Distributed digital twin architecture cluster unmanned aerial vehicle confrontation training collaborative optimization method, system and device, and storage medium

Abstract

The invention relates to the technical field of cluster unmanned aerial vehicle confrontation training, provides a distributed digital twin architecture cluster unmanned aerial vehicle confrontation training collaborative optimization method, system and device, and a storage medium, and solves the problem of collaborative failure. The method comprises the following steps: correcting control parameters of each unmanned aerial vehicle individual according to abnormal fluctuation amplitude and space gradient change information; constructing an adversarial training scene through a digital twinning model of a distributed digital twinning architecture; the corresponding corrected control parameters are loaded, and actual response results of the unmanned aerial vehicle individuals in the confrontation training scene are recorded; comparing the deviation between the actual response result and the expected behavior of the digital twin model, and reconstructing the correction logic of the control parameter when the deviation exceeds a dynamic fault-tolerant threshold value; and cooperatively generating an anti-interference group flight strategy based on the corrected control parameters and actual response results of the plurality of unmanned aerial vehicle individuals exchanged by the distributed nodes. According to the invention, the anti-interference capability and group cooperative flight performance of the cluster unmanned aerial vehicle are improved.

Description

Technical Field

[0001] The present application relates to the technical field of swarm drone adversarial training, and in particular to a swarm drone adversarial training collaborative optimization method and system based on a distributed digital twin architecture. Background Art

[0002] In the swarm drone confrontation training scenario, the technical requirements are mainly reflected in three dimensions: first, it is necessary to integrate the drone power system status data (such as the speed fluctuation amplitude change rate, voltage anomaly offset) and fluid dynamic environment parameters (such as the air pressure change rate vector, temperature field contour density) in real time to construct a confrontation scenario that can dynamically reflect the changes in spatial gradients; second, the control parameter correction mechanism is required to be fault-tolerant, and when the actual response result exceeds the expected deviation from the digital twin model, the correction logic can be automatically reconstructed; third, it is necessary to realize data interaction and collaborative decision-making of distributed nodes, generate anti-interference swarm flight strategies while ensuring communication efficiency, and meet the real-time confrontation training needs of large-scale clusters.

[0003] A typical solution currently addressing these needs is a hybrid digital twin training architecture based on edge-cloud collaboration. This approach uses edge nodes to collect drone power and environmental data, pre-processes it, and uploads it to a cloud-based digital twin model. The cloud then centrally generates adversarial training scenarios and calculates control parameter adjustments, which are then distributed to each drone for execution. For example, a cluster adversarial simulation platform developed by a company utilizes this architecture, with the edge responsible for data filtering and the cloud utilizing a parallel computing engine to generate virtual adversarial rules, enabling simultaneous simulation of virtual and real scenarios.

[0004] However, this solution has three significant flaws: First, centralized processing in the cloud causes data transmission delays to increase exponentially with the growth of cluster size. When the number of drones is large, the control parameter correction lag time is long, which cannot meet the real-time requirements of highly dynamic confrontation scenarios; second, environmental modeling relies on global parameters in the cloud, which makes it difficult to adapt to local spatial gradient differences. For example, the virtual scene in the area where the air pressure change rate vector suddenly changes deviates greatly from the physical entity; third, the fault-tolerant mechanism relies on a single decision-making node in the cloud. Once communication is interrupted or computing fails, the cluster will lose its ability to coordinate and adjust, and the efficiency of generating anti-interference group strategies is low. Summary of the Invention

[0005] The present application provides a distributed digital twin architecture cluster drone adversarial training collaborative optimization method and system to solve the problems in the existing technology such as centralized processing leading to delays, difficulty in adapting global parameters to local differences, and single node failure leading to collaborative failure.

[0006] In the first aspect, the present application provides a distributed digital twin architecture cluster drone adversarial training collaborative optimization method, including: Obtain the power system status data and fluid dynamic environment parameters of multiple UAVs in the swarm; Correcting the control parameters of each individual drone based on the abnormal fluctuation amplitude of the power system state data and the spatial gradient change information of the fluid dynamic environment parameters; Through the digital twin model of the distributed digital twin architecture, adversarial training scenarios are constructed based on the fluctuation characteristics of the power system state data and the spatial gradient distribution of the fluid dynamic environment parameters; After loading the corresponding corrected control parameters into each of the drone individuals, the actual response results of the drone individuals in the confrontation training scenario are recorded; Comparing the deviation between the actual response result and the expected behavior of the digital twin model, when the deviation exceeds a dynamic fault tolerance threshold, triggering a fault tolerance adjustment mechanism to reconstruct the correction logic of the control parameters; Based on the modified control parameters of multiple UAV individuals exchanged by distributed nodes and the actual response results, an anti-interference swarm flight strategy is collaboratively generated.

[0007] Optionally, the digital twin model of the distributed digital twin architecture constructs an adversarial training scenario based on the fluctuation characteristics of the power system state data and the spatial gradient distribution of the fluid dynamic environment parameters, including: The feature extraction module in the digital twin model extracts corresponding fluctuation features from the power state parameters of each individual drone. These fluctuation features include the speed fluctuation amplitude change rate and the voltage anomaly offset. Simultaneously, the spatial gradient distribution is extracted from the fluid dynamic environment parameters of each individual drone. The spatial gradient distribution includes the pressure change rate vector and the temperature field contour line density. The first rule setting module in the digital twin model maps the speed fluctuation amplitude change rate to a maneuverability attenuation coefficient of the virtual entity, and generates a first boundary constraint rule for characterizing the collision volume of the virtual entity based on the maneuverability attenuation coefficient; The air pressure change rate vector is converted into a virtual aerodynamic disturbance direction by a second rule setting module in the digital twin model, and a second boundary constraint rule for characterizing the range of environmental disturbance is generated according to the virtual aerodynamic disturbance direction; Resetting a communication delay parameter based on the voltage anomaly offset by a first strategy generation module in the digital twin model to obtain an anti-interference communication strategy; The second strategy generation module in the digital twin model redistributes the virtual obstacle coordinates according to the density of the temperature field contour lines to obtain an environment-adaptive obstacle avoidance strategy; Through the fusion module in the digital twin model, an adversarial training scenario is constructed based on the first boundary constraint rule, the second boundary constraint rule, the anti-interference communication strategy and the environment adaptive obstacle avoidance strategy.

[0008] Optionally, mapping the speed fluctuation amplitude change rate to a maneuverability attenuation coefficient of the virtual entity through a first rule setting module in the digital twin model, and generating a first boundary constraint rule for characterizing the collision volume of the virtual entity according to the maneuverability attenuation coefficient, includes: The fluctuation amplitude change rate is divided into different levels according to the threshold value corresponding to the preset interval by the mapping unit in the first rule setting module, so as to obtain the interval corresponding to the speed fluctuation amplitude change rate; generating a propulsion force reduction degree value based on a calculation method corresponding to the interval; After mapping the propulsion force reduction degree value into grades, the values ​​are converted into a maneuverability attenuation coefficient of the virtual entity; determining, by a rule generating unit in a first rule setting module, a baseline collision volume parameter of the virtual entity in a normal state and an adjustment ratio corresponding to the baseline collision volume parameter based on the maneuverability attenuation coefficient, the baseline collision volume parameter including a baseline length, a baseline width, and a baseline height; Adjusting the reference length, the reference width, and the reference height respectively according to the corresponding adjustment ratios to obtain an adjusted length, an adjusted width, and an adjusted height; Converting the three-dimensional space defined by the adjusted length, the adjusted width, and the adjusted height into maximum offset values ​​along the X-axis, the Y-axis, and the Z-axis, wherein the maximum offset values ​​are insurmountable boundary values ​​when the virtual entity moves; Based on the boundary value, a first boundary constraint rule for characterizing the collision volume of the virtual entity is generated.

[0009] Optionally, the second rule setting module in the digital twin model converts the air pressure change rate vector into a virtual aerodynamic interference direction, and generates a second boundary constraint rule for characterizing the range of environmental disturbance according to the aerodynamic interference direction, including: Decomposing the direction information of the air pressure change in the air pressure change rate vector into components along the X-axis, Y-axis, and Z-axis directions by a conversion unit in the second rule setting module; Setting a corresponding interference weight coefficient for the component in each coordinate axis direction, wherein the interference weight coefficient is used to indicate the degree of influence of the component in the corresponding coordinate axis direction on the virtual aerodynamic interference direction; Converting the intensity information in the air pressure change rate vector into an intensity coefficient; Multiplying the intensity coefficient by the interference weight coefficient in each coordinate axis direction to obtain a weighted coefficient in each coordinate axis direction; Based on the weighting coefficients, weighted calculations are performed on the components of the directions of the coordinate axes to obtain weighted direction components; synthesizing the weighted directional components to obtain a virtual aerodynamic interference direction; Calculating, by a rule generation unit in a second rule setting module, an angular deviation between the virtual aerodynamic interference direction and a preset environmental disturbance action range reference direction; According to the angle deviation, the preset environmental disturbance action range is rotated and adjusted so that the central axis of the adjusted environmental disturbance action range is consistent with the virtual aerodynamic force interference direction, thereby obtaining boundary parameters of the adjusted environmental disturbance action range; The adjusted boundary parameters of the environmental disturbance effect range are integrated to generate a second boundary constraint rule for characterizing the environmental disturbance effect range.

[0010] Optionally, the modifying of the control parameters of each individual UAV according to the abnormal fluctuation amplitude of the power system state data and the spatial gradient change information of the fluid dynamic environment parameters includes: Extract the magnitude and frequency of abnormal fluctuations from the power system status data; Normalizing the magnitude of the abnormal fluctuation amplitude to obtain a corresponding basic value, converting the occurrence frequency into a correction coefficient according to a preset ratio, and generating a fluctuation characteristic value by multiplying the basic value by the correction coefficient; Analyzing the changes of the fluid dynamic environment parameters at different spatial gradients to obtain spatial gradient change information; Associating the fluctuation characteristic value with the spatial gradient change information to form a comprehensive information set; Based on the comprehensive information set, determining the control parameter adjustment direction and initial adjustment value of each individual drone; According to the preset control parameter adjustment restriction conditions, the initial adjustment value is adjusted to obtain a target adjustment value that meets the restriction conditions; According to the target adjustment value, the control parameters of each individual drone are modified according to the control parameter adjustment direction to obtain the modified control parameters of each individual drone.

[0011] Optionally, associating the fluctuation characteristic value with the spatial gradient change information to form a comprehensive information set includes: For each individual drone, the corresponding fluctuation characteristic value is matched with the value at the same position in the spatial gradient change information according to the spatial gradient position of the individual drone, and the corresponding relationship between each pair of matching data is recorded; Counting the occurrence frequency of the corresponding relationship of each pair of matching data in the corresponding drone individuals, and marking the corresponding relationship with an occurrence frequency greater than or equal to a preset frequency threshold as a strong association; Group the strongly correlated matching data according to the individual identifiers of each drone to obtain data groups; Counting the extreme values ​​and median values ​​of the fluctuation characteristic values ​​in the data group to form a characteristic interval, and simultaneously calculating the upper and lower limits and the mean value of the spatial gradient change information to form a numerical interval; A comprehensive information set is formed based on the individual identifiers of the drones, the characteristic intervals, and the numerical intervals.

