing the overall response speed and combat effectiveness of combat missions. Attached Figure Description

[0062] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0063] Figure 1 A flowchart illustrating a resource allocation method for kill network nodes provided in an embodiment of this application;

[0064] Figure 2 A flowchart illustrating another node resource allocation method provided in this application embodiment;

[0065] Figure 3 This is a schematic diagram of a resource allocation device for kill network nodes provided in an embodiment of this application. Detailed Implementation

[0066] To enable those skilled in the art to better understand the present application, the technical solutions in this embodiment will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0067] As mentioned earlier, current kill network node resource allocation is often based on historical combat experience, combined with pre-set rules or static scheduling algorithms. When the battlefield situation changes, such as when the enemy adjusts its attack strategy, new combat targets are added, or some nodes are damaged, the above methods cannot make timely adjustments to the allocation strategy, resulting in an unreasonable resource allocation strategy, which in turn affects the overall response speed and combat effectiveness of the combat mission.

[0068] In view of this, embodiments of this application provide a resource allocation method for kill network nodes. By utilizing real-time battlefield situational data of combat resource node information and combat target information within the kill network, and based on this real-time battlefield situational data, combat targets are allocated to the kill network nodes to obtain multiple candidate allocation strategies. Based on the expected execution effect of these candidate allocation strategies, the candidate allocation strategy that meets the expectations is selected from the multiple candidate allocation strategies and used as the target allocation strategy, ensuring that nodes are allocated to the most suitable combat target. This approach, on the one hand, can quickly obtain node allocation strategies synchronized with the actual situation of the current battlefield based on real-time battlefield situational data, avoiding the information lag characteristic of determining node resource allocation based on historical experience. On the other hand, by constructing the correlation between combat targets and node information, this method can filter out the strategy with the best execution effect from multiple candidate strategies, improving the rationality and accuracy of resource allocation in nodes, thereby adapting to complex and ever-changing battlefield environments and improving the overall response speed and combat effectiveness of combat missions.

[0069] The resource allocation method for kill network nodes provided in this application embodiment will be described in detail below with reference to the accompanying drawings.

[0070] The resource allocation method for kill network nodes provided in this application can be applied to electronic devices with computing capabilities, such as servers, server clusters, computers, laptops, etc., and is not specifically limited in this application. For ease of explanation, the resource allocation method for kill nodes is illustrated below using a server as an example.

[0071] Appendix Figure 1 A flowchart illustrating a resource allocation method for kill network nodes provided in this application embodiment is included, comprising the following steps:

[0072] S10 acquires real-time battlefield situational data.

[0073] Real-time battlefield situation data is used to describe the current battlefield situation. In this embodiment, real-time battlefield situation data includes information on combat resource nodes within the kill network and combat target information.

[0074] The combat resource node information within the kill network is used to describe the equipment information of the kill network. In the embodiments of this application, the combat resource node information within the kill network includes, but is not limited to, node type (also known as equipment type), performance parameters (such as detection range, strike accuracy, and mobility), current status (such as available, unavailable, and damaged), and location information of combat resources.

[0075] Operational target information is used to describe the situation of operational targets. In this embodiment, operational target information includes the target characteristics of the operational targets and the environmental constraints of the operational targets.

[0076] The target characteristics of combat targets include, but are not limited to, the type of combat target (e.g., fighter jets, missile launchers, ships, etc.), the number of combat targets, the movement trajectory of each combat type, the completion time limit of each combat type, the priority of each combat type, the equipment performance of each combat type (e.g., range, stealth capability, etc.), and the current status (e.g., ready to launch, in motion, concealed, etc.).

[0077] In this embodiment, the target characteristics of the combat target are an overall combat mission within the kill network, including multiple combat targets. Each combat target is a combat type and contains its unique target information. For example, the enemy target information includes three combat targets: a high-altitude reconnaissance aircraft, a cruise missile group, and an electronic jamming aircraft. Among them, the high-altitude reconnaissance aircraft is located at position 1, its movement state is high-altitude uniform flight, and its target threat state is to perform reconnaissance without direct attack capability; the cruise missile group is located at position 2, its movement state is low-altitude penetration, and its target threat state is to have direct destruction capability; the electronic jamming aircraft is located at position 3, its movement state is circling and waiting, and its target threat state is to suppress friendly detection capabilities and assist other targets in penetration.

[0078] Environmental constraint information indicates the battlefield environment in which the combat target is located. In the embodiments of this application, environmental constraint information includes, but is not limited to, the geographical environment of the combat target (such as the terrain, landforms, and vegetation cover of the enemy's position), meteorological conditions (such as wind speed and visibility in the enemy area), the artificial environment constructed by the enemy (such as fortifications, radar deployment density, electronic jamming areas, etc.), the remaining time of the enemy's combat target, and the distance between the combat target and combat resources.

