A method and system for controlling a UAV cluster based on a collaborative emotional decision network

Abstract

The application discloses a kind of unmanned aerial vehicle cluster control method and system based on collaborative emotional decision network, applied to unmanned aerial vehicle flight control field, comprising: fuzzy feature matrix is obtained by fuzzy processing to fusion feature vector;Determine the target weight of each fuzzy feature and each preset control action in the double decision channel of collaborative emotional decision network;Determine the preliminary action decision vector of each unmanned aerial vehicle based on fuzzy feature matrix and target weight;The consensus fusion of the target tracking information and preliminary action decision vector of each unmanned aerial vehicle obtains the collaborative control instruction of unmanned aerial vehicle cluster;Collaborative control instruction is converted into the action control instruction executed by each unmanned aerial vehicle.The application converts the perception feature of each unmanned aerial vehicle into action decision by the cooperation of double decision channel of collaborative emotional decision network, and improves the collaborative control accuracy of unmanned aerial vehicle cluster as a whole by consensus fusion and instruction conversion to the action decision of each unmanned aerial vehicle.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) flight control technology, and in particular to a UAV swarm control method and system based on a collaborative emotion decision network. Background Technology

[0002] Existing UAV swarm control methods mainly include centralized control and distributed control. Centralized control relies on a single central node for global decision-making. While this offers good decision consistency, it suffers from high single-point-of-failure risk and limited system scalability. Distributed control, on the other hand, relies on local information exchange and autonomous decision-making, making it difficult to guarantee overall swarm coordination and task execution efficiency in dynamic and complex environments. Therefore, existing technologies suffer from low accuracy in UAV swarm control tasks due to poor overall control coordination. Summary of the Invention

[0003] This invention provides a method and system for controlling a drone swarm based on a collaborative emotion decision network, in order to improve the overall collaborative control accuracy of the drone swarm.

[0004] To address the aforementioned technical problems, embodiments of the present invention provide a method for controlling a drone swarm based on a collaborative emotion decision network, comprising: The fusion feature vector obtained by fusing multi-source sensing data is fuzzified to obtain a feature matrix containing multiple fuzzy features. Determine the target weights connecting each fuzzy feature and each preset control action in the dual decision-making channels of the collaborative emotion decision-making network; Based on the feature matrix and the target weight, the tendency and suppression of each preset control action are determined, and based on the tendency and suppression, the preliminary action decision vector of each UAV node is determined. Consensus fusion is performed on the target tracking information and preliminary action decision vectors of each UAV node to obtain the collaborative control command of the UAV cluster; The collaborative control commands are converted into action control commands for each of the UAV nodes, and each UAV node is controlled to execute the action control commands.

[0005] As one preferred embodiment, the step of fuzzifying the fused feature vector obtained from the fusion of multi-source sensing data to obtain a feature matrix containing multiple fuzzy features includes: Based on each preset fuzzy semantic representation, determine the membership degree of each feature in the fused feature vector to each preset fuzzy semantic representation; The membership degree is processed to obtain a feature matrix containing multiple fuzzy features.

[0006] As one preferred embodiment, the dual decision-making channels include an intuitive channel and a rational channel; determining the tendency and inhibition level of each preset control action based on the feature matrix and the target weights includes: Obtain the effectiveness evaluation results, original impulse intensity, and original inhibition value for each preset control action; Based on the learning rate of the intuitive channel, the feature matrix, the performance evaluation results, and the original impulse intensity, the degree of tendency for each preset control action is determined. Based on the learning rate of the rational channel, the feature matrix, the original impulse intensity, and the original inhibition value, the inhibition amount of each preset control action is determined.

[0007] As one preferred embodiment, determining the preliminary action decision vector for each UAV node based on the degree of inclination and the amount of suppression includes: The initial decision value is determined based on the difference between the degree of tendency and the amount of inhibition. If the effectiveness assessment value of the initial decision value is less than the expected assessment value, the target weight is adjusted, and the step of determining the initial decision value based on the difference between the degree of propensity and the amount of inhibition is returned until the effectiveness assessment value is greater than or equal to the expected assessment value. If the performance evaluation value of the initial decision value is greater than or equal to the expected evaluation value, the initial decision value of each preset control action is vectorized to obtain the preliminary action decision vector of each UAV node.