[0012] Optionally, the modified control parameters of the multiple UAV individuals exchanged based on the distributed nodes and the actual response results are used to collaboratively generate an anti-interference swarm flight strategy, including: Based on distributed nodes, each drone is controlled to send its actual position information, revised control parameters and actual response results to other drones in the cluster at preset time intervals, and at the same time receive the actual position information, revised control parameters and actual response results sent by other drones to form a cluster dataset containing the actual position information, revised control parameters and actual response results of all drones; Extracting the adjustment amplitude and adjustment frequency of the corrected control parameters of each individual drone from the cluster data set; Comparing the deviation between the adjustment amplitude and the adjustment frequency and the actual response result, and determining the maneuvering requirements of each individual UAV according to the deviation, wherein different maneuvering requirements correspond to different urgency levels; Extract the actual position information of each drone individual from the cluster data set, and calculate the distance and relative position change rate between adjacent drone individuals; Determining a cluster formation constraint based on the distances and relative position change rates between the adjacent individual drones, wherein the distances between the adjacent individual drones are inversely proportional to the strength of the corresponding cluster formation constraint; Based on the urgency of each individual drone's maneuvering needs and the strength of the cluster formation constraints, the adjustment plans of all individual drones are integrated into an anti-interference swarm flight strategy.

[0013] In the second aspect, the present application provides a distributed digital twin architecture cluster drone adversarial training collaborative optimization system, including: An acquisition module is used to obtain the power system status data and fluid dynamic environment parameters of multiple individual drones in the cluster; A correction module, configured to correct the control parameters of each individual UAV based on the abnormal fluctuation amplitude of the power system state data and the spatial gradient change information of the fluid dynamic environment parameters; A construction module for constructing adversarial training scenarios based on the fluctuation characteristics of power system state data and the spatial gradient distribution of fluid dynamic environment parameters through the digital twin model of the distributed digital twin architecture; A recording module is used to load the corresponding corrected control parameters into each of the drone individuals and record the actual response results of the drone individuals in the confrontation training scenario; An adjustment module is configured to compare the deviation between the actual response result and the expected behavior of the digital twin model, and when the deviation exceeds a dynamic fault tolerance threshold, trigger a fault tolerance adjustment mechanism to reconstruct the correction logic of the control parameters; A generation module is used to collaboratively generate an anti-interference swarm flight strategy based on the corrected control parameters of multiple drone individuals exchanged by distributed nodes and the actual response results.

[0014] In a third aspect, the present application provides a computing device comprising a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a cluster drone adversarial training collaborative optimization method with a distributed digital twin architecture as described in any one of the first aspects.

[0015] In a fourth aspect, the present application provides a computer storage medium storing a computer program. When the computer program is executed by a computer, it implements a cluster drone adversarial training collaborative optimization method with a distributed digital twin architecture as described in any one of the first aspects.

[0016] This application provides a collaborative optimization method for adversarial training of swarm drones based on a distributed digital twin architecture, including: obtaining power system state data and fluid dynamic environment parameters of multiple individual drones in the swarm; correcting the control parameters of each individual drone based on the abnormal fluctuation amplitude of the power system state data and the spatial gradient change information of the fluid dynamic environment parameters; constructing an adversarial training scenario based on the fluctuation characteristics of the power system state data and the spatial gradient distribution of the fluid dynamic environment parameters through a digital twin model of the distributed digital twin architecture; loading the corresponding corrected control parameters into each individual drone, and recording the actual response results of the individual drone in the adversarial training scenario; By comparing the deviation between the actual response results and the expected behavior of the digital twin model, when the deviation exceeds the dynamic fault tolerance threshold, the fault tolerance adjustment mechanism is triggered to reconstruct the correction logic of the control parameters; based on the corrected control parameters and actual response results of multiple drone individuals exchanged through distributed nodes, an anti-interference swarm flight strategy is collaboratively generated.

[0017] This application has the following advantages: By obtaining the power system status data and fluid dynamic environment parameters of multiple individual drones in the cluster, basic data support can be provided for subsequent control parameter correction and confrontation scenario construction; by correcting the control parameters of each individual drone based on the abnormal fluctuation amplitude of the power system status data and the spatial gradient change information of the fluid dynamic environment parameters, the control parameters can be adapted to the actual operating status and environmental changes; by constructing confrontation training scenarios through the digital twin model of the distributed digital twin architecture, a high-fidelity virtual confrontation environment can be generated based on real data; by loading the corrected control parameters and recording the actual response results of individual drones in the confrontation training scenarios, a measured basis can be provided for fault-tolerant adjustment and group strategy generation; by comparing the deviation between the actual response results and the expected behavior of the digital twin model and triggering the fault-tolerant adjustment mechanism, the correction logic can be dynamically reconstructed when the deviation exceeds the limit, thereby improving the system fault tolerance; by collaboratively generating anti-interference group flight strategies based on distributed node exchange of data, the overall anti-interference collaborative optimization of the cluster can be achieved.

[0018] Furthermore, the digital twin model first extracts fluctuation characteristics such as the speed fluctuation amplitude change rate and the voltage anomaly offset from the power state parameters, and extracts spatial gradient distributions such as the air pressure change rate vector and the temperature field contour line density from the fluid dynamic environment parameters. The first rule setting module then maps the speed fluctuation amplitude change rate to the virtual entity's maneuverability attenuation coefficient and generates a first boundary constraint rule. This module divides the fluctuation amplitude change rate into grades to generate a propulsion force reduction degree value, which is then converted into a maneuverability attenuation coefficient. Based on this, the virtual entity's baseline collision volume parameters are adjusted and boundary values ​​are generated to determine the first rule. By refining the feature extraction and graded mapping rules, the boundary constraints of the virtual entity's collision volume can be made more consistent with the actual maneuverability of the physical drone, improving the authenticity of the virtual entity in the adversarial training scenario and the accuracy of the scenario construction.

[0019] These and other aspects of the present application will become more readily apparent from the description of the following embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0021] Figure 1 A flowchart of a distributed digital twin architecture cluster drone adversarial training collaborative optimization method provided in an embodiment of the present application; Figure 2 A schematic diagram of the structure of a distributed digital twin architecture swarm drone adversarial training collaborative optimization system provided in an embodiment of the present application; Figure 3 A schematic diagram of the structure of a computing device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0022] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0023] In some of the processes described in the specification and claims of this application and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this document or may be executed in parallel. The serial numbers of the operations, such as 11, 12, etc., are only used to distinguish between different operations, and the serial numbers themselves do not represent any order of execution. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to being different types.

[0024] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0025] Figure 1 A flowchart of a distributed digital twin architecture cluster drone adversarial training collaborative optimization method provided in an embodiment of the present application is shown in FIG. Figure 1 As shown, the method includes: S11. Obtain power system status data and fluid dynamic environment parameters of multiple individual drones in the cluster.

[0026] Among them, the power system status data is the operating data of the UAV's power part, including changes in motor speed and power supply voltage, etc., reflecting the working status of the power system; the fluid dynamic environment parameters are the data of the fluid environment such as the air around the UAV, including the air pressure and temperature distribution at different positions, reflecting the characteristics of the environmental fluid; the cluster is a whole composed of multiple UAVs; the individual UAV is a single UAV in the cluster; through this step, a collection of power and environmental data of all UAVs is finally generated.

[0027] In an embodiment of the present application, each drone in a swarm is equipped with equipment that can measure its own powertrain operating conditions (e.g., motor speed and voltage) and the surrounding environment (e.g., air pressure and temperature). These devices record data in real time, and each drone sends this data to the cluster's distributed processing node, which aggregates the data from all drones. For example, in a swarm of 20 drones, each drone uses instruments to record changes in motor speed, voltage fluctuations, and the surrounding air pressure and temperature. This data is then sent to the processing node, which compiles a collection of data from all drones.

[0028] S12. Correct the control parameters of each UAV based on the abnormal fluctuation amplitude of the power system status data and the spatial gradient change information of the fluid dynamic environment parameters.

[0029] Among them, the abnormal fluctuation amplitude is the size of the change in the power system status data that exceeds the normal range, such as the degree of sudden change in motor speed; the spatial gradient change information is the change of fluid dynamic environment parameters at different locations, such as the spatial change trend of air pressure; the control parameters are the parameters that control the flight of the UAV, such as speed and steering related parameters; through this step, the corrected control parameters of each UAV are finally generated.

[0030] In this embodiment, the magnitude of abnormal fluctuations is determined from the power data, and the spatial variation trends of environmental parameters are analyzed. These two pieces of information are combined to determine the adjustment direction and initial adjustment value of the control parameters. These initial values ​​are then modified based on the actual limitations of the drone (such as the maximum motor speed) to obtain the corrected control parameters. For example, if a drone experiences abnormally large fluctuations in motor speed and significant spatial variations in surrounding air pressure, a speed reduction is determined. The initial adjustment value is 5, but this is changed to 3 to account for motor limitations, ultimately slowing the speed parameter by 3.

[0031] S13. Through the digital twin model of the distributed digital twin architecture, an adversarial training scenario is constructed based on the fluctuation characteristics of the power system state data and the spatial gradient distribution of the fluid dynamic environment parameters.

[0032] Among them, the distributed digital twin architecture is a virtual simulation system jointly constructed by multiple processing nodes; the digital twin model is a simulation model of the real UAV and environment in the system; the fluctuation characteristics are the changing characteristics of the power system state data, such as the speed fluctuation speed and voltage offset; the spatial gradient distribution is the spatial distribution characteristics of the fluid environment parameters, such as the direction of air pressure change and temperature distribution density; through this step, the adversarial training scenario is finally generated.

[0033] In the embodiment of the present application, the digital twin model extracts features such as speed fluctuation speed and voltage offset from power data, and extracts information such as the direction of air pressure change and temperature distribution density from environmental data. The first rule setting module maps the speed fluctuation to the attenuation coefficient of the virtual drone's motion capability, adjusts the collision volume, and generates a first constraint rule (first, the fluctuation is divided into grades to generate a propulsion force reduction value, which is then converted into an attenuation coefficient, and the reference volume parameter is adjusted to obtain a boundary value). The second rule setting module converts the direction of air pressure change to the direction of virtual aerodynamic interference and generates a second constraint rule for the environmental disturbance range. The first strategy generation module resets the communication delay based on the voltage offset to obtain an anti-interference communication strategy. The second strategy generation module redistributes virtual obstacles based on the temperature distribution to obtain an obstacle avoidance strategy. Finally, these rules and strategies are integrated to construct an adversarial training scenario. For example, the model extracts that a certain drone has rapid speed fluctuations and large voltage offsets, and that the air pressure in a certain direction changes significantly and the temperature distribution is uneven in the environment. It then generates corresponding constraint rules and strategies, which are integrated to construct a virtual adversarial scenario.

[0034] S14. After loading the corresponding corrected control parameters into each individual drone, the actual response results of the individual drone in the adversarial training scenario are recorded.

[0035] Among them, the corrected control parameters are the flight control parameters adjusted by S12; the adversarial training scenario is the virtual training environment constructed by S13; the actual response result is the actual performance of the UAV flying in the scenario according to the corrected parameters, such as trajectory, speed change, etc.; through this step, the recorded actual response result is finally generated.

[0036] In this embodiment of the present application, the corrected control parameters obtained in S12 are input into the drone control system, and the drone is allowed to fly in the scene constructed in S13. The recording device records the actual performance of the flight trajectory, speed changes, obstacle avoidance, etc. in real time, and then sends these records to the cluster processing node. For example, the drone loads the corrected speed and steering parameters, flies in the virtual scene, and the recording device records the flight path, whether it avoids obstacles when encountering obstacles, etc., and then sends the records to the node.