[0079] It is understandable that the aforementioned real-time battlefield situation data can comprehensively and accurately reflect the current battlefield situation map, which helps to provide a reasonable node allocation strategy and obtain an accurate and reasonable resource allocation strategy.

[0080] S20 determines multiple candidate allocation strategies based on real-time battlefield situation data.

[0081] The candidate allocation strategy indicates a candidate allocation strategy for assigning combat targets to nodes in a kill network. In this embodiment, the candidate allocation strategy refers to the strategy for assigning candidate combat targets to idle nodes. Idle nodes refer to nodes within the kill network that have not been assigned combat missions and are in an available state. Electronic devices can determine multiple candidate allocation strategies based on real-time battlefield situation data.

[0082] In one specific implementation, the electronic device determines the state space and action space based on real-time battlefield situation data. The state space consists of quantified combat resource node information and combat target information.

[0083] The action space indicates the behavioral state of each combat resource node within the kill network under the current battlefield conditions. The behavioral states of combat resources include, but are not limited to, standby, maneuvering, combat, and withdrawal. The action space provides the current behavioral state of combat resources, offering a selectable operational range for subsequently providing candidate combat targets.

[0084] Next, the electronic equipment can determine the candidate combat targets of each combat resource node in the kill network based on the state space and action space; among which the candidate combat targets of the combat resource nodes include those in standby, maneuver, combat, and withdrawal.

[0085] In this embodiment, each combat resource node includes one or more candidate allocation strategies. For example, given two nodes (device A and device B) and two combat targets (equipment X and equipment Y), the candidate combat targets for each combat resource node may include: Allocation Strategy 1: Device A → Device X, Device B → Device Y; Allocation Strategy 2: Device A → Device Y, Device B → Device X; Allocation Strategy 3: Device A → Device X (Device B is not allocated for now). All these possible allocation combinations together constitute the candidate combat targets for each combat resource node within the kill network.

[0086] Electronic devices can calculate the matching degree between each combat resource node and its corresponding combat target in the state space. Based on this matching degree, multiple candidate combat targets are assigned to the combat resource nodes to obtain multiple candidate allocation strategies. Among them, the candidate combat targets are combat targets whose matching degree with idle nodes is greater than or equal to a preset matching degree threshold.

[0087] In one example, the state space includes the quantitative characteristics of nodes in the kill network, the quantitative characteristics of combat targets, and the quantitative characteristics of environmental constraint information; the quantitative characteristics of a node include the remaining available quantity of combat resources in that node, its damage effectiveness, and its range constraint; the quantitative characteristics of a combat target include the target threat level, target value weight, and vulnerability coefficient of that combat target; the combat target includes multiple combat targets, and each combat target corresponds to a target type of a combat target.

[0088] For example, the state space S can be represented as shown in formula (1):

[0089] S=[T1,T2,...,T n ,W1,W2...,W m [C](1)

[0090] Where T1 represents the quantitative characteristics of the first operational target, ..., T n Let quantify the characteristics of the nth operational target, where n is the number of operational targets. The quantified characteristics of an operational target include its target threat level, target value weight, and vulnerability coefficient. For example, the quantified characteristics of the nth operational target include its target threat level t. n Target value weight v n d, the coefficient of vulnerability n .

[0091] In one specific implementation, the target threat level of an operational target can be determined based on the target's priority and target type. The higher the priority of the operational target and the target type of the operational target is an attack target (such as a cruise missile), the higher the target threat level of the operational target. If the operational target is of low priority and the target type is a transport target, the target threat level of the operational target is low.

[0092] The target value weight of an operational objective is used to quantify its strategic value. For example, if the strategic values ​​of operational objectives are determined from highest to lowest as follows: command and control center, missile launch site, electronic warfare aircraft, reconnaissance aircraft, logistics supply convoy, and patrol vehicle, then the target value weight of a command and control center is the highest, while the target value weight of a patrol vehicle is the lowest.

[0093] The vulnerability coefficient of an operational target is used to represent the target's protective capability. The higher the protective capability, the lower the vulnerability coefficient of the operational target, and vice versa.

[0094] W1 represents the quantization feature of the first node, ..., W mLet w be the quantified characteristics of the m-th node. Here, m is the number of nodes within the kill network. The quantified characteristics of a node include the remaining available quantity of combat resources, its damage effectiveness, and its range constraints. For example, the quantified characteristics of the m-th node might include the remaining available quantity w of combat resources. n Damage effectiveness k n Range constraint r n C represents environmental constraint information, including the remaining time t for the combat objective, the distance d between the enemy and friendly forces, etc.

[0095] In this embodiment of the application, the electronic device can determine the matching degree between idle nodes and combat targets based on the state space.