[0008] As one preferred embodiment, the consensus fusion of target tracking information and preliminary action decision vectors of each UAV node to obtain the collaborative control commands for the UAV swarm includes: The normalized values ​​determined based on target tracking information are weighted and fused to obtain the target confidence scores of each UAV node. Consensus fusion is performed on the target confidence and preliminary action decision vector of each UAV node to obtain the consensus support of each preset control action; Based on the consensus support, target actions are selected from each preset control action, and collaborative control instructions for the UAV swarm are generated according to the target actions.

[0009] As one preferred embodiment, the step of converting the collaborative control command into action control commands for each of the UAV nodes includes: Based on the cooperative control command, the composite weights of each UAV node on each preset motion component are adjusted; Based on the adjusted synthesis weights, the preset motion components are weighted and fused to obtain the total control force vector of each UAV node. Based on the total control force vector, action control commands are generated for each of the UAV nodes.

[0010] Another embodiment of the present invention provides a drone swarm control system based on a collaborative emotion decision network, comprising: The fuzzing processing module is used to fuzzify the fused feature vector obtained by fusing multi-source sensing data to obtain a feature matrix containing multiple fuzzy features. The target weight determination module is used to determine the target weights connecting each fuzzy feature and each preset control action in the dual decision-making channels of the collaborative emotion decision-making network. The preliminary action decision vector determination module is used to determine the tendency and suppression of each preset control action based on the feature matrix and the target weight, and to determine the preliminary action decision vector of each UAV node based on the tendency and the suppression. The collaborative control command determination module is used to perform consensus fusion on the target tracking information and preliminary action decision vectors of each UAV node to obtain the collaborative control command of the UAV cluster. The motion control command execution module is used to convert the collaborative control command into motion control commands for each of the UAV nodes, and to control each of the UAV nodes to execute the motion control commands.

[0011] As one preferred embodiment, the preliminary action decision vector determination module includes: An initial decision value determination unit is used to determine an initial decision value based on the difference between the degree of tendency and the amount of inhibition. The target weight adjustment unit is used to adjust the target weight when the effectiveness evaluation value of the initial decision value is less than the expected evaluation value, and return to the step of determining the initial decision value based on the difference between the degree of tendency and the amount of inhibition, until the effectiveness evaluation value is greater than or equal to the expected evaluation value; The vectorization processing unit is used to vectorize the initial decision values ​​of each preset control action when the performance evaluation value of the initial decision value is greater than or equal to the expected evaluation value, so as to obtain the preliminary action decision vector of each UAV node.

[0012] As one preferred embodiment, the cooperative control command determination module includes: The target confidence determination unit is used to perform weighted fusion processing on the normalized values ​​determined based on target tracking information to obtain the target confidence of each UAV node; The consensus fusion unit is used to perform consensus fusion on the target confidence and preliminary action decision vector of each UAV node to obtain the consensus support of each preset control action. The collaborative control command determination unit is used to filter out target actions from each preset control action based on the consensus support, and generate collaborative control commands for the UAV swarm based on the target actions.

[0013] As one preferred embodiment, the motion control command execution module includes: The composite weight adjustment unit is used to adjust the composite weight of each UAV node on each preset motion component based on the cooperative control command. The total control force vector determination unit is used to perform weighted fusion of each preset motion component force based on the adjusted synthetic weights to obtain the total control force vector of each UAV node. The motion control command generation unit is used to generate motion control commands for each of the UAV nodes based on the total control force vector.