[0037] S15. Compare the deviation between the actual response result and the expected behavior of the digital twin model. When the deviation exceeds the dynamic fault tolerance threshold, trigger the fault tolerance adjustment mechanism to reconstruct the correction logic of the control parameters.

[0038] Among them, the actual response result is the actual performance of the drone recorded by S14; the expected behavior of the digital twin model is the performance of the drone predicted by the model; the deviation is the difference between the actual and expected; the dynamic fault tolerance threshold is the acceptable difference range set according to the actual situation; the fault tolerance adjustment mechanism is the way to adjust the correction logic when the deviation exceeds the limit; the correction logic of the control parameters is the method of adjusting the control parameters; through this step, the reconstructed correction logic will be generated when the deviation exceeds the limit.

[0039] In this embodiment, the actual response recorded in S14 is compared with the expected behavior of the digital twin model, the deviation between the two is calculated, and a determination is made as to whether the deviation exceeds a dynamic fault tolerance threshold. If so, the fault tolerance adjustment mechanism is activated, and the correction logic for the control parameters is readjusted. For example, if the actual flight trajectory of a drone differs significantly from the model's expected behavior and exceeds a threshold, the system will readjust the method for adjusting parameters based on speed and air pressure.

[0040] S16. Based on the corrected control parameters and actual response results of multiple UAV individuals exchanged by distributed nodes, collaboratively generate an anti-interference swarm flight strategy.

[0041] Among them, distributed nodes are multiple units in the cluster that process and exchange data; the anti-interference swarm flight strategy is the collaborative flight method of the cluster to deal with interference; through this step, the anti-interference swarm flight strategy is finally generated.

[0042] In an embodiment of the present application, distributed nodes control drones, sending each other their positions, corrected parameters, and actual response results at fixed intervals to form a cluster data set. The control parameter adjustment amplitude and frequency are extracted from this data to determine the flight requirements and urgency of each drone. The distances and position change rates of adjacent drones are also calculated to determine formation constraints. Finally, the adjustment schemes are integrated to generate an anti-interference swarm flight strategy, combining the urgency of the flight requirements and the formation constraints. For example, a cluster of drones sends information to each other to form a data set. After analysis, it is found that some drones require emergency speed adjustments and some require steering. Combined with the formation constraints, this integration generates a flight strategy for collaborative obstacle avoidance in the cluster.

[0043] For example, when a cluster of 25 drones undergoes adversarial training, S11 is first used to collect the motor speed changes, voltage fluctuations, and surrounding air pressure and temperature data of each drone, and summarize them to the distributed processing node; S12 analyzes the data and adjusts the control parameters of drones with large speed fluctuations and obvious changes in surrounding air pressure, such as slowing down the speed of a drone by 2; S13 uses the digital twin model to extract features, generate constraint rules and strategies, and build a virtual adversarial scenario; S14 allows the drone to load the correction parameters and fly in the scenario, record the trajectory, obstacle avoidance, etc. and send them to the node; S15 compares the actual flight with the model expectation, and readjusts the parameter correction logic for drones whose deviation exceeds the threshold; in S16, drones send information to each other to form a data set, analyze flight requirements and formation constraints, and finally generate a cluster anti-interference collaborative flight strategy.

[0044] By executing S11 to S16, the embodiment of the present application provides support for subsequent steps by collecting basic data, and the control parameters are corrected to adapt the drone to the actual state and environment. The confrontation scenario constructed based on real data improves the authenticity of training. The actual response records provide a basis for fault-tolerant adjustment and strategy generation. The fault-tolerant mechanism enhances system stability. The distributed collaboratively generated group strategy improves the overall anti-interference ability and coordination of the cluster, and ultimately achieves efficient collaborative optimization of cluster drone confrontation training.

[0045] In one possible embodiment, S13, using a digital twin model of a distributed digital twin architecture, constructing an adversarial training scenario based on the fluctuation characteristics of the power system state data and the spatial gradient distribution of the fluid dynamic environment parameters, includes: Step 131: Use the feature extraction module in the digital twin model to extract corresponding fluctuation features from the power state parameters of each individual drone. The fluctuation features include the speed fluctuation amplitude change rate and the voltage anomaly offset. Simultaneously, extract the spatial gradient distribution from the fluid dynamic environment parameters of each individual drone. The spatial gradient distribution includes the pressure change rate vector and the temperature field contour line density.

[0046] Among them, the digital twin model is a virtual model that simulates the real drone and environment, and the feature extraction module is the part of the model that extracts data characteristics; the power state parameters are the operating data of the drone power system, and the fluctuation characteristics are the change characteristics extracted from them, including the speed fluctuation amplitude change rate (the speed of the speed change amplitude) and the voltage anomaly offset (the value of the voltage deviating from the normal range); the fluid dynamic environment parameters are the fluid environment data around the drone, and the spatial gradient distribution is the spatial change extracted from it, including the air pressure change rate vector (the speed and direction of the change of air pressure in different directions) and the temperature field contour line density (the density of lines with the same temperature); through this step, the fluctuation characteristics and spatial gradient distribution are finally generated.

[0047] In this embodiment, the feature extraction module first analyzes the power state parameters of each drone, calculating the speed fluctuation amplitude change rate (e.g., if the speed changes from 1000 rpm to 1200 rpm in 10 seconds, the change rate is 200 ÷ 1000 ÷ 10 = 0.02, or 2% per second) and the voltage anomaly offset (e.g., if the normal voltage is 12 volts and the measured voltage is 12.3 volts, the offset is 0.3 volt), thereby obtaining the fluctuation characteristics. Simultaneously, the fluid dynamic environment parameters are analyzed to determine the changes in air pressure in different directions and the density of temperature contours, thereby obtaining the spatial gradient distribution. For example, a drone's speed fluctuation amplitude change rate is 3% per second, the voltage offset is 0.4 volts, the surrounding air pressure changes by 0.2 kPa per meter eastward, and there are 3 temperature contours per meter. These are extracted as the fluctuation characteristics and spatial gradient distribution.

[0048] Step 132: Map the speed fluctuation amplitude change rate to the maneuverability attenuation coefficient of the virtual entity through the first rule setting module in the digital twin model, and generate a first boundary constraint rule for characterizing the collision volume of the virtual entity based on the maneuverability attenuation coefficient.

[0049] Among them, the first rule setting module is the part that sets the virtual rules in the digital twin model; the speed fluctuation amplitude change rate is the speed of the speed change amplitude, and the virtual entity is the object that simulates the real drone in the virtual scene; the maneuverability attenuation coefficient is the degree of decline in the virtual entity's mobility; the first boundary constraint rule is the rule that stipulates the collision range of the virtual entity, and the collision volume is the spatial range in which the virtual entity may collide; through this step, the first boundary constraint rule is finally generated.

[0050] In the embodiment of the present application, the first rule setting module first divides the speed fluctuation amplitude change rate into different levels (e.g., 0-5% for the first level, 5%-10% for the second level), generates corresponding propulsion reduction values ​​(10% for the first level, 20% for the second level), and then converts them into maneuverability attenuation coefficients (0.1 for the first level, 0.2 for the second level). The virtual entity's baseline collision volume (length, width, and height) is then adjusted based on the attenuation coefficients. The larger the attenuation coefficient, the greater the volume adjustment ratio (e.g., 0.2 corresponds to 1.2 times). Finally, the first boundary constraint rule is determined based on the adjusted volume. For example, if a drone has an 8% speed fluctuation amplitude change rate (second level) and a attenuation coefficient of 0.2, its baseline volume is 2 meters long, 1 meter wide, and 0.5 meters high. After adjustment, it becomes 2.4 meters long, 1.2 meters wide, and 0.6 meters high. This is used to generate the first rule.

[0051] Step 133: Convert the air pressure change rate vector into a virtual aerodynamic interference direction through the second rule setting module in the digital twin model, and generate a second boundary constraint rule for characterizing the range of environmental disturbance effects based on the aerodynamic interference direction.

[0052] Among them, the second rule setting module is the part that sets the environmental rules in the digital twin model; the air pressure change rate vector is the speed and direction of the change of air pressure in different directions, and the virtual aerodynamic interference direction is the thrust direction of the air on the drone in the virtual scene; the second boundary constraint rule is the rule that stipulates the range of environmental disturbance, and the range of environmental disturbance is the area where the environment affects the drone; through this step, the second boundary constraint rule is finally generated.

[0053] In this embodiment of the present application, the second rule-setting module first decomposes the pressure rate of change vector into components in different directions, assigning weights to each direction (e.g., 0.7 horizontally and 0.3 vertically). The components are multiplied by the weights to create a virtual aerodynamic interference direction. The module then adjusts the preset environmental disturbance range based on this direction, aligning the central axis of the range with the interference direction. Boundary parameters are determined to generate the second boundary constraint rule. For example, if the pressure rate of change vector is 0.3 kPa / m horizontally and 0.1 kPa / m vertically, with weights of 0.7 and 0.3, respectively, the resulting weights are 0.21 horizontally and 0.03 vertically. The resulting direction is biased toward the horizontal, and the environmental disturbance range is adjusted accordingly, generating the second rule.

[0054] Step 134: Reset the communication delay parameters based on the voltage anomaly offset through the first strategy generation module in the digital twin model to obtain an anti-interference communication strategy.

[0055] Among them, the first strategy generation module is the part of the digital twin model that generates the communication strategy; the voltage anomaly offset is the value of the voltage deviating from the normal range, and the communication delay parameter is the allowable delay time for signal transmission; the anti-interference communication strategy is a method to reduce communication interference, and the anti-interference communication strategy is finally generated through this step.

[0056] In this embodiment, the first strategy generation module first determines the normal voltage range (e.g., 12 ± 0.2 volts) and calculates the abnormal voltage offset (e.g., a measured voltage of 12.4 volts with a 0.2 volt offset). The module then adjusts the communication delay parameter based on the offset. The larger the offset, the larger the delay parameter (e.g., from 50 milliseconds to 70 milliseconds for a 0.2 volt offset), thus forming an anti-interference communication strategy. For example, if a drone experiences a 0.3 volt voltage offset, the module adjusts its communication delay parameter from 40 milliseconds to 60 milliseconds to reduce interference.

[0057] Step 135: Through the second strategy generation module in the digital twin model, the virtual obstacle coordinates are redistributed according to the density of the temperature field contour lines to obtain an environment-adaptive obstacle avoidance strategy.

[0058] Among them, the second strategy generation module is the part of the digital twin model that generates the obstacle avoidance strategy; the temperature field contour line density is the density of lines with the same temperature (reflecting the speed of temperature change), and the virtual obstacle coordinates are the positions of objects that hinder flight in the virtual scene; the environment-adaptive obstacle avoidance strategy is an obstacle avoidance method adjusted according to the environment. Through this step, the environment-adaptive obstacle avoidance strategy is finally generated.

[0059] In this embodiment, the second strategy generation module first analyzes the density of temperature contour lines. A higher density indicates a more complex environment. It then increases the number of virtual obstacles in high-density areas (for example, in an area with three contour lines per meter, the number of obstacles increases from five to eight) and adjusts their distribution. Finally, based on this new distribution, it develops an obstacle avoidance strategy (such as reducing speed or increasing detection frequency), forming an adaptive obstacle avoidance strategy. For example, if a region has four temperature contour lines per meter, the module adds five obstacles to the area, and the strategy requires the drone to slow down by 20% in this area.