[0096] In one example, the matching degree calculation between a single node and a single combat target measures the matching degree of the i-th node with the j-th combat target. As shown in formula (2):

[0097] (2)

[0098] Let i be the i-th node, which is an idle node, and j be the j-th combat target. Both i and j are positive integers. Let i be the damage performance of the i-th node. Vulnerability of the j-th combat target Let be the distance between the i-th node and the j-th combat target. Let the target value weight be the j-th operational target. Let the target threat level be the j-th operational target. Let represent the remaining amount of combat resources in the i-th node. The total combat resources in the i-th node. The first weighting coefficient, These are the second weighting coefficients, all of which are greater than 0.

[0099] Electronic devices can select targets with a matching degree greater than a preset matching degree threshold as candidate targets for an idle node based on the matching degree between the idle node and the combat target.

[0100] It should be noted that in this embodiment of the application, the global matching degree (also known as the overall situation value) of all nodes and all combat targets in the current battlefield can also be calculated. Based on this global matching degree, the quality of resource allocation of nodes in the current battlefield can be determined, as shown in formula (3):

[0101] (3)

[0102] in, This indicates that the i-th node is assigned to the j-th combat target. This indicates that the i-th node was not assigned to the j-th combat target; Indicates time-based reward items. It is a positive number. is the overall situational awareness value; m is the number of combat targets included in the overall combat mission, m is a positive integer; n is the number of nodes in the kill network, n is a positive integer.

[0103] In this embodiment of the application, it is determined whether the resource allocation strategy for nodes needs to be adjusted based on the overall situation of the current battlefield. For example, if... If the situation exceeds the threshold, no adjustment to the node's resource allocation strategy is needed; otherwise, the node's resource allocation strategy needs to be adjusted.

[0104] S30, determine the expected execution effect of each candidate allocation strategy, and based on the execution effect, obtain the target allocation strategy from multiple candidate allocation strategies.

[0105] In one specific implementation, a reward function can be constructed, using the reward function value of each candidate allocation strategy to indicate the expected performance of that strategy. Specifically, the higher the reward function value, the better the expected performance. The electronic device can determine the target allocation strategy based on the reward function value of each candidate allocation strategy. The target allocation strategy is the candidate allocation strategy with the highest reward function value.

[0106] In this embodiment, the reward function is positively correlated with at least one of the following: completion time, success rate, resource utilization rate, and / or degree of achievement of the overall combat mission corresponding to the candidate allocation strategy. When the resource allocation scheme can quickly and efficiently complete high-priority objectives and has high resource utilization, it will obtain a higher reward value; conversely, the reward value will be lower or even negative.

[0107] In one specific implementation, the reward function aims to maximize target damage, minimize resource consumption (ammunition, fuel, etc.), and minimize the risk to the kill web.

[0108] For example, the reward function is shown in formula (4):

[0109] (4)

[0110] R is the reward function. As a reward for damage, For risk penalty function, A time-sensitive reward function for the completion of the overall combat mission. This is a resource consumption penalty function. For the third weighting coefficient, It is the fourth weighting coefficient. The fifth weighting coefficient is greater than 0;

[0111] Using formula (5), obtain :

[0112] (5)

[0113] Where n is the number of the combat targets, This represents the hit probability of the i-th node against the j-th combat target. The damage performance of the i-th node is... Vulnerability of the j-th combat target Let the target value weight be the j-th operational target. For 0-1 variables, This indicates that the i-th node is assigned to the j-th combat target. This indicates that the i-th node was not assigned to the j-th combat target;

[0114] Obtain using formula (6) :

[0115] (6)

[0116] in, This represents the probability that the j-th operational target has not been allocated resources; This represents a 0-1 function where the target value is below a threshold. The time is 1; This represents the penalty coefficient for low-value targets; Let m be the target threat level of the j-th operational target, and m be the number of operational targets in the entire operational system.

[0117] Obtain using formula (7)

[0118] (7)

[0119] in, This represents the unit cost of the i-th node; Indicates the remaining proportion of resources

[0120] Obtain using formula (8) :

[0121] (8)

[0122] in, Indicates the total validity period of node allocation; This indicates the time taken to complete the resource allocation for the node; This represents the time-based reward coefficient.

[0123] In actual combat scenarios, the reward function needs to be dynamically adjusted according to changes in the battlefield situation. If the remaining combat resources are insufficient, the reward function should be increased. If a high-threat target appears, reduce the threat level. Prioritize strikes against high-threat targets.

[0124] Therefore, the embodiments of this application can accurately obtain the target allocation strategy by using the reward function value of each candidate allocation strategy.

[0125] S40, Based on the target allocation strategy, perform resource allocation for the nodes.