[0014] This invention provides a method and system for controlling a drone swarm based on a collaborative emotion decision network. By fuzzifying the fused feature vector, a feature matrix containing multiple fuzzy features is obtained. This feature matrix is ​​then input into the collaborative emotion decision network deployed on each drone node. Using the target weights of the dual decision channels of the collaborative emotion decision network and the fuzzy feature matrix, the tendency and suppression of each preset control action are determined. These tendencies and suppressions are used to determine the preliminary action decision vector for each drone. Through consensus fusion of the target tracking information and preliminary action decision vectors of each drone, the collaborative control action commands for the entire drone swarm are obtained. During execution by each drone, the collaborative control action commands are transformed into individual drone action control commands, and each drone is controlled to execute the corresponding action control commands. This invention, through the collaboration of the dual decision channels of the collaborative emotion decision network, maps the perceptual features of each drone to action decisions, and then performs consensus fusion and transformation on the action decisions of each drone to obtain the drone swarm control commands and the corresponding execution action commands for each drone, thereby improving the overall collaborative control accuracy of the drone swarm. Attached Figure Description

[0015] Figure 1 This is one of the flowcharts of the UAV swarm control method based on collaborative emotion decision network provided by the present invention; Figure 2 This is the second flowchart of the UAV swarm control method based on collaborative emotion decision network provided by the present invention; Figure 3 This is a schematic diagram of the structure of the UAV swarm control system based on collaborative emotion decision-making network provided by the present invention.

[0016] Figure label: Among them, 301 is the fuzzification processing module; 302 is the target weight determination module; 303 is the preliminary action decision vector determination module; 304 is the collaborative control instruction determination module; and 305 is the action control instruction execution module. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0018] In the description of this application, the terms "first," "second," "third," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined with "first," "second," "third," etc., may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0019] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; they can refer to mechanical connections or electrical connections; they can refer to direct connections or indirect connections through an intermediate medium; and they can refer to the internal communication between two components. The terms "vertical," "horizontal," "left," "right," "upper," "lower," and similar expressions used herein are for illustrative purposes only and do not indicate or imply that the device or component referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as limiting the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0020] In the description of this application, it should be noted that, unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in this specification is for the purpose of describing specific embodiments only and is not intended to limit the invention. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0021] See Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the UAV swarm control method based on a collaborative emotion decision network provided by the present invention, as shown below. Figure 1 As shown, this embodiment includes steps 100 to 500, and the specific steps are as follows: Step 100: The fusion feature vector obtained by fusing multi-source sensing data is fuzzified to obtain a feature matrix containing multiple fuzzy features; In terms of data perception, multi-source perception data is acquired through sensors integrated into the UAVs. After feature extraction and fusion of the preprocessed multi-source perception data, a fused feature vector containing the target's motion features, signal features, image features, and behavioral context features is obtained. The fused feature vectors of each UAV are input into the Collaborative Emotional Decision Network (CEDN) deployed locally on each UAV. The CEDN performs fuzzification processing on the fused feature vectors to obtain a feature matrix containing multiple fuzzy features.

[0022] Step 200: Determine the target weights connecting each fuzzy feature and each preset control action in the dual decision-making channels of the collaborative emotion decision-making network; Specifically, the collaborative affective decision-making network has a dual-decision-channel mechanism of intuitive and rational channels, where the channel weight of the intuitive channel is... The target weight connecting the kj-th fuzzy feature and the m-th preset control action; the channel weight of the rational channel. This represents the target weight connecting the kj-th fuzzy feature and the m-th preset control action. (The target weight...) and During the learning and updating process, the reward index It played a key role. The calculation is as follows: As shown, where, The value is 0 or 1, representing the assessment result of whether the objective has been effectively terminated. The energy consumed in this control action; The time from identification to taking effective action; This is an estimate of the risk of collateral damage. , , and These are the corresponding weighting coefficients. This invention improves the balance of action decision-making through the reward index.

[0023] Step 300: Based on the feature matrix and the target weight, determine the tendency and suppression of each preset control action; based on the tendency and suppression, determine the preliminary action decision vector of each UAV node. Specifically, after inputting the feature matrix and updated target weights into the intuition channel, the output of the intuition channel is the original impulse intensity of each UAV node for each preset control action. By comparing the original impulse intensity with the actual reward, the target weights of the intuition channel are adjusted to update the impulse intensity. The actual reward is derived from the performance evaluation after executing each preset control action. Through the target weight update, the intuition channel finally outputs the tendency of each preset control action.