[0060] Step 136: Through the fusion module in the digital twin model, an adversarial training scenario is constructed based on the first boundary constraint rule, the second boundary constraint rule, the anti-interference communication strategy, and the environment adaptive obstacle avoidance strategy.

[0061] Among them, the fusion module is the part that integrates information in the digital twin model; the first and second boundary constraint rules respectively stipulate the virtual entity collision range and environmental disturbance range; the anti-interference communication strategy and the environment adaptive obstacle avoidance strategy are the corresponding communication and obstacle avoidance methods; the adversarial training scenario is the virtual environment used for training, and this step finally generates the adversarial training scenario.

[0062] In this embodiment, the fusion module first collects the first and second boundary constraint rules, anti-interference communication strategy, and environment-adaptive obstacle avoidance strategy generated in steps 132-135. These rules and strategies are then integrated into the same virtual environment to ensure consistency, forming an adversarial training scenario that includes collision rules, environmental perturbations, communication, and obstacle avoidance methods. For example, the module integrates the virtual entity collision range, northeastern environmental perturbations, 60 millisecond communication delay, and obstacle avoidance requirements in complex temperature areas to construct a complete scenario.

[0063] For example, in a cluster UAV confrontation training, step 131 first extracts the speed fluctuation amplitude change rate of 5% per second, the voltage abnormal offset of 0.2 volts, the air pressure change rate vector of 0.2 kilopascals per meter to the south, and the temperature field contour line density of 3 per meter from UAV A; step 132 sets the maneuverability performance attenuation coefficient of the virtual entity of UAV A to 0.1 based on this, and generates the first boundary constraint rule after adjusting the collision volume; step 133 determines that the virtual aerodynamic interference direction is southward according to the southward air pressure change, and generates the second boundary constraint rule; step 134 adjusts the communication delay parameter of UAV A from 50 milliseconds to 60 milliseconds to obtain an anti-interference communication strategy; step 135 adds three obstacles in the area with dense temperature contour lines to generate an environment-adaptive obstacle avoidance strategy; finally, step 136 integrates the above rules and strategies to construct a complete confrontation training scenario.

[0064] By executing steps 131 to 136, the embodiment of the present application extracts the key features of the drone power and environment, generates rules and strategies for the virtual scene in a targeted manner, and integrates them into a unified adversarial training scene, so that the scene can accurately reflect the actual power state and environmental characteristics, thereby improving the authenticity and reliability of the scene, providing a virtual platform that fits reality for cluster drone adversarial training, and helping to improve the effectiveness of training.

[0065] In one possible embodiment, step 132, mapping the speed fluctuation amplitude change rate to the maneuverability attenuation coefficient of the virtual entity through a first rule setting module in the digital twin model, and generating a first boundary constraint rule for characterizing the collision volume of the virtual entity based on the maneuverability attenuation coefficient, includes: b1. Using the mapping unit in the first rule setting module, the fluctuation amplitude change rate is divided into different levels according to the thresholds corresponding to the preset intervals, and the intervals corresponding to the speed fluctuation amplitude change rate are obtained.

[0066] Among them, the mapping unit in the first rule setting module is the part used to classify the data. The fluctuation amplitude change rate refers to the speed of the speed change amplitude. The preset interval is several numerical ranges divided in advance. The threshold is the boundary value of each interval. The interval is the numerical range to which the fluctuation amplitude change rate belongs. Through this step, the interval corresponding to the fluctuation amplitude change rate will be finally determined.

[0067] In an embodiment of the present application, first, the mapping unit will refer to several pre-set intervals and the boundary values ​​of each interval. For example, the three intervals of 0-5%, 5%-10%, and above 10% are set, and the boundary values ​​are 5% and 10% respectively. Then, the fluctuation amplitude change rate is compared with these boundary values ​​to determine which interval it belongs to. For example, the fluctuation amplitude change rate of a certain drone is 7%. After comparing it with the boundary values, it is determined that it belongs to the 5%-10% interval.

[0068] b2. Generate a propulsion force reduction value based on the calculation method corresponding to the interval.

[0069] Among them, the calculation method corresponding to the interval means that each interval has a corresponding calculation method, and the propulsion force reduction degree value refers to the degree of reduction in the propulsion force of the drone. Through this step, the propulsion force reduction degree value will eventually be obtained.

[0070] In an embodiment of the present application, first, a corresponding calculation method is set for each interval. For example, the fluctuation amplitude change rate is multiplied by 0.1 for the 0-5% interval, and the fluctuation amplitude change rate is multiplied by 0.2 for the 5%-10% interval. Then, the calculation method corresponding to the interval determined in step b1 is substituted into the fluctuation amplitude change rate for calculation to obtain the propulsion force reduction degree value. For example, if the interval determined in step b1 is 5%-10%, and the fluctuation amplitude change rate is 7%, according to the calculation method of this interval, 7% is multiplied by 0.2, and the propulsion force reduction degree value is 1.4%.

[0071] b3. After mapping the propulsion force reduction degree value into different grades, it is converted into the maneuverability attenuation coefficient of the virtual entity.

[0072] Among them, the graded mapping refers to mapping the propulsion force reduction degree value to different values ​​according to different ranges. The maneuverability attenuation coefficient of the virtual entity is a value indicating the degree of reduction in the virtual entity's movement ability. Through this step, the maneuverability attenuation coefficient will eventually be obtained.

[0073] In the embodiment of the present application, first, several ranges of propulsion force reduction values ​​and the attenuation coefficients corresponding to each range are set in advance, for example, 0-1% corresponds to 0.1, 1%-2% corresponds to 0.2, and then the range of the propulsion force reduction value obtained in step b2 is checked to obtain the corresponding maneuverability attenuation coefficient. For example, the propulsion force reduction value obtained in step b2 is 1.4%, which belongs to the range of 1%-2%, and the corresponding attenuation coefficient is 0.2.

[0074] b4. Determine, by means of a rule generation unit in the first rule setting module, a baseline collision volume parameter of the virtual entity in a normal state and an adjustment ratio corresponding to the baseline collision volume parameter based on the maneuverability attenuation coefficient. The baseline collision volume parameter includes a baseline length, a baseline width, and a baseline height.

[0075] Among them, the rule generation unit in the first rule setting module is the part used to formulate rules. The baseline collision volume parameter of the virtual entity in the normal state refers to the basic size of the space where a collision may occur when the virtual entity moves normally, including the baseline length, baseline width and baseline height. The adjustment ratio corresponding to the baseline collision volume parameter refers to the amplification ratio determined according to the attenuation coefficient. Through this step, the baseline collision volume parameter and the corresponding adjustment ratio will be finally determined.

[0076] In an embodiment of the present application, first, the rule generation unit determines the baseline collision volume parameters of the virtual entity under normal conditions, such as a baseline length of 2 meters, a baseline width of 1 meter, and a baseline height of 0.5 meters, and then determines the adjustment ratio based on the maneuverability attenuation coefficient obtained in step b3. For example, an attenuation coefficient of 0.1 corresponds to 1.1 times, and 0.2 corresponds to 1.2 times. For example, if the attenuation coefficient obtained in step b3 is 0.2, the rule generation unit will determine the adjustment ratio to be 1.2 times, and at the same time determine the baseline length to be 2 meters, the baseline width to be 1 meter, and the baseline height to be 0.5 meters.

[0077] b5. Adjust the reference length, reference width, and reference height respectively according to the corresponding adjustment ratios to obtain the adjusted length, adjusted width, and adjusted height.

[0078] The adjusted length, adjusted width, and adjusted height refer to the sizes of the virtual entity collision volume after proportional adjustment. This step will ultimately yield the adjusted length, width, and height.

[0079] In an embodiment of the present application, first, the adjustment ratio obtained in step b4 is multiplied by the reference length, reference width and reference height respectively to calculate the adjusted size. For example, if step b4 determines that the adjustment ratio is 1.2 times, the reference length is 2 meters, the reference width is 1 meter, and the reference height is 0.5 meters, then the adjusted length is 2 meters multiplied by 1.2, which is equal to 2.4 meters, the adjusted width is 1 meter multiplied by 1.2, which is equal to 1.2 meters, and the adjusted height is 0.5 meters multiplied by 1.2, which is equal to 0.6 meters.

[0080] b6. Convert the three-dimensional space range defined by the adjusted length, adjusted width, and adjusted height into maximum offset values ​​along the X-axis, Y-axis, and Z-axis directions. The maximum offset value is the boundary value that cannot be crossed when the virtual entity moves.

[0081] Among them, the three-dimensional space range defined by the adjusted length, adjusted width, and adjusted height refers to the size of the space where virtual entities may collide. The maximum offset values ​​along the X-axis, Y-axis, and Z-axis refer to the distances that the virtual entity cannot exceed when moving in these three directions. The boundary values ​​that cannot be crossed when the virtual entity moves are these maximum offset values. Through this step, the maximum offset values ​​in the three directions will eventually be obtained.

[0082] In an embodiment of the present application, first, the adjusted length, adjusted width, and adjusted height obtained in step b5 are used as the total ranges in the X-axis, Y-axis, and Z-axis directions, respectively, and then the total range in each direction is divided by 2 to obtain the maximum offset values ​​in the positive and negative directions in each direction. For example, when the adjusted length is 2.4 meters, the maximum offset value in the X-axis direction is 2.4 meters divided by 2, which is equal to ±1.2 meters. When the adjusted width is 1.2 meters, the maximum offset value in the Y-axis direction is 1.2 meters divided by 2, which is equal to ±0.6 meters. When the adjusted height is 0.6 meters, the maximum offset value in the Z-axis direction is 0.6 meters divided by 2, which is equal to ±0.3 meters.

[0083] b7. Based on the boundary value, generate a first boundary constraint rule for characterizing the collision volume of the virtual entity.

[0084] Among them, the boundary values ​​refer to the maximum offset values ​​in the three directions obtained in step b6. The first boundary constraint rule used to characterize the collision volume of the virtual entity is a rule that stipulates that the virtual entity cannot exceed these offset values ​​when moving. Through this step, the first boundary constraint rule will eventually be generated.

[0085] In an embodiment of the present application, first, the maximum offset values ​​of the X-axis, Y-axis and Z-axis directions obtained in step b6 are collected, and then these offset values ​​are integrated into a rule to clarify that the virtual entity cannot exceed these offset values ​​when moving. For example, the offset values ​​of ±1.2 meters on the X-axis, ±0.6 meters on the Y-axis, and ±0.3 meters on the Z-axis are integrated to form the first boundary constraint rule that "when the virtual entity moves, the offset in the X-axis direction cannot exceed ±1.2 meters, the offset in the Y-axis direction cannot exceed ±0.6 meters, and the offset in the Z-axis direction cannot exceed ±0.3 meters."

[0086] For example, in a swarm drone confrontation training, when processing the speed fluctuation amplitude change rate of drone B, the mapping unit in step b1 compares the fluctuation amplitude change rate of 8% with the preset intervals of 0-5%, 5%-10%, and above 10%, and determines that it belongs to the 5%-10% interval; the calculation method corresponding to this interval in step b2 is multiplication by 0.2, so the propulsion force reduction value is 8% × 0.2 = 1.6%; in step b3, 1.6% belongs to the 1%-2% gear, corresponding to the maneuverability attenuation coefficient of 0.2; in step b4, the rule generation unit determines the reference collision volume parameters as a reference length of 3 meters and a reference width of 1. 5 meters, the base height is 0.8 meters, and the adjustment ratio is determined to be 1.2 times according to the attenuation coefficient of 0.2; in step b5, the adjustment ratio is used to calculate the adjusted length 3×1.2=3.6 meters, the width 1.5×1.2=1.8 meters, and the height 0.8×1.2=0.96 meters; in step b6, the adjusted volume is converted into maximum offset values ​​of ±1.8 meters on the X axis, ±0.9 meters on the Y axis, and ±0.48 meters on the Z axis; in step b7, based on these offset values, the first boundary constraint rule of "when the virtual entity moves, the X-axis offset does not exceed ±1.8 meters, the Y-axis does not exceed ±0.9 meters, and the Z-axis does not exceed ±0.48 meters" is generated.