[0126] The node resource allocation method provided in this application utilizes real-time battlefield situational data of combat resource node information and combat target information within a kill network. Based on this real-time battlefield situational data, it allocates combat targets to nodes in the kill network to obtain multiple candidate allocation strategies. Based on the expected execution effect of these candidate allocation strategies, it selects the one that meets the expectations from among the multiple candidate strategies and uses it as the target allocation strategy. This ensures that nodes are allocated to the most suitable combat target. This approach, on the one hand, can quickly obtain node allocation strategies synchronized with the actual situation on the current battlefield based on real-time battlefield situational data, avoiding the information lag characteristic of determining node resource allocation based on historical experience. On the other hand, by constructing the correlation between combat targets and node information, this method can filter out the strategy with the best execution effect from multiple candidate strategies, improving the rationality and accuracy of resource allocation within nodes. This adapts to complex and ever-changing battlefield environments, improving the overall response speed and combat effectiveness of combat missions.

[0127] Furthermore, embodiments of this application also provide a reinforcement learning algorithm, based on which node resource allocation can be achieved.

[0128] Appendix Figure 2 A flowchart illustrating another node resource allocation method provided in this application embodiment, the method comprising the following:

[0129] S210 electronic equipment acquires real-time battlefield situational data.

[0130] In this embodiment, the electronic device collects battlefield situation data in real time, i.e., real-time battlefield situation data. The specific content of the real-time battlefield situation data is described in step S10, and will not be repeated here.

[0131] The S220 and electronic equipment use reinforcement learning algorithms to process real-time battlefield situation data and obtain target allocation strategies.

[0132] In this embodiment, the reinforcement learning algorithm is used to obtain a state space and an action space based on real-time battlefield situational data; and based on the state space and action space, and a preset reward function, to process multiple candidate allocation strategies to obtain a target allocation strategy; the state space consists of quantified combat resource node information and combat target information; the action space refers to the actions that a combat node can currently perform, for example, the action space of a currently idle combat node can be maneuver, position deployment, or attack status. The reward function is positively correlated with at least one of the following: overall combat mission completion time, success rate, resource utilization rate, and / or target achievement degree.

[0133] In one specific implementation, the reinforcement learning algorithm is the Deep Q-Network algorithm. The Deep Q-Network algorithm can handle high-dimensional state spaces and is suitable for state spaces containing a large amount of combat resource node information and combat target information; at the same time, through experience replay and target network mechanisms, it has high stability and convergence, and can better adapt to the dynamic changes in the battlefield environment.

[0134] The reinforcement learning algorithm's network structure includes an input layer, a hidden layer, and an output layer. The input layer converts real-time battlefield situational data into a state space and an action space, where the state space consists of quantified features that the neural network can recognize. The input layer then sends the state space and action space to the hidden layer.

[0135] The hidden layer consists of three fully connected neural networks, where the number of neurons in the first fully connected neural network is greater than that in the second fully connected neural network, and the number of neurons in the second fully connected neural network is greater than that in the third fully connected neural network. The hidden layer processes the state space and action space to obtain the adaptation relationship between the combat target and the node, thereby determining multiple candidate allocation strategies.

[0136] The output layer is the same size as the action space. The output layer is used to determine the reward function value of each candidate allocation policy among multiple candidate allocation policies, and outputs the target allocation policy based on the reward function value.

[0137] Specifically, in the network structure of the reinforcement learning algorithm, the number of neurons in the input layer is consistent with the dimension of the state space. The three-layer hidden layer has 256, 128, and 64 neurons respectively, all using the ReLU activation function. This is specifically used to perform nonlinear transformations and higher-order feature extraction on the input features, capturing the complex mapping relationship between "combat targets and nodes," thereby obtaining the action space. The number of neurons in the output layer is consistent with the size of the action space, and a pre-defined reward function is used to obtain the reward function value corresponding to the candidate combat targets of idle nodes and output the target allocation strategy.

[0138] In the embodiments of this application, reinforcement learning algorithms can be pre-trained.

[0139] It should be noted that before training, the core parameters must first be set. These core parameters include the experience replay pool, training batch size, learning rate, discount factor γ, and ε-greedy policy.

[0140] The experience replay pool is used to store experience data from interactions with the environment. By randomly sampling, the temporal correlation of samples is broken, reducing training fluctuations. The capacity of the experience replay pool can be set to a preset value, such as 10,000, and the storage format can be (state, action, reward, next state).

[0141] Training batches refer to randomly selecting a preset number of samples (e.g., 32) from the experience replay pool for each training session, balancing training efficiency and parameter update stability.

[0142] The initial learning rate is set to a preset learning rate, for example, 0.001, and then an adaptive learning rate strategy is adopted, which gradually decreases as the training process progresses.

[0143] The discount factor γ is set to a preset discount value, such as 0.9, to weigh the importance of current rewards and future rewards, making reinforcement learning focus more on long-term cumulative rewards.