[0024] After inputting the feature matrix and updated target weights into the rational channel, the output of the rational channel is the original inhibition value of each UAV node for each preset control action. The difference between the original impulse intensity and the original inhibition value is used as the initial decision value. At the same time, the effectiveness of the initial decision value is evaluated to obtain the reward of the initial decision value. The reward of the initial decision value is compared with the expected reward of the rational channel for the initial decision value. The target weight of the rational channel is adjusted according to the comparison result to update the inhibition value of each UAV node for each preset control action. Through the target weight update, the rational channel finally outputs the inhibition amount of each preset control action.

[0025] After obtaining the tendency and inhibition of each UAV node for each preset control action, the preliminary action decision vector for all m preset control actions is obtained by synthesizing the tendency and inhibition through decision analysis.

[0026] Step 400: Consensus fusion is performed on the target tracking information and preliminary action decision vectors of each UAV node to obtain the collaborative control command of the UAV cluster; Specifically, after obtaining the preliminary action decision vectors of each UAV node, the confidence level of each UAV node is calculated using its target tracking information. The confidence level reflects the importance of various target tracking indicators in evaluating the reliability of the UAV node's decision-making. By collecting the preliminary action decision vectors and confidence levels of all UAV nodes and performing weighted consensus fusion on these vectors and levels, a consensus support level at the UAV swarm level is obtained. Finally, by selecting several actions with higher consensus support from among the preset control actions, collaborative control commands for the UAV swarm are generated.

[0027] Step 500: Convert the collaborative control command into action control commands for each of the UAV nodes, and control each of the UAV nodes to execute the action control commands.

[0028] Specifically, the collaborative control commands of the UAV swarm are transformed into specific executable motion control commands for each UAV. Through the control actions in the collaborative control commands, the composite weights of various components in the underlying motion control of each UAV node are dynamically adjusted to obtain the total control force vector of each UAV node. Finally, the motion control commands of each UAV node are generated based on the total control force vector, thereby controlling each UAV node to execute the motion control commands.

[0029] This embodiment fuzzifies the fused feature vector to obtain a feature matrix containing multiple fuzzy features. This feature matrix is ​​then input into a collaborative emotion decision network deployed across all UAV nodes. Using the target weights and fuzzy feature matrix from the dual decision channels of the collaborative emotion decision network, the tendency and suppression of each preset control action are determined. These tendencies and suppressions are used to determine the initial action decision vector for each UAV. Through consensus fusion of the target tracking information and the initial action decision vector of each UAV, the collaborative control action command for the entire UAV swarm is obtained. During execution by each UAV, the collaborative control action command is transformed into the action control command for each UAV, and each UAV is controlled to execute the corresponding action control command. This invention, through the collaboration of the dual decision channels of the collaborative emotion decision network, maps the perceptual features of each UAV to action decisions, and then performs consensus fusion and transformation on the action decisions of each UAV to obtain the UAV swarm control command and the corresponding execution action commands for each UAV, thus improving the coordination of UAV swarm control.

[0030] In another embodiment of the UAV swarm control method based on a collaborative emotion decision network provided by the present invention, step 100 specifically includes: Step 110: Based on each preset fuzzy semantic representation, determine the membership degree of each feature in the fused feature vector to each preset fuzzy semantic representation; Step 120: Process the membership degree to obtain a feature matrix containing multiple fuzzy features.

[0031] Specifically, fuzzification transforms the fused feature vector into robust fuzzy semantic concepts (i.e., the preset fuzzy semantic representation in this embodiment). For example, speed feature values ​​are transformed into membership degrees of fuzzy semantic representations such as low speed, medium speed, and high speed. The specific process of obtaining a feature matrix containing multiple fuzzy features through CEDAN is as follows: For the k-th feature in the fused feature vector... Its membership degree to the j-th fuzzy semantic representation The Gaussian membership function, as shown in Formula 1, is used to calculate, where, The center of the j-th fuzzy semantic representation; Let be the width of the j-th fuzzy semantic representation.

[0032] (1) After calculating all membership degrees using Formula 1, the membership degrees are processed to obtain a feature matrix containing multiple fuzzy features.