[0087] By executing steps b1 to b7, the embodiment of the present application classifies the rate of change of the speed fluctuation amplitude, calculates the degree of propulsion weakening, and determines the coefficient of motion reduction. Based on this, the collision volume parameters of the virtual entity are adjusted and constraint rules are generated. This allows the collision range of the virtual entity to accurately reflect the changes in its motion ability, making the collision rules in the virtual scene more in line with the actual situation, improving the rationality of the virtual scene, and providing a more reliable virtual environment basis for drone confrontation training.

[0088] In one possible embodiment, step 133, using a second rule setting module in the digital twin model, converts the air pressure change rate vector into a virtual aerodynamic interference direction, and generates a second boundary constraint rule for characterizing the range of environmental disturbance effects based on the aerodynamic interference direction, including: c1. Decomposing the direction information of the air pressure change in the air pressure change rate vector into components along the X-axis, Y-axis, and Z-axis directions through the conversion unit in the second rule setting module.

[0089] Among them, the conversion unit in the second rule setting module is the part that processes the information on the direction of air pressure change. The air pressure change rate vector contains information on the direction and speed of air pressure change. The components along the X-axis, Y-axis, and Z-axis are the specific values ​​that split the direction of air pressure change into three directions: horizontal front and back, horizontal left and right, and vertical up and down. Through this step, the components of these three directions will eventually be obtained.

[0090] In an embodiment of the present application, first, the conversion unit receives an air pressure change rate vector, which can reflect the direction in which the air pressure changes and how fast the change is. Then, the directional information of this vector is split into specific values ​​in three directions: horizontal front and back (X-axis), horizontal left and right (Y-axis), and vertical up and down (Z-axis). For example, if a certain air pressure change rate vector shows that the air pressure mainly changes in the horizontal front and back direction, the conversion unit will split it into an X-axis component of 0.3, a Y-axis component of 0.1, and a Z-axis component of 0.05.

[0091] c2. Set the corresponding interference weight coefficient for the component in each coordinate axis direction. The interference weight coefficient is used to indicate the degree of influence of the component in the corresponding coordinate axis direction on the virtual aerodynamic interference direction.

[0092] Among them, the components of each coordinate axis direction are the air pressure change values ​​in the X-axis, Y-axis, and Z-axis directions. The interference weight coefficient is a value that represents the influence of the air pressure change in each direction on the virtual aerodynamic interference direction. Through this step, the interference weight coefficient corresponding to each direction will be finally determined.

[0093] In an embodiment of the present application, first, according to the actual situation of the influence of air pressure in various directions during the flight of the drone, interference weight coefficients are set for the X-axis, Y-axis, and Z-axis. For example, the horizontal direction has a greater impact on the flight of the drone, so the weight coefficients of the X-axis and Y-axis are set to 0.4, and the Z-axis is set to 0.2. For example, interference weight coefficients of 0.4, 0.4, and 0.2 are set for the X-axis, the Y-axis, and the Z-axis.

[0094] c3. Convert the intensity information in the pressure change rate vector into an intensity coefficient.

[0095] The intensity information in the pressure change rate vector refers to the speed of the pressure change, and the intensity coefficient is a specific numerical value converted from this speed. The intensity coefficient will eventually be obtained through this step.

[0096] In an embodiment of the present application, first, the correspondence between the range of air pressure change speed and the intensity coefficient is set. For example, when the air pressure change rate is 0-0.5 kPa / m, the intensity coefficient is 0.5, and when it is 0.5-1 kPa / m, the intensity coefficient is 1. Then, the corresponding intensity coefficient is found according to the intensity information in the air pressure change rate vector. For example, when the air pressure change rate is 0.6 kPa / m, the corresponding intensity coefficient is 1.

[0097] c4. Multiply the intensity coefficient by the interference weight coefficient of each coordinate axis direction to obtain the weighted coefficient of each coordinate axis direction.

[0098] Among them, the intensity coefficient is a numerical value indicating the speed of air pressure change, the interference weight coefficient in each coordinate axis direction is the numerical value of the influence of each direction on the virtual aerodynamic force, and the weighted coefficient in each coordinate axis direction is the result of multiplying the intensity coefficient by the corresponding weight coefficient. Through this step, the weighted coefficients of the three directions will eventually be obtained.

[0099] In an embodiment of the present application, first, the intensity coefficient obtained in step c3 is taken out, and then multiplied by the X-axis, Y-axis, and Z-axis interference weight coefficients set in step c2 respectively to obtain the weighted coefficient in each direction. For example, if the intensity coefficient is 1, the X-axis weight is 0.4, the Y-axis weight is 0.4, and the Z-axis weight is 0.2, then the X-axis weighting coefficient is 1×0.4=0.4, the Y-axis is 1×0.4=0.4, and the Z-axis is 1×0.2=0.2.

[0100] c5. Based on the weighting coefficient, perform weighted calculation on the components of each coordinate axis direction to obtain the weighted direction components.

[0101] Among them, the weighting coefficient is the numerical value of the influence degree in each direction, the components in the directions of each coordinate axis are the numerical values ​​of the air pressure changes in the three directions, and the weighted directional components are the results of multiplying the components by the corresponding weighting coefficients. Through this step, the weighted components in the three directions will eventually be obtained.

[0102] In an embodiment of the present application, first, the X-axis, Y-axis, and Z-axis components obtained in step c1 are taken out, and then multiplied by the corresponding directional weighting coefficients obtained in step c4 respectively to obtain the weighted directional components. For example, if the X-axis component is 0.3 and the weighting coefficient is 0.4, then the weighted X-axis component is 0.3×0.4=0.12; the Y-axis component is 0.1 and the weighting coefficient is 0.4, and 0.1×0.4=0.04 is obtained; the Z-axis component is 0.05 and the weighting coefficient is 0.2, and 0.05×0.2=0.01 is obtained.

[0103] c6. Synthesize the weighted directional components to obtain the virtual aerodynamic interference direction.

[0104] Among them, the weighted directional components are the influence values ​​of the air pressure changes in the three directions, and the virtual aerodynamic interference direction is the virtual aerodynamic action direction obtained by combining these three components. Through this step, the virtual aerodynamic interference direction will eventually be obtained.

[0105] In the embodiment of the present application, first, the weighted components of the X-axis, Y-axis, and Z-axis obtained in step c5 are collected, and then these three components are merged to determine a total direction. This direction is the interference direction of the virtual aerodynamic force on the drone. For example, the weighted components of the X-axis are 0.12, the Y-axis is 0.04, and the Z-axis is 0.01. The direction obtained after merging is mainly along the X-axis direction.

[0106] c7. Calculate the angular deviation between the virtual aerodynamic force interference direction and the preset environmental disturbance action range reference direction through the rule generation unit in the second rule setting module.

[0107] Among them, the rule generation unit in the second rule setting module is the part that formulates the environmental interference rules. The virtual aerodynamic interference direction is the direction of action of the virtual aerodynamic force. The preset environmental disturbance action range reference direction is the environmental interference reference direction set in advance. The angle deviation is the angle between the two directions. Through this step, the angle deviation will eventually be obtained.

[0108] In an embodiment of the present application, first, the rule generation unit determines a preset reference direction, such as the horizontal front-to-back direction (positive direction of the X-axis), and then measures the angle between the virtual aerodynamic interference direction obtained in step c6 and the reference direction. This angle is the angle deviation. For example, if the virtual aerodynamic interference direction deviates 10 degrees toward the horizontal left and right direction (positive direction of the Y-axis), and the reference direction is the positive direction of the X-axis, then the angle deviation is 10 degrees.

[0109] c8. According to the angle deviation, the preset environmental disturbance action range is rotated and adjusted so that the central axis of the adjusted environmental disturbance action range is consistent with the virtual aerodynamic force interference direction, thereby obtaining the boundary parameters of the adjusted environmental disturbance action range.

[0110] Among them, the angle deviation is the angle between the virtual aerodynamic interference direction and the reference direction, the preset environmental disturbance effect range is the environmental disturbance influence range set in advance, the central axis is the central direction line of this range, and the boundary parameter of the adjusted environmental disturbance effect range is the boundary data of the adjusted range. Through this step, the adjusted boundary parameters will eventually be obtained.

[0111] In an embodiment of the present application, first, according to the angular deviation obtained in step c7, the preset environmental disturbance action range is rotated around the center so that the central axis of the range is consistent with the virtual aerodynamic interference direction obtained in step c6. For example, if the angular deviation is 10 degrees, the preset range is rotated 10 degrees, and then the boundary data of the range after rotation is determined. For example, the preset range is a sector with a radius of 5 meters. After rotating 10 degrees, the new boundary parameters include the angular range and radius of the sector.

[0112] c9. Integrate the adjusted boundary parameters of the environmental disturbance effect range to generate a second boundary constraint rule for characterizing the environmental disturbance effect range.

[0113] Among them, the boundary parameter of the adjusted environmental disturbance range is the boundary data of the adjusted disturbance range, and the second boundary constraint rule used to characterize the environmental disturbance range is the rule that stipulates the environmental disturbance influence range. Through this step, the second boundary constraint rule will eventually be generated.

[0114] In an embodiment of the present application, first, the adjusted boundary parameters obtained in step c8 are collected, and then these parameters are organized into a rule to clarify the spatial range of the environmental disturbance. For example, if the boundary parameters show that the interference range is a sector with a radius of 5 meters and a central axis 10 degrees off the Y axis along the positive direction of the X axis, it is organized into a second boundary constraint rule of "the environmental disturbance effect range is a sector with a radius of 5 meters and a central direction 10 degrees off the Y axis along the positive direction of the X axis".

[0115] For example, in a certain swarm drone confrontation training, when processing the air pressure data around drone A, in step c1, the conversion unit decomposes the air pressure change rate vector into an X-axis component of 0.2, a Y-axis component of 0.15, and a Z-axis component of 0.08; in step c2, the X-axis interference weight coefficient is set to 0.4, the Y-axis is 0.4, and the Z-axis is 0.2; in step c3, the air pressure change rate is 0.7 kPa / m, corresponding to the intensity coefficient of 1; in step c4, the weighting coefficient is calculated, X-axis 1×0.4=0.4, Y-axis 1×0.4=0.4, and Z-axis 1×0.2=0.2; in step c5, the weighted components are obtained, X-axis 0.2×0.4=0.08, Y-axis 0.15× 0.4=0.06, Z axis 0.08×0.2=0.016; in step c6, the direction of the synthetic virtual aerodynamic disturbance is 37 degrees from the positive direction of the X-axis to the positive direction of the Y-axis (by calculating the arc tangent value of the ratio of the Y-axis to the X-axis components, 0.06÷0.08=0.75, corresponding to an angle of approximately 37 degrees); in step c7, the reference direction is the positive direction of the X-axis, and the angle deviation is 37 degrees; in step c8, the preset 6-meter radius sector disturbance range is rotated 37 degrees to obtain the adjusted boundary parameters; in step c9, the parameters are integrated to generate the second boundary constraint rule of "the environmental disturbance action range is a sector with a radius of 6 meters, and the center direction is 37 degrees along the positive direction of the X-axis and offset from the Y-axis".