[0144] In one specific example, this application embodiment also provides a training process for a reinforcement learning algorithm, which includes the following steps:

[0145] Step 1: Initialize the main network parameter θ and the target network parameter θ′. The initial parameters of both are the same. Initialize the experience replay pool to be empty.

[0146] Step 2: Action selection and sample collection.

[0147] According to the ε-greedy policy, an action is selected in the current state. After executing the action, a reward and the next state are obtained, and the sample is stored in the experience replay pool. At the current time t, the action selected by the combat resources... The expression is:

[0148] (9)

[0149] in, for It decays linearly with the number of steps. =max(0.1, 1.0 - 0.0001 × t), where t is the number of training steps. The state-action value output by the main network.

[0150] Step 3: Model training.

[0151] Training begins when the number of samples in the experience replay pool reaches a preset threshold. Each time, a batch of samples is randomly drawn from the experience replay pool, and the maximum Q-value of the next state is calculated using the target network. This maximum Q-value is then combined with the current reward value to calculate the target Q-value. The expression for the target Q-value is shown in formula (10):

[0152] (10)

[0153] in, =0.9 is the discount factor. This represents the Q-value estimate of the target network for the optimal action in the next state. Let Q be the target Q value at the current moment. The immediate reward obtained from the environment after performing an action at time t. The state to be entered after the current action is performed. For the next state The possible actions to choose from.

[0154] The main network outputs a predicted Q-value based on the current state and action, and calculates the loss between the predicted Q-value and the target Q-value using the mean squared error loss function. The parameters of the main network are updated using gradient descent. The mean squared error loss function is expressed as equation (11):

[0155] (12)

[0156] Update the main network parameters using gradient descent. : ,in The initial learning rate, This represents the gradient of the loss function with respect to the parameters. Number of batch samples A batch of samples drawn from the experience replay pool.

[0157] The parameters of the main network are copied to the target network at a set update frequency to update the baseline for calculating the target Q-value. Training stops when the number of training steps reaches a preset maximum or the algorithm converges, resulting in an optimized reinforcement learning algorithm for practical resource allocation decisions.

[0158] S230. Based on the target allocation strategy, allocate node resources of the kill network.

[0159] Specifically, the higher-level command and control system distributes the target allocation strategy to the nodes of each combat resource within the kill network, and each node begins to execute combat missions according to the target allocation strategy.

[0160] Furthermore, electronic equipment can also perform preliminary verification of the feasibility of the operational plan. If problems such as resource node task scheduling conflicts or lack of execution conditions are found, the plan can be readjusted. If the plan is feasible, it can be sent to the superior command and control system.

[0161] S240. During the execution of combat missions at each node, real-time execution data such as mission progress, actual resource consumption, and combat effectiveness monitoring data are acquired.

[0162] S250: Using a pre-built reward function, the execution data such as the task progress, actual resource consumption, and combat effect monitoring data of each node are processed to obtain the reward function value for this resource allocation and the issues in the task execution process.

[0163] S260. Based on the reward function value of the secondary resource allocation and the problem of task execution, the electronic device updates the network parameters of the reinforcement learning algorithm to achieve dynamic updating of the resource allocation strategy.

[0164] This invention, based on reinforcement learning algorithms, can adjust the allocation strategy of combat resource nodes according to the real-time battlefield situation. When unexpected situations arise, it can quickly learn and adapt to new environments, making optimal resource allocation decisions. This overcomes the poor adaptability of traditional fixed rules and static algorithms, greatly improving the flexibility of resource allocation. This invention can continuously update and optimize its strategy; with the accumulation of input examples, the combat resource allocation strategy will gradually improve, thereby enhancing the overall combat effectiveness of the kill network. Furthermore, embodiments of this application can dynamically adjust the network parameters in the reinforcement learning algorithm based on the reward function value and problems encountered during mission execution, thereby achieving dynamic updates to the resource allocation strategy.

[0165] In addition, this application embodiment also provides a resource allocation device for kill network nodes.

[0166] Appendix Figure 3 This application provides a schematic diagram of a resource allocation device for a kill network node, the device 300 comprising:

[0167] The information acquisition unit 301 is used to acquire real-time battlefield situation data; the real-time battlefield situation data includes information on combat resource nodes and combat targets within the kill network.

[0168] The strategy determination unit 302 is used to determine multiple candidate allocation strategies based on the real-time battlefield situation data; the candidate allocation strategies indicate candidate allocation strategies for allocating combat targets to nodes within the kill network; determine the expected execution effect of each candidate allocation strategy; and obtain a target allocation strategy from the multiple candidate allocation strategies based on the expected execution effect of each candidate allocation strategy.

[0169] The strategy allocation unit 303 is used to allocate resources to nodes based on the target allocation strategy.