[0033] This embodiment transforms precise feature values ​​into robust fuzzy semantic representations through feature fuzzification.

[0034] See Figure 2 , Figure 2 This is a flowchart illustrating another embodiment of the UAV swarm control method based on a collaborative emotion decision-making network provided by the present invention, wherein the aforementioned dual decision-making channels include an intuitive channel and a rational channel, such as... Figure 2 As shown, this embodiment includes steps 310 to 360, and the specific steps are as follows: Step 310: Obtain the effectiveness evaluation results, original impulse intensity, and original inhibition value for each preset control action; Step 320: Based on the learning rate of the intuitive channel, the feature matrix, the performance evaluation results, and the original impulse intensity, determine the degree of tendency for each preset control action; Step 330: Based on the learning rate of the rational channel, the feature matrix, the original impulse intensity, and the original inhibition value, determine the inhibition amount of each preset control action; Step 340: Determine the initial decision value based on the difference between the degree of tendency and the amount of inhibition; Step 350: If the effectiveness evaluation value of the initial decision value is less than the expected evaluation value, adjust the target weight and return to the step of determining the initial decision value based on the difference between the degree of tendency and the amount of inhibition, until the effectiveness evaluation value is greater than or equal to the expected evaluation value; Step 360: If the performance evaluation value of the initial decision value is greater than or equal to the expected evaluation value, the initial decision value of each preset control action is vectorized to obtain the preliminary action decision vector of each UAV node.

[0035] Specifically, CEDN employs a dual-channel decision-making mechanism, employing both intuitive and rational channels. The intuitive channel simulates rapid, experience-based responses. The output of the intuitive channel... The calculation is shown in Formula 2. This represents the initial impulse intensity for the m-th preset control action; This represents the kj-th fuzzy feature in the feature matrix; It is the weight connecting the kj-th fuzzy feature and the m-th preset control action in the intuitive channel (i.e., one of the target weights in this embodiment). The learning rules are shown in Formula 3. The learning rate for the weights of the intuitive channel; The actual reward comes from the performance evaluation results after the system executes a preset control action m. When the actual reward... Higher than At that time, weight That's when it will increase.

[0036] (2) (3) The rational channel simulates prudent risk assessment; the output of the rational channel. Let represent the original inhibition value for the m-th action, where It is the weight connecting the kj-th fuzzy feature and the m-th preset control action in the rational channel (i.e., one of the target weights in this embodiment). The learning rules are quite flexible and used to correct errors, as shown in Formula 5. The learning rate for the weights of the intuitive channel; , which is the initial decision value in this embodiment; This represents the expected reward from the rational channel for the initial decision value, i.e., the expected evaluation value in this embodiment. If the initial decision value... The rewards brought Below expectations Then increase the suppression weight. Conversely, it decreases.

[0037] (4) (5) (6) For all m preset control actions, CEDN outputs a decision vector. This refers to the preliminary action decision vector in this embodiment, as shown in Formula 6. The preliminary action decision vector represents the comprehensive evaluation of each UAV for each preset control action.

[0038] This embodiment uses a dual-decision-channel mechanism to map multi-source perception features into action strategies in real time, and can autonomously optimize and learn channel weights based on environmental feedback.

[0039] In another embodiment of the UAV swarm control method based on a collaborative emotion decision network provided by the present invention, step 400 specifically includes: Step 410: Perform weighted fusion processing on the normalized values ​​determined based on target tracking information to obtain the target confidence of each UAV node; Step 420: Perform consensus fusion on the target confidence and preliminary action decision vector of each UAV node to obtain the consensus support of each preset control action; Step 430: Based on the consensus support, select the target action from each preset control action, and generate the collaborative control command for the UAV swarm according to the target action.

[0040] Specifically, the initial action decision vector of a single UAV node's CEDN may be limited by a local perspective. Cluster optimization can improve the robustness and global optimality of collaborative decision-making. This involves obtaining the initial action decision vector for each UAV node i. At the same time, calculate the confidence level. Confidence level It is a scalar in the interval [0,1], generated by a weighted linear combination, as shown in Formula 7. Wherein, the index... For the continuity of UAV node i's tracking of the target, such as the percentage of consecutive tracking frames; indicators The sensor signal-to-noise ratio of UAV node i; metric This indicates the health status of drone node i, such as battery level and communication quality.