[0116] By executing c1 to c9, the embodiment of the present application splits the direction of air pressure change, calculates the degree of influence of each direction, determines the direction of virtual aerodynamic interference, and then adjusts the range of environmental disturbance and generates rules, so that the interference direction and range in the virtual environment match the actual air pressure change situation, making the environmental disturbance closer to the real situation, improving the accuracy of the virtual scene, and providing a more realistic environmental reference for UAV confrontation training.

[0117] In one possible embodiment, S12, based on the abnormal fluctuation amplitude of the power system state data and the spatial gradient change information of the fluid dynamic environment parameters, correcting the control parameters of each individual drone includes: Step 121: Extract the magnitude and frequency of abnormal fluctuations from the power system state data.

[0118] Among them, the power system status data is the operating information of the UAV power part, including changes in motor speed, power supply voltage, etc.; the magnitude of the abnormal fluctuation amplitude is the degree of change in these data beyond the normal range; the frequency of occurrence is the number of times this abnormal change occurs; through this step, the magnitude and frequency of the abnormal fluctuation amplitude will eventually be obtained.

[0119] In an embodiment of the present application, the normal range of the power system status data is first determined, for example, the motor speed is normally between 1000-1500 rpm and the voltage is normally between 12-14 volts. Then, the parts outside this range are found from the collected data, the degree of change of these parts is measured, and the number of occurrences within a certain period of time is counted. For example, the motor speed of a certain drone reaches 1600 rpm several times, which is 100 rpm beyond the normal range. This is the magnitude of the abnormal fluctuation amplitude. It occurs 3 times within 1 hour, which is the frequency of occurrence.

[0120] Step 122: Normalize the magnitude of the abnormal fluctuation amplitude to obtain the corresponding basic value, convert the occurrence frequency into a correction coefficient according to a preset ratio, and generate the fluctuation characteristic value by multiplying the basic value and the correction coefficient.

[0121] Among them, normalization processing is to convert the size of the abnormal fluctuation amplitude into a value between 0 and 1; the basic value is the result obtained after normalization processing; the preset ratio is the ratio set in advance to convert the occurrence frequency into a numerical value; the correction coefficient is the numerical value after the occurrence frequency is converted according to the preset ratio; the fluctuation characteristic value is the result of multiplying the basic value and the correction coefficient; through this step, the fluctuation characteristic value will eventually be obtained.

[0122] In the embodiment of the present application, the size of the abnormal fluctuation amplitude obtained in step 121 is first normalized by the calculation method of (size of the abnormal fluctuation amplitude - minimum abnormal value) ÷ (maximum abnormal value - minimum abnormal value) to obtain a basic value, and then the occurrence frequency is converted into a correction coefficient according to a preset ratio, for example, each occurrence corresponds to 0.2, and finally the two are multiplied to obtain a fluctuation characteristic value. For example, the size of the abnormal fluctuation amplitude is 100 rpm, the minimum abnormal value is 50 rpm, and the maximum is 150 rpm. The basic value is (100-50) ÷ (150-50) = 0.5, the occurrence frequency is 3 times, the correction coefficient is 3×0.2=0.6, and the fluctuation characteristic value is 0.5×0.6=0.3.

[0123] Step 123: Analyze the changes of fluid dynamic environment parameters at different spatial gradients to obtain spatial gradient change information.

[0124] Among them, the fluid dynamic environment parameters are the environmental information of the air and other fluids around the UAV, including the air pressure and temperature at different locations; the spatial gradient change is the changing trend of these parameters at different locations; the spatial gradient change information is a summary of these changing trends; through this step, the spatial gradient change information will eventually be obtained.

[0125] In an embodiment of the present application, the fluid dynamic environment parameters at different positions around the drone are first collected, such as the air pressure and temperature in front, behind, left and right. Then, the parameters at these positions are compared and the change trend is analyzed. For example, the air pressure around a drone is 101.3 kPa in the north, 101.1 kPa in the south, the temperature is 25 degrees in the east, and 23 degrees in the west. The spatial gradient change information obtained after analysis is "the air pressure in the north is 0.2 kPa higher than that in the south, and the temperature in the east is 2 degrees higher than that in the west."

[0126] Step 124: Associate the fluctuation characteristic value with the spatial gradient change information to form a comprehensive information set.

[0127] Among them, the fluctuation characteristic value is the value obtained in step 122 that reflects the abnormal situation of the power system; the spatial gradient change information is the information obtained in step 123 that reflects the environmental change; the association is to find the corresponding relationship between the two; the comprehensive information set is the information combination that integrates these corresponding relationships; through this step, a comprehensive information set will eventually be obtained.

[0128] In an embodiment of the present application, the fluctuation characteristic value obtained in step 122 is first matched with the spatial gradient change information obtained in step 123 according to the position of the drone. For example, the fluctuation characteristic value at a certain position is 0.3, and the spatial gradient change at the same position is "air pressure high in the north and low in the south". Then, the number of times this matching relationship occurs is counted, and the one with a large number of occurrences is regarded as a strong correlation. Finally, these strongly correlated information are sorted according to individual drones. For example, a fluctuation characteristic value of 0.3 appears multiple times on a drone at a position where "air pressure is high in the north and low in the south". This relationship is sorted into a comprehensive information set.

[0129] Step 125: Based on the comprehensive information set, determine the control parameter adjustment direction and initial adjustment value of each individual drone.

[0130] Among them, the comprehensive information set is the set of integrated power and environment-related information obtained in step 124; the control parameters are the parameters for controlling the flight of the UAV, such as speed and steering angle; the adjustment direction is the direction in which the parameter needs to be increased or decreased; the initial adjustment value is the preliminarily determined adjustment amount; through this step, the adjustment direction and initial adjustment value of the control parameter will eventually be obtained.

[0131] In an embodiment of the present application, the comprehensive information set obtained in step 124 is first analyzed to see the combination of the fluctuation characteristic value and the spatial gradient change information. For example, when the fluctuation characteristic value is large and the air pressure changes drastically, it is judged that the speed needs to be slowed down, and then the adjustment direction and the initial adjustment amount are determined based on this judgment. For example, if the comprehensive information set shows that a certain drone is unstable in flight when the fluctuation characteristic value is 0.3 and the "air pressure is high in the north and low in the south", the adjustment direction is determined to be slowing down, and the initial adjustment value is set to 2 meters / second.

[0132] Step 126: Adjust the initial adjustment value according to the preset control parameter adjustment restriction conditions to obtain a target adjustment value that meets the restriction conditions.

[0133] Among them, the preset control parameter adjustment restriction conditions are the adjustment range set to ensure the safety of the drone, such as the speed cannot be lower than 1 meter per second; the initial adjustment value is the preliminary adjustment amount obtained in step 125; the target adjustment value is the final adjustment amount that meets the restriction conditions; through this step, the target adjustment value will eventually be obtained.

[0134] In the embodiment of the present application, the preset restriction conditions are first clarified, such as the speed cannot be lower than 1 m / s after adjustment, and then the initial adjustment value obtained in step 125 is checked to see whether it meets these conditions. If it does not meet the requirements, it is adjusted until it meets the requirements. For example, if the initial adjustment value is 2 m / s and the original speed is 5 m / s, the adjusted speed is 3 m / s, which meets the requirements, and the target adjustment value is 2 m / s; if the initial adjustment value is 4 m / s, the adjusted speed is 1 m / s, which just meets the requirements, and the target value is 4 m / s.

[0135] Step 127: According to the target adjustment value, the control parameters of each individual UAV are modified according to the control parameter adjustment direction to obtain the modified control parameters of each individual UAV.

[0136] Among them, the target adjustment value is the adjustment amount that meets the restriction conditions obtained in step 126; the control parameter adjustment direction is the increase or decrease direction obtained in step 125; the modified control parameter is the parameter ultimately used to control the flight of the UAV after adjustment; through this step, the modified control parameters of each individual UAV will eventually be obtained.

[0137] In the embodiment of the present application, the adjustment direction obtained in step 125 and the target adjustment value obtained in step 126 are first determined, and then the original control parameters are modified according to this direction and value. For example, the original speed parameter of a drone is 5 meters / second, the adjustment direction is slowing down, and the target adjustment value is 2 meters / second. The modified speed parameter is 5-2=3 meters / second.

[0138] For example, in a swarm UAV confrontation training, when processing the control parameter adjustment of UAV A, step 121 extracts from its power data the magnitude of the abnormal fluctuation amplitude of 100 rpm (outside the normal range), which occurs three times within 1 hour; in step 122, the magnitude of the abnormal fluctuation amplitude is normalized, with the minimum abnormal value of 50 rpm and the maximum of 150 rpm. The basic value is (100-50) ÷ (150-50) = 0.5. The frequency of occurrence of 3 times is converted to a correction coefficient of 0.6 according to the preset ratio (each time corresponds to 0.2), and the fluctuation characteristic value is 0.5×0.6=0.3 ; Step 123 analyzes its surrounding environment and obtains the spatial gradient change information that "the air pressure in the north is 0.2 kPa higher than that in the south, and the temperature in the east is 2 degrees higher than that in the west"; Step 124 associates the fluctuation characteristic value 0.3 with the environmental information to form a comprehensive information set; Step 125 determines that the adjustment direction is to slow down the speed based on the set, and the initial adjustment value is 2 meters / second; Step 126 checks and finds that the original speed is 5 meters / second, and the adjusted speed is 3 meters / second, which meets the restriction condition of "not less than 1 meter / second", and the target adjustment value is 2 meters / second; Step 127 modifies according to the direction and target value to obtain a modified speed parameter of 3 meters / second.

[0139] By executing steps 121 through 127, the present embodiment extracts information about power system anomalies and environmental spatial changes, correlates the two, and determines the adjustment direction and value of control parameters, ensuring that the adjustments comply with safety limits, ultimately obtaining appropriate control parameters. This allows the drone's control parameters to better adapt to power system anomalies and changes in the surrounding environment, allowing the drone to fly more stably during confrontation training and better cope with complex situations.

[0140] In a possible embodiment, step 124, associating the fluctuation characteristic value with the spatial gradient change information to form a comprehensive information set, includes: d1. For each individual drone, match the corresponding fluctuation characteristic value with the value at the same position in the spatial gradient change information according to the spatial gradient position of the individual drone, and record the corresponding relationship between each pair of matching data.

[0141] Among them, the individual drone is a single drone in the cluster; the spatial gradient position is the specific spatial location of the drone; the fluctuation characteristic value is the value reflecting the abnormality of the drone's power system; the spatial gradient change information is the environmental change data at different locations; the value at the same position is the environmental data corresponding to the location of the drone; matching is to match the fluctuation characteristic value and environmental data at the same location; the corresponding relationship is the correlation between the two; through this step, the corresponding relationship of each pair of matching data will be finally recorded.

[0142] In an embodiment of the present application, first, the specific spatial position of each drone is determined, and then the environmental data corresponding to the position is found in the spatial gradient change information, and then the fluctuation characteristic value of the drone is matched with these environmental data, and this association is recorded. For example, the fluctuation characteristic value of a drone at position P is 0.4, and the temperature change in the spatial gradient change information of position P is 2°C / meter, then the correspondence between "fluctuation characteristic value 0.4 and temperature change 2°C / meter" is recorded.