[0170] The strategy determination unit 302 is specifically used to: determine a state space and an action space based on the real-time battlefield situation data, wherein the state space is quantified combat resource node information and combat target information; and the action space indicates the behavioral state of the combat resource nodes within the kill network.

[0171] Based on the state space and the action space, determine the set of candidate combat targets for each combat resource node within the kill network;

[0172] Obtain the matching degree between each combat resource node and its corresponding combat target; based on the matching degree between the combat resource node and its corresponding combat target, assign multiple candidate combat targets to the combat resource node to obtain multiple candidate allocation strategies; the candidate combat targets of the combat resource node indicate combat targets whose matching degree with the combat resource node is greater than or equal to a preset matching degree threshold.

[0173] The operational target information includes the target characteristics of the operational target and the environmental constraint information of the operational target, and the operational target includes multiple operational targets;

[0174] The state space includes the quantitative characteristics of nodes in the kill network, the quantitative characteristics of combat targets, and the quantitative characteristics of environmental constraint information. The quantitative characteristics of a node include the remaining available quantity of combat resources, damage effectiveness, and range constraints of the node. The quantitative characteristics of a combat target include the target threat level, target value weight, and vulnerability coefficient of the combat target. The combat target includes multiple combat targets, and each combat target corresponds to a target type of the combat target.

[0175] The process of obtaining the matching degree between each combat resource node and its corresponding combat target includes:

[0176] According to formula (1), the matching degree between the combat resource node and the corresponding combat target is obtained:

[0177] (1)

[0178] Where i represents the i-th node, which is the idle node, and j represents the j-th combat target, where i and j are both positive integers. The damage performance of the i-th node is... Vulnerability of the j-th combat target Let be the distance between the i-th node and the j-th combat target. Let the target value weight be the j-th operational target. Let the target threat level be the j-th operational target. Let be the remaining amount of combat resources in the i-th node. The total combat resources in the i-th node. The first weighting coefficient, These are the second weighting coefficients, all of which are greater than 0.

[0179] Optionally, determining the expected execution effect of each of the candidate allocation strategies includes:

[0180] Based on the pre-constructed reward function, the reward function value of each candidate allocation strategy is determined;

[0181] The reward function value of each candidate allocation strategy is positively correlated with at least one of the following: completion time, success rate, resource utilization rate, and / or target achievement degree of the overall combat mission corresponding to the candidate allocation strategy; the reward function value of each candidate allocation strategy indicates the execution effect of the candidate allocation strategy.

[0182] The step of obtaining the target allocation strategy from a plurality of candidate allocation strategies based on the expected execution effect of each candidate allocation strategy includes:

[0183] Based on the reward function value of each candidate allocation strategy, the allocation strategy with the largest reward function value among the multiple candidate allocation strategies is selected as the target allocation strategy.

[0184] Optionally, the reward function is used to maximize the target damage outcome, minimize resource consumption, and minimize the risk of the kill net.

[0185] Optionally, the reward function is as shown in formula (2):

[0186] (2)

[0187] R is the reward function. As a reward for damage, For risk penalty function, A time-sensitive reward function for the completion of the overall combat mission. This is a resource consumption penalty function. For the third weighting coefficient, It is the fourth weighting coefficient. The fifth weighting coefficient is greater than 0;

[0188] Using formula (3), obtain :

[0189] (3)

[0190] Where n is the number of the combat targets, This represents the hit probability of the i-th node against the j-th combat target. The damage performance of the i-th node is... Vulnerability of the j-th combat target Let the target value weight be the j-th operational target. For 0-1 variables, This indicates that the i-th node is assigned to the j-th combat target. This indicates that the i-th node was not assigned to the j-th combat target;

[0191] Obtain using formula (4) :

[0192] (4)

[0193] in, This represents the probability that the j-th operational target has not been allocated resources; This represents a 0-1 function where the target value is below a threshold. The time is 1; This represents the penalty coefficient for low-value targets; Let m be the target threat level of the j-th operational target, and m be the number of operational targets in the entire operational system.

[0194] Obtain using formula (5) :

[0195] (5)

[0196] in, This represents the unit cost of the i-th node; Indicates the remaining proportion of resources

[0197] Obtain using formula (6) :

[0198] (6)

[0199] in, Indicates the total validity period of node allocation; This indicates the time taken to complete the resource allocation for the node; This represents the time-based reward coefficient.

[0200] Optionally, the strategy determination unit 302 is further configured to: use a reinforcement learning algorithm to determine multiple candidate allocation strategies for the real-time battlefield situation data; determine the expected execution effect of each candidate allocation strategy; and obtain a target allocation strategy from the multiple candidate allocation strategies based on the expected execution effect of each candidate allocation strategy.