[0041] (7) in, This means normalizing the corresponding indicators to the [0,1] interval; , and For example, the confidence factor weighting coefficients for each item. The confidence factor weighting coefficient is used to reflect the importance of each indicator in assessing the reliability of the initial action decision vector of the UAV node.

[0042] (8) For each drone node Perform weighted consensus fusion and calculate the consensus support of the m-th preset control action at the cluster level. As shown in Formula 8, where, The initial decision value for drone node i The system selects one or more actions with the highest consensus support among the preset control actions, which are the target actions in this embodiment. Based on one or more target actions, it generates collaborative control instructions for the drone swarm.

[0043] This embodiment improves the coordination of drone swarm control by filtering out individual decisions that cause problems through a consensus decision-making mechanism built on the confidence of drone nodes.

[0044] In another embodiment of the UAV swarm control method based on a collaborative emotion decision network provided by the present invention, step 500 specifically includes: Step 510: Based on the cooperative control command, adjust the composite weight of each UAV node on each preset motion component; Step 520: Based on the adjusted synthesis weights, the preset motion components are weighted and fused to obtain the total control force vector of each UAV node; Step 530: Based on the total control force vector, generate motion control commands for each of the UAV nodes.

[0045] Specifically, the cooperative control commands of the drone swarm obtained above need to be converted into executable motion control commands for each drone node. Based on the cooperative actions in the cooperative control commands, the composite weights of various components in the underlying motion control of each drone node are dynamically adjusted, as shown in Formula 9.

[0046] (9) in, This represents the total control force vector for each UAV node. The gravitational weight of the target determines the strength of the drone's intention to fly towards the target location; The gravitational vector of the target is the attractive force pointing towards the target's expected location; To consolidate the weight of the cluster and control the intensity of the willingness to move towards the center of the drone cluster; The cluster cohesion vector is the attractive force pointing towards the average position of the local neighboring drones (cluster center); Individual separation weights control the strength of the willingness to maintain a safe distance from neighboring drones (or obstacles); The individual separation force vector represents the repulsive force from neighboring drones or obstacles that are too close. For velocity alignment weights, control the strength of the willingness to keep the average velocity direction of neighboring drones in line with the target. The force vector for aligning velocity is the turning force that causes the direction of one's own velocity to align with the direction of the average velocity of its neighbor; The obstacle rejection weight controls the intensity of the willingness to stay away from environmental obstacles; It is the vector of the repulsive force of the obstacle, representing the repulsive force away from the known environmental obstacle.

[0047] The weights in Formula 9 are related to the motion control command. If the motion control command is a long-distance interference, then... The reduction in size means that drones are focusing more on maintaining swarm formation. , and (Raise) and linger at a safe distance.

[0048] This embodiment improves the flexibility of drone swarm control by adjusting the composite weights of each drone node on each preset motion component force to generate motion control commands for each drone node.

[0049] The UAV swarm control system based on collaborative emotion decision network provided by the present invention will be described below. The UAV swarm control system based on collaborative emotion decision network described below can be referred to in correspondence with the UAV swarm control method based on collaborative emotion decision network described above.

[0050] Please refer to Figure 3 The present invention also provides a drone swarm control system based on a collaborative emotion decision-making network, comprising: The fuzzing processing module 301 is used to fuzzify the fused feature vector obtained by fusing multi-source sensing data to obtain a feature matrix containing multiple fuzzy features. The target weight determination module 302 is used to determine the target weights connecting each fuzzy feature and each preset control action in the dual decision-making channels of the collaborative emotion decision-making network. The preliminary action decision vector determination module 303 is used to determine the tendency and suppression of each preset control action based on the feature matrix and the target weight, and to determine the preliminary action decision vector of each UAV node based on the tendency and the suppression. The collaborative control command determination module 304 is used to perform consensus fusion on the target tracking information and preliminary action decision vectors of each UAV node to obtain the collaborative control command of the UAV cluster. The motion control command execution module 305 is used to convert the collaborative control command into motion control commands for each of the UAV nodes, and control each of the UAV nodes to execute the motion control commands.