[0143] d2. Count the occurrence frequencies of the corresponding relationships of each pair of matching data in the corresponding drone individuals, and mark the corresponding relationships whose occurrence frequencies are greater than or equal to the preset frequency threshold as strong associations.

[0144] Among them, the correspondence between each pair of matching data is the association between the fluctuation characteristic value recorded in step d1 and the environmental data; the frequency of occurrence is the number of times this correspondence occurs in the same drone; the preset frequency threshold is the standard set in advance for judging whether it is a strong correlation; a strong correlation is a correspondence whose frequency reaches or exceeds the threshold; through this step, the correspondence with a strong correlation will eventually be marked.

[0145] In the embodiment of the present application, first, the number of times each pair of corresponding relationships recorded in step d1 appears in the same drone is counted, and then this number is compared with a preset frequency threshold (for example, 4 times). If the number reaches or exceeds the threshold, the pair of corresponding relationships is marked as a strong association. For example, if a corresponding relationship appears 5 times in the same drone and the preset threshold is 4 times, it is marked as a strong association.

[0146] d3. Group the strongly correlated matching data according to the individual identifiers of each drone to obtain data groups.

[0147] Among them, the drone individual identifier is a mark (such as a number) that distinguishes different drones; the strongly associated matching data is the association relationship marked in step d2; data grouping is to group the strongly associated data of the same drone together; through this step, the data grouping divided by individual drones will be finally obtained.

[0148] In the embodiment of the present application, first, a unique identifier (such as numbers A, B, and C) is set for each drone, and then the strongly correlated data marked in step d2 are classified according to the corresponding identifiers, and the strongly correlated data belonging to the same identifier are put together to form a data group. For example, if the drone numbered A has two pairs of strongly correlated data, these two pairs of data are grouped together, called the data group numbered A.

[0149] d4. The extreme values ​​and median values ​​of the fluctuation characteristic values ​​in the statistical data group are used to form the characteristic interval. The upper and lower limits and mean values ​​of the spatial gradient change information are simultaneously calculated to form the numerical interval.

[0150] Among them, the data group is the strongly correlated data divided by individual drones in step d3; the extreme values ​​of the fluctuation eigenvalue are the maximum and minimum fluctuation eigenvalues ​​in the data; the median is the middle value after the fluctuation eigenvalues ​​are sorted by size; the characteristic interval is the range of fluctuation eigenvalues ​​determined by the extreme values ​​and the median; the upper and lower limits of the spatial gradient change information are the maximum and minimum values ​​in the environmental data; the mean is the average value of the environmental data; the numerical interval is the range of environmental data determined by the upper and lower limits and the mean; through this step, the characteristic interval and numerical interval will eventually be formed.

[0151] In an embodiment of the present application, first, the maximum, minimum, and middle values ​​of the fluctuation characteristic value are found from each data group (such as the maximum 0.6, the minimum 0.2, and the middle 0.4 in a certain group), and these three values ​​are used to determine the characteristic interval (0.2-0.6); at the same time, the maximum, minimum, and average values ​​of the spatial gradient change information in the group are calculated (such as the maximum pressure change is 0.4, the minimum is 0.1, and the average is 0.25), and these three values ​​are used to determine the value interval (0.1-0.4).

[0152] d5. Based on the individual identification, characteristic interval and value interval of each drone, a comprehensive information set is formed.

[0153] Among them, the individual drone identification is a mark to distinguish different drones; the characteristic interval is the range of fluctuation characteristic values ​​determined in step d4; the numerical interval is the range of environmental data determined in step d4; the comprehensive information set is an information combination that integrates the identification, characteristic interval and numerical interval; through this step, a comprehensive information set will eventually be formed.

[0154] In an embodiment of the present application, first, the identification of each drone, the corresponding feature interval and the numerical interval are integrated together to form a set containing these three parts of information. For example, the drone numbered B has a feature interval of 0.3-0.7 and a numerical interval of 0.2-0.5. These three are integrated into the comprehensive information of numbered B. This type of information of all drones is aggregated to form a comprehensive information set.

[0155] For example, a cluster consists of three drones (numbered 1, 2, and 3). The processing process is as follows: in step d1, the spatial position of each drone is determined, and the fluctuation characteristic value of 0.5 of number 1 at position Q is matched with the air pressure change of 0.3 kPa / m at that position, and the corresponding relationship is recorded; in step d2, it is statistically found that this corresponding relationship appears 5 times in number 1, and the preset frequency threshold is 4 times, which is marked as a strong correlation; in step d3, all strongly correlated data of number 1 are grouped together, and numbers 2 and 3 are processed similarly to obtain three data groups; in step d4, the fluctuation characteristic value of number 1 is maximum 0.6, minimum 0.4, median 0.5, and the characteristic range is 0.4-0.6. The air pressure change in the spatial gradient change information is maximum 0.4, minimum 0.2, mean 0.3, and the numerical range is 0.2-0.4; in step d5, the identifier, characteristic range 0.4-0.6, and numerical range 0.2-0.4 of number 1 are integrated, and the other numbers are processed similarly to form a comprehensive information set.

[0156] By executing d1 to d5, the embodiment of the present application matches the power abnormality data of the drone with the environmental data of its location, screens out frequently occurring correlations, and then classifies the drones and determines the data range. The final comprehensive information set can clearly reflect the correlation between the power characteristics and environmental changes of each drone in different environments, providing a clear and reliable reference basis for the subsequent adjustment of the drone control parameters, so that the adjustment is more in line with the actual operating conditions.

[0157] In one possible embodiment, S16, based on the modified control parameters and actual response results of the multiple individual drones exchanged by the distributed nodes, collaboratively generating an anti-interference swarm flight strategy includes: Step 161: Based on the distributed nodes, control each individual drone to send its actual position information, corrected control parameters, and actual response results to other individuals in the cluster at preset time intervals, and at the same time receive the actual position information, corrected control parameters, and actual response results sent by other individuals to form a cluster data set containing the actual position information, corrected control parameters, and actual response results of all individual drones.

[0158] Among them, distributed nodes are multiple units in the cluster responsible for processing and exchanging information. The preset time interval is the fixed time for sending information between drones. The actual location information is the current location of the drone. The corrected control parameters are the adjusted flight control data. The actual response result is the performance of the drone flying according to the corrected parameters. The cluster dataset is a summary of this information from all drones. Through this step, a cluster dataset containing information from all drones will eventually be formed.

[0159] In an embodiment of the present application, first, the distributed node controls each drone to send its actual position, corrected control parameters and actual flight performance to other drones in the cluster at a fixed time (for example, 8 seconds). At the same time, each drone receives this information sent by other drones. Finally, all received information and its own information are aggregated together to form a collection containing all drone data. For example, a cluster has 15 drones, and each drone sends its own position coordinates, adjusted flight speed and actual flight trajectory every 8 seconds, and receives information from other 14 drones. This information is aggregated to form a cluster data set.

[0160] Step 162: Extract the adjustment amplitude and adjustment frequency of the corrected control parameters of each individual drone from the cluster data set.

[0161] The cluster data set is the information of all drones summarized in step 161, the corrected control parameters are the adjusted flight control data, the adjustment amplitude is the change in the control parameters compared with the original parameters, and the adjustment frequency is the number of times the control parameters change per unit time. Through this step, the adjustment amplitude and adjustment frequency of each drone will be finally extracted.

[0162] In an embodiment of the present application, first, the corrected control parameters of each drone are found from the cluster data set, and these parameters are subtracted from their original parameters. The difference obtained is the adjustment amplitude; at the same time, the number of times the control parameters of each drone change in unit time (for example, 1 minute) is counted to obtain the adjustment frequency. For example, the original flight speed of a drone is 6 meters / second, and the corrected speed is 4 meters / second. The adjustment amplitude is 2 meters / second. The speed is adjusted 3 times in 1 minute, and the adjustment frequency is 3 times / minute.

[0163] Step 163: Compare the deviations between the adjustment amplitude and the adjustment frequency and the actual response results, and determine the maneuvering requirements of each UAV according to the deviations. Different maneuvering requirements correspond to different urgency levels.

[0164] Among them, the adjustment amplitude is the size of the change in the control parameter, the adjustment frequency is the number of times the parameter changes per unit time, the actual response result is the flight performance of the drone, the deviation degree is the gap between the adjusted parameter and the actual flight performance, the maneuverability requirement is the necessity of the drone to adjust the flight status, and the urgency is the priority of the maneuverability requirement. Through this step, the maneuverability requirement and the corresponding urgency of each drone will be finally determined.

[0165] In an embodiment of the present application, first, the adjustment amplitude, adjustment frequency and actual response results of each drone are compared, and the difference between them is calculated (for example, the adjustment amplitude is 2 meters / second, but the speed in actual flight is only reduced by 1 meter / second, the deviation degree is 1 meter / second), and then different thresholds are set according to the deviation degree (for example, the deviation degree exceeds 1.2 meters / second for high emergency), and the maneuvering requirements and corresponding urgency of each drone are determined according to the threshold. For example, the deviation degree of a drone is 1.5 meters / second, which exceeds the set high emergency threshold and is determined to be a high-urgency deceleration requirement.

[0166] Step 164: extract the actual position information of each drone individual from the cluster data set, and calculate the distance and relative position change rate between adjacent drone individuals.

[0167] Among them, the actual position information is the current location of the drone, the adjacent drone refers to the drone that is closer, the distance is the spatial interval between adjacent drones, and the relative position change rate is the speed at which the distance between adjacent drones changes. Through this step, the distance and relative position change rate between adjacent drones will eventually be calculated.

[0168] In an embodiment of the present application, first, the actual position information of each drone (usually expressed as coordinates) is extracted from the cluster data set, and the spatial interval (distance) between the coordinates of adjacent drones is calculated by taking the square root of the sum of the squares of the difference between the two coordinates; then, the distances at different time points are recorded, and the change in distance is divided by the change in time to obtain the relative position change rate.

[0169] Step 165: Determine the cluster formation constraint based on the distance and relative position change rate between adjacent UAV individuals, wherein the distance between adjacent UAV individuals is inversely proportional to the strength of the corresponding cluster formation constraint.

[0170] Among them, the distance between adjacent drones is the spatial interval between them, the relative position change rate is the speed of distance change, the cluster formation constraint is the rule for maintaining the cluster formation, and the strength refers to the strictness of these rules. The distance between adjacent drones is inversely proportional to the strength of the cluster formation constraint, which means that the closer the distance, the stricter the rules, and the farther the distance, the rules can be appropriately relaxed. Through this step, the cluster formation constraint will eventually be determined.

[0171] In the embodiment of the present application, first, different constraint strengths are set according to the distance and relative position change rate between adjacent drones obtained in step 164. For example, when the distance is 2-3 meters and the change rate is small (such as within 0.2 meters / second), the constraint strength is high (the distance must be strictly maintained); when the distance is 4-5 meters and the change rate is large (such as about 0.5 meters / second), the constraint strength is low (the distance can be appropriately relaxed). For example, if the distance between adjacent drones is 3.5 meters and the relative position change rate is 0.3 meters / second, the constraint strength is determined to be medium, requiring the distance between the two to be maintained between 3-4 meters.

[0172] Step 166: Based on the urgency of the maneuvering requirements of each individual drone and the strength of the cluster formation constraints, the adjustment plans of all individual drones are integrated into an anti-interference group flight strategy.

[0173] Among them, the urgency of maneuverability requirements refers to the priority of the drone to adjust its flight status, the strength of the cluster formation constraint refers to the strictness of maintaining the cluster formation rules, the adjustment plan is the flight adjustment plan of a single drone, and the anti-interference group flight strategy is the overall collaborative flight plan of the cluster. Through this step, an anti-interference group flight strategy will eventually be generated.