[0201] The reinforcement learning algorithm is used to obtain a state space and an action space based on the real-time battlefield situation data; and to process multiple candidate allocation strategies based on the state space, the action space, and a preset reward function to obtain the target allocation strategy; the state space consists of quantified combat resource node information and combat target information; the matching degree between each combat resource node and its corresponding combat target is obtained; the reward function is positively correlated with at least one of the following: overall combat mission completion time, success rate, resource utilization rate, and / or target achievement degree.

[0202] Optionally, the reinforcement learning algorithm includes a deep Q-network.

[0203] Optionally, the device 300 further includes a feedback unit for:

[0204] Based on the real-time battlefield situation data, the overall situation value of the current battlefield is determined according to formula (7):

[0205] (7)

[0206] in, This indicates that the i-th node is assigned to the j-th combat target. This indicates that the i-th node was not assigned to the j-th combat target; Indicates time-based reward items. It is a positive number. The overall situational awareness value; m is the number of combat targets included in the overall combat mission, where m is a positive integer; n is the number of nodes in the kill network, where n is a positive integer;

[0207] Let the matching degree between the i-th node and the j-th combat target be:

[0208] (8)

[0209] Where i represents the i-th node, which is the target node, and j represents the j-th combat target, where i and j are both positive integers. The damage performance of the i-th node is... Vulnerability of the j-th combat target Let be the distance between the i-th node and the j-th combat target. Let the target value weight be the j-th operational target. Let the target threat level be the j-th operational target. Let be the remaining amount of combat resources in the i-th node. The total combat resources in the i-th node. The first weighting coefficient, These are the second weighting coefficients, all of which are greater than 0;

[0210] The parameters of the reinforcement learning algorithm are adjusted based on the overall situation of the current battlefield.

[0211] The node resource allocation device provided in this application utilizes real-time battlefield situational data of combat resource node information and combat target information within a kill network. Based on this real-time battlefield situational data, it allocates combat targets to nodes in the kill network to obtain multiple candidate allocation strategies. Based on the expected execution effect of these candidate allocation strategies, it selects the candidate allocation strategy that meets the expectations from among the multiple candidate strategies and uses it as the target allocation strategy. This ensures that nodes are allocated to the most suitable combat target. This approach, on the one hand, can quickly obtain node allocation strategies synchronized with the actual situation on the current battlefield based on real-time battlefield situational data, avoiding the information lag characteristic of determining node resource allocation based on historical experience. On the other hand, by constructing the correlation between combat targets and node information, this method can filter out the strategy with the best execution effect from multiple candidate strategies, improving the rationality and accuracy of resource allocation in nodes, thereby adapting to complex and ever-changing battlefield environments and enhancing the overall response speed and combat effectiveness of combat missions.

[0212] In the embodiments of this application, the terms and English abbreviations are exemplary examples given for ease of description and should not be construed as limiting the application in any way. This application does not preclude the possibility of defining other terms that can achieve the same or similar functions in existing or future agreements.

[0213] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When these computer instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated.

[0214] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

Claims

1. A method for resource allocation for a sink node, characterized in that, The method comprises: obtaining real-time battlefield situation data; the real-time battlefield situation data comprises node information of operational resources in the kill web and operational target information; determining a plurality of candidate allocation strategies according to the real-time battlefield situation data; the candidate allocation strategy indicates a candidate allocation strategy for allocating operational targets to nodes in the kill web; determining the expected execution effect of each candidate allocation strategy; obtaining a target allocation strategy from the plurality of candidate allocation strategies according to the expected execution effect of each candidate allocation strategy, and performing resource allocation of nodes based on the target allocation strategy.

2. The method of claim 1, wherein, The method comprises: determining a state space and an action space according to the real-time battlefield situation data; the state space is quantized operational resource node information and operational target information; the action space indicates the behavior state of the operational resource node in the kill web; determining a candidate operational target set of each operational resource node in the kill web according to the state space and the action space; obtaining the matching degree between each operational resource node and the corresponding operational target; according to the matching degree between the operational resource node and the corresponding operational target, a plurality of candidate operational targets are allocated to the operational resource node to obtain a plurality of candidate allocation strategies; the candidate operational target of the operational resource node indicates an operational target with a matching degree greater than or equal to a preset matching degree threshold.

3. The method of claim 2, wherein, The operational target information comprises target characteristics of the operational target and environmental constraint information of the operational target; the overall operational task comprises a plurality of operational targets; The state space comprises quantized characteristics of nodes in the kill web, quantized characteristics of operational targets, and quantized characteristics of environmental constraints; the quantized characteristics of the node comprise the remaining available number of operational resources in the node, damage effectiveness, and range constraints; the quantized characteristics of the operational target comprise target threat degree, target value weight, and vulnerability coefficient of the operational target; each operational target corresponds to a target type of the operational target; The method comprises: According to formula (1), the matching degree between the operational resource node and the corresponding operational target is obtained: （1） wherein, is a matching degree of the ith node and the jth combat target, i and j are both positive integers, is a damage effectiveness of the ith node, is a vulnerability of the jth combat target is a distance between the ith node and the jth combat target, is a target value weight of the jth combat target, is a target threat degree of the jth combat target, is a remaining number of combat resources in the ith node, is a total combat resource in the ith node, is a first weight coefficient, is a second weight coefficient, both greater than 0.