[0051] Optionally, the preliminary action decision vector determination module includes: An initial decision value determination unit is used to determine an initial decision value based on the difference between the degree of tendency and the amount of inhibition. The target weight adjustment unit is used to adjust the target weight when the effectiveness evaluation value of the initial decision value is less than the expected evaluation value, and return to the step of determining the initial decision value based on the difference between the degree of tendency and the amount of inhibition, until the effectiveness evaluation value is greater than or equal to the expected evaluation value; The vectorization processing unit is used to vectorize the initial decision values ​​of each preset control action when the performance evaluation value of the initial decision value is greater than or equal to the expected evaluation value, so as to obtain the preliminary action decision vector of each UAV node.

[0052] Optionally, the cooperative control command determination module includes: The target confidence determination unit is used to perform weighted fusion processing on the normalized values ​​determined based on target tracking information to obtain the target confidence of each UAV node; The consensus fusion unit is used to perform consensus fusion on the target confidence and preliminary action decision vector of each UAV node to obtain the consensus support of each preset control action. The collaborative control command determination unit is used to filter out target actions from each preset control action based on the consensus support, and generate collaborative control commands for the UAV swarm based on the target actions.

[0053] Optionally, the motion control command execution module includes: The composite weight adjustment unit is used to adjust the composite weight of each UAV node on each preset motion component based on the cooperative control command. The total control force vector determination unit is used to perform weighted fusion of each preset motion component force based on the adjusted synthetic weights to obtain the total control force vector of each UAV node. The motion control command generation unit is used to generate motion control commands for each of the UAV nodes based on the total control force vector.

[0054] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method for controlling a swarm of unmanned aerial vehicles (UAVs) based on a collaborative emotion-based decision-making network, characterized in that, include: The fusion feature vector obtained by fusing multi-source sensing data is fuzzified to obtain a feature matrix containing multiple fuzzy features. Determine the target weights connecting each fuzzy feature and each preset control action in the dual decision-making channels of the collaborative emotion decision-making network; Based on the feature matrix and the target weight, the tendency and suppression of each preset control action are determined, and based on the tendency and suppression, the preliminary action decision vector of each UAV node is determined. Consensus fusion is performed on the target tracking information and preliminary action decision vectors of each UAV node to obtain the collaborative control command of the UAV cluster; The collaborative control commands are converted into action control commands for each of the UAV nodes, and each UAV node is controlled to execute the action control commands.

2. The UAV swarm control method based on collaborative emotion decision-making network as described in claim 1, characterized in that, The process of blurring the fused feature vector obtained from the fusion of multi-source sensing data to obtain a feature matrix containing multiple blurred features includes: Based on each preset fuzzy semantic representation, determine the membership degree of each feature in the fused feature vector to each preset fuzzy semantic representation; The membership degree is processed to obtain a feature matrix containing multiple fuzzy features.

3. The UAV swarm control method based on collaborative emotion decision-making network as described in claim 1, characterized in that, The dual decision-making channels include an intuitive channel and a rational channel; The determination of the tendency and inhibition of each preset control action based on the feature matrix and the target weights includes: Obtain the effectiveness evaluation results, original impulse intensity, and original inhibition value for each preset control action; Based on the learning rate of the intuitive channel, the feature matrix, the performance evaluation results, and the original impulse intensity, the degree of tendency for each preset control action is determined. Based on the learning rate of the rational channel, the feature matrix, the original impulse intensity, and the original inhibition value, the inhibition amount of each preset control action is determined.

4. The UAV swarm control method based on collaborative emotion decision-making network as described in claim 1, characterized in that, The determination of the preliminary action decision vector for each UAV node based on the degree of tendency and the amount of inhibition includes: The initial decision value is determined based on the difference between the degree of tendency and the amount of inhibition. If the effectiveness assessment value of the initial decision value is less than the expected assessment value, the target weight is adjusted, and the step of determining the initial decision value based on the difference between the degree of propensity and the amount of inhibition is returned until the effectiveness assessment value is greater than or equal to the expected assessment value. If the performance evaluation value of the initial decision value is greater than or equal to the expected evaluation value, the initial decision value of each preset control action is vectorized to obtain the preliminary action decision vector of each UAV node.

5. The UAV swarm control method based on collaborative emotion decision-making network as described in claim 1, characterized in that, The consensus fusion of target tracking information and preliminary action decision vectors of each UAV node to obtain the collaborative control commands for the UAV cluster includes: The normalized values ​​determined based on target tracking information are weighted and fused to obtain the target confidence scores of each UAV node. Consensus fusion is performed on the target confidence and preliminary action decision vector of each UAV node to obtain the consensus support of each preset control action; Based on the consensus support, target actions are selected from each preset control action, and collaborative control instructions for the UAV swarm are generated according to the target actions.

6. The UAV swarm control method based on collaborative emotion decision-making network as described in claim 1, characterized in that, The step of converting the collaborative control command into action control commands for each of the UAV nodes includes: Based on the cooperative control command, the composite weights of each UAV node on each preset motion component are adjusted; Based on the adjusted synthesis weights, the preset motion components are weighted and fused to obtain the total control force vector of each UAV node. Based on the total control force vector, action control commands are generated for each of the UAV nodes.

7. A drone swarm control system based on a collaborative emotion-based decision-making network, characterized in that, include: The fuzzing processing module is used to fuzzify the fused feature vector obtained by fusing multi-source sensing data to obtain a feature matrix containing multiple fuzzy features. The target weight determination module is used to determine the target weights connecting each fuzzy feature and each preset control action in the dual decision-making channels of the collaborative emotion decision-making network. The preliminary action decision vector determination module is used to determine the tendency and suppression of each preset control action based on the feature matrix and the target weight, and to determine the preliminary action decision vector of each UAV node based on the tendency and the suppression. The collaborative control command determination module is used to perform consensus fusion on the target tracking information and preliminary action decision vectors of each UAV node to obtain the collaborative control command of the UAV cluster. The motion control command execution module is used to convert the collaborative control command into motion control commands for each of the UAV nodes, and to control each of the UAV nodes to execute the motion control commands.

8. The UAV swarm control system based on a collaborative emotion-based decision-making network as described in claim 7, characterized in that, The preliminary action decision vector determination module includes: An initial decision value determination unit is used to determine an initial decision value based on the difference between the degree of tendency and the amount of inhibition. The target weight adjustment unit is used to adjust the target weight when the effectiveness evaluation value of the initial decision value is less than the expected evaluation value, and return to the step of determining the initial decision value based on the difference between the degree of tendency and the amount of inhibition, until the effectiveness evaluation value is greater than or equal to the expected evaluation value; The vectorization processing unit is used to vectorize the initial decision values ​​of each preset control action when the performance evaluation value of the initial decision value is greater than or equal to the expected evaluation value, so as to obtain the preliminary action decision vector of each UAV node.

9. The UAV swarm control system based on a collaborative emotion-based decision-making network as described in claim 7, characterized in that, The cooperative control command determination module includes: The target confidence determination unit is used to perform weighted fusion processing on the normalized values ​​determined based on target tracking information to obtain the target confidence of each UAV node; The consensus fusion unit is used to perform consensus fusion on the target confidence and preliminary action decision vector of each UAV node to obtain the consensus support of each preset control action. The collaborative control command determination unit is used to filter out target actions from each preset control action based on the consensus support, and generate collaborative control commands for the UAV swarm based on the target actions.

10. The UAV swarm control system based on a collaborative emotion-based decision-making network as described in claim 7, characterized in that, The motion control command execution module includes: The composite weight adjustment unit is used to adjust the composite weight of each UAV node on each preset motion component based on the cooperative control command. The total control force vector determination unit is used to perform weighted fusion of each preset motion component force based on the adjusted synthetic weights to obtain the total control force vector of each UAV node. The motion control command generation unit is used to generate motion control commands for each of the UAV nodes based on the total control force vector.

Images