4. The method of claim 1, wherein, The method comprises: determining the reward function value of each candidate allocation strategy based on a pre-constructed reward function; The reward function value of each candidate allocation strategy is positively correlated with at least one of the completion time, success rate, resource utilization rate, and / or target achievement degree of the overall operational task corresponding to the candidate allocation strategy; the reward function value of each candidate allocation strategy indicates the execution effect of the candidate allocation strategy; The method comprises: According to the reward function value of each candidate allocation strategy, the allocation strategy with the maximum reward function value in the plurality of candidate allocation strategies is taken as the target allocation strategy.

5. The method of claim 4, wherein, The reward function is used to maximize the target damage result, minimize resource consumption and minimize the risk of the killing net.

6. The method of claim 5, wherein, The reward function is shown in formula (2): （2） R is a reward function, is a damage reward function, is a risk penalty function, is a timeliness reward function of overall combat task completion, is a resource consumption penalty function, is a third weight coefficient, is a fourth weight coefficient, is a fifth weight coefficient, all greater than 0; Using equation (3), we obtain : （3） wherein n is the number of the combat targets, denotes the hit probability of the i-th node to the j-th combat target is the damage effectiveness of the i-th node, is the vulnerability of the j-th combat target is the target value weight of the j-th combat target, is a 0-1 variable, denotes the assignment of the i-th node to the j-th combat target, denotes the non-assignment of the i-th node to the j-th combat target. Using equation (4) to obtain : （4） wherein, Pj represents the probability that the jth target is not allocated resources; I represents a 0-1 function, which is 1 when the target value is below a threshold value I represents a 0-1 function, which is 1 when the target value is below a threshold value I represents a 0-1 function, which is 1 when the target value is below a threshold value Pj represents the probability that the jth target is not allocated resources; Using equation (5) to obtain （5） wherein, represents the unit cost of the i-th node; represents the resource remaining ratio; Using equation (6) to obtain : （6） wherein, represents the total duration of the effective period of node allocation; represents the time consumption of completing resource allocation of the node; represents the time reward coefficient.

7. The method of claim 1, wherein, The method comprises the following steps: The method comprises the following steps: The reinforcement learning algorithm is used to obtain a state space and an action space based on the real-time battlefield situation data, and based on the state space and the action space and a preset reward function, the target distribution strategy is obtained by processing a plurality of candidate distribution strategies; the state space is the quantized combat resource node information and combat target information; the action space indicates the behavior state of the combat resource node in the killing net; the reward function is positively correlated with at least one of the completion time, success rate, resource utilization rate and / or target achievement degree of the overall combat task.

8. The method of claim 7, wherein, The reinforcement learning algorithm comprises a deep Q network.

9. The method of claim 7, wherein, The method further comprises: Based on formula (7), the overall situation value of the current battlefield is determined according to the real-time battlefield situation data: （7） wherein, represents that the ith node is assigned to the jth combat target, represents that the ith node is not assigned to the jth combat target; represents a time reward term, is positive, is an overall situation value; m is the number of combat targets included in an overall combat mission, m is a positive integer, and n is the number of nodes in the kill web, n is a positive integer. Matching degree of the ith node and the jth combat target: （8） wherein i is an ith node, the ith node is a target node, j is a jth combat target, i and j are positive integers, a damage effectiveness of the ith node, a vulnerability of the jth combat target a distance between the ith node and the jth combat target, a target value weight of the jth combat target, a target threat degree of the jth combat target, a remaining quantity of combat resources in the ith node, total combat resources in the ith node, a first weight coefficient, a second weight coefficient, both greater than 0. According to the overall situation value of the current battlefield, the parameters of the reinforcement learning algorithm are adjusted.

10. A resource allocation apparatus for killing a mesh node, characterized by, The device comprises: An information acquisition unit is configured to acquire real-time battlefield situation data; the real-time battlefield situation data comprises combat resource node information and combat target information in the killing net; A strategy determination unit is configured to determine a plurality of candidate distribution strategies according to the real-time battlefield situation data; the candidate distribution strategy indicates a candidate distribution strategy for distributing combat targets to nodes in the killing net; determine the expected execution effect of each candidate distribution strategy; and obtain a target distribution strategy from the plurality of candidate distribution strategies according to the expected execution effect of each candidate distribution strategy; A strategy distribution unit is configured to distribute resources of nodes based on the target distribution strategy.

Images