[0174] In the embodiment of the present application, first, the urgency of the maneuvering needs of each drone (high-urgency needs are given priority) and the strength of the cluster formation constraints (high-strength constraints must prioritize formation), and then the adjustment plans of all drones are integrated together to ensure that both the emergency adjustment needs of a single drone and the formation rules of the cluster are met. For example, if a drone has a high-urgency deceleration need, and its adjacent drones have medium-strength formation constraints (need to maintain a distance of 3 meters), the strategy formed after integration may be "the drone decelerates by 2 meters / second, and the adjacent drones synchronously decelerate by 1 meter / second to maintain a distance of 3 meters."

[0175] For example, a cluster consists of 12 drones. In step 161, each drone sends its own position coordinates, corrected flight speed, and actual flight trajectory to other drones every 10 seconds, and receives information from the other 11 drones at the same time to form a cluster data set. In step 162, it is extracted from the data set that the speed adjustment amplitude of drone C is 1.5 m / s (original speed 5 m / s, corrected to 3.5 m / s), and the adjustment frequency is 1 time / minute. In step 163, it is found that the deviation between the adjustment amplitude of drone C and the actual response result is 1.3 m / s, which is determined to be a high-urgency deceleration requirement. In step 164, the speed of drone C is calculated. The distance between UAV C and its neighboring UAV D is 4 meters. After 10 seconds, the distance becomes 4.5 meters, and the relative position change rate is (4.5-4) / 10=0.05 meters / second. In step 165, based on the distance of 4 meters and the change rate of 0.05 meters / second, the formation constraint between the two is determined to be medium (maintaining a distance of 3.5-4.5 meters). In step 166, the high emergency deceleration requirement of UAV C and the medium formation constraint are combined, and the adjustment plans of all UAVs are integrated to generate an anti-interference swarm flight strategy: UAV C decelerates by 1.5 meters / second, UAV D decelerates by 0.5 meters / second to maintain a distance of 4 meters, and other UAVs maintain their current speed.

[0176] By executing steps 161 to 166, the embodiment of the present application realizes information sharing between drones through distributed nodes, extracts key adjustment parameters and determines the emergency needs of individuals in combination with actual flight performance, and formulates the formation constraints of the cluster based on the position relationship. Finally, the generated anti-interference swarm flight strategy is integrated, which can not only meet the emergency adjustment needs of a single drone, but also ensure the stability of the cluster formation, thereby improving the cluster's ability to coordinate and cooperate in complex environments and its ability to cope with interference.

[0177] Figure 2 A schematic diagram of the structure of an XX device (or system) provided in an embodiment of the present application is shown as follows: Figure 2 As shown, the device includes: The acquisition module 21 is used to obtain the power system status data and fluid dynamic environment parameters of multiple individual drones in the cluster.

[0178] The correction module 22 is used to correct the control parameters of each UAV according to the abnormal fluctuation amplitude of the power system state data and the spatial gradient change information of the fluid dynamic environment parameters.

[0179] Construction module 23 is used to construct an adversarial training scenario based on the fluctuation characteristics of the power system state data and the spatial gradient distribution of the fluid dynamic environment parameters through the digital twin model of the distributed digital twin architecture.

[0180] The recording module 24 is used to load the corresponding corrected control parameters into each individual drone and record the actual response results of the individual drone in the confrontation training scenario.

[0181] The adjustment module 25 is used to compare the deviation between the actual response result and the expected behavior of the digital twin model. When the deviation exceeds the dynamic fault tolerance threshold, the fault tolerance adjustment mechanism is triggered to reconstruct the correction logic of the control parameters.

[0182] The generation module 26 is used to collaboratively generate an anti-interference swarm flight strategy based on the modified control parameters and actual response results of multiple UAV individuals exchanged by distributed nodes.

[0183] Figure 2 The distributed digital twin architecture cluster UAV adversarial training collaborative optimization system can perform Figure 1 The implementation principle and technical effects of the distributed digital twin architecture cluster drone adversarial training collaborative optimization method described in the illustrated embodiment will not be repeated here. The specific manner in which each module and unit performs operations in the distributed digital twin architecture cluster drone adversarial training collaborative optimization system in the above embodiment has been described in detail in the embodiment of the method and will not be elaborated here.

[0184] In one possible design, Figure 2 The distributed digital twin architecture cluster UAV adversarial training collaborative optimization system of the embodiment shown can be implemented as a computing device, such as Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32 .

[0185] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32 .

[0186] The processing component 32 is configured to execute the following process: obtaining power system status data and fluid dynamic environment parameters of multiple individual drones in the cluster; and modifying the control parameters of each individual drone based on the abnormal fluctuation amplitude of the power system status data and the spatial gradient variation information of the fluid dynamic environment parameters; Through the digital twin model of the distributed digital twin architecture, an adversarial training scenario is constructed based on the fluctuation characteristics of the power system state data and the spatial gradient distribution of the fluid dynamic environment parameters; after loading the corresponding corrected control parameters in each individual drone, the actual response results of the individual drone in the adversarial training scenario are recorded; the deviation between the actual response results and the expected behavior of the digital twin model is compared. When the deviation exceeds the dynamic fault tolerance threshold, the fault tolerance adjustment mechanism is triggered to reconstruct the correction logic of the control parameters; based on the corrected control parameters and actual response results of multiple individual drones exchanged by distributed nodes, an anti-interference swarm flight strategy is collaboratively generated.

[0187] The processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component may also be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above method.

[0188] The storage component 31 is configured to store various types of data to support operations on the terminal. The storage component can be implemented by any type of volatile or non-volatile memory device, or a combination thereof, such as random access memory (RAM), static random access memory (SRAM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0189] Of course, a computing device may also include other components, such as input / output interfaces, display components, communication components, etc.

[0190] The input / output interface provides an interface between the processing component and the peripheral interface module, which can be an output device, an input device, etc.

[0191] The communication component is configured to facilitate, among other things, wired or wireless communications between the computing device and other devices.

[0192] Among them, the computing device can be a physical device or an elastic computing host provided by a cloud computing platform, etc. In this case, the computing device can refer to a cloud server, and the above-mentioned processing components, storage components, etc. can be basic server resources rented or purchased from the cloud computing platform.

[0193] The present application also provides a computer storage medium storing a computer program, wherein the computer program can achieve the above-mentioned Figure 1 The embodiment shown is a cluster drone adversarial training collaborative optimization method based on a distributed digital twin architecture.

[0194] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0195] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0196] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0197] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A distributed digital twin architecture cluster UAV adversarial training collaborative optimization method, characterized by: include: Obtain the power system status data and fluid dynamic environment parameters of multiple UAVs in the swarm; Correcting the control parameters of each individual drone based on the abnormal fluctuation amplitude of the power system state data and the spatial gradient change information of the fluid dynamic environment parameters; Through the digital twin model of the distributed digital twin architecture, adversarial training scenarios are constructed based on the fluctuation characteristics of the power system state data and the spatial gradient distribution of the fluid dynamic environment parameters; After loading the corresponding corrected control parameters into each of the drone individuals, the actual response results of the drone individuals in the confrontation training scenario are recorded; Comparing the deviation between the actual response result and the expected behavior of the digital twin model, when the deviation exceeds a dynamic fault tolerance threshold, triggering a fault tolerance adjustment mechanism to reconstruct the correction logic of the control parameters; Based on the modified control parameters of multiple UAV individuals exchanged by distributed nodes and the actual response results, an anti-interference swarm flight strategy is collaboratively generated.

2. The distributed digital twin architecture cluster UAV adversarial training collaborative optimization method according to claim 1 is characterized in that: The digital twin model of the distributed digital twin architecture constructs an adversarial training scenario based on the fluctuation characteristics of the power system state data and the spatial gradient distribution of the fluid dynamic environment parameters, including: The feature extraction module in the digital twin model extracts corresponding fluctuation features from the power state parameters of each individual drone. These fluctuation features include the speed fluctuation amplitude change rate and the voltage anomaly offset. Simultaneously, the spatial gradient distribution is extracted from the fluid dynamic environment parameters of each individual drone. The spatial gradient distribution includes the pressure change rate vector and the temperature field contour line density. The first rule setting module in the digital twin model maps the speed fluctuation amplitude change rate to a maneuverability attenuation coefficient of the virtual entity, and generates a first boundary constraint rule for characterizing the collision volume of the virtual entity based on the maneuverability attenuation coefficient; The air pressure change rate vector is converted into a virtual aerodynamic disturbance direction by a second rule setting module in the digital twin model, and a second boundary constraint rule for characterizing the range of environmental disturbance is generated according to the virtual aerodynamic disturbance direction; Resetting a communication delay parameter based on the voltage anomaly offset by a first strategy generation module in the digital twin model to obtain an anti-interference communication strategy; The second strategy generation module in the digital twin model redistributes the virtual obstacle coordinates according to the density of the temperature field contour lines to obtain an environment-adaptive obstacle avoidance strategy; Through the fusion module in the digital twin model, an adversarial training scenario is constructed based on the first boundary constraint rule, the second boundary constraint rule, the anti-interference communication strategy and the environment adaptive obstacle avoidance strategy.

3. The distributed digital twin architecture cluster UAV adversarial training collaborative optimization method according to claim 2 is characterized in that: The first rule setting module in the digital twin model maps the speed fluctuation amplitude change rate to a maneuverability attenuation coefficient of the virtual entity, and generates a first boundary constraint rule for characterizing the collision volume of the virtual entity based on the maneuverability attenuation coefficient, including: The fluctuation amplitude change rate is divided into different levels according to the threshold value corresponding to the preset interval by the mapping unit in the first rule setting module, so as to obtain the interval corresponding to the speed fluctuation amplitude change rate; generating a propulsion force reduction degree value based on a calculation method corresponding to the interval; After mapping the propulsion force reduction degree value into grades, the values ​​are converted into a maneuverability attenuation coefficient of the virtual entity; determining, by a rule generating unit in a first rule setting module, a baseline collision volume parameter of the virtual entity in a normal state and an adjustment ratio corresponding to the baseline collision volume parameter based on the maneuverability attenuation coefficient, the baseline collision volume parameter including a baseline length, a baseline width, and a baseline height; Adjusting the reference length, the reference width, and the reference height respectively according to the corresponding adjustment ratios to obtain an adjusted length, an adjusted width, and an adjusted height; Converting the three-dimensional space defined by the adjusted length, the adjusted width, and the adjusted height into maximum offset values ​​along the X-axis, the Y-axis, and the Z-axis, wherein the maximum offset values ​​are insurmountable boundary values ​​when the virtual entity moves; Based on the boundary value, a first boundary constraint rule for characterizing the collision volume of the virtual entity is generated.

4. The distributed digital twin architecture cluster UAV adversarial training collaborative optimization method according to claim 2 is characterized in that: The second rule setting module in the digital twin model converts the air pressure change rate vector into a virtual aerodynamic interference direction, and generates a second boundary constraint rule for characterizing the range of environmental disturbance according to the aerodynamic interference direction, including: Decomposing the direction information of the air pressure change in the air pressure change rate vector into components along the X-axis, Y-axis, and Z-axis directions by a conversion unit in the second rule setting module; Setting a corresponding interference weight coefficient for the component in each coordinate axis direction, wherein the interference weight coefficient is used to indicate the degree of influence of the component in the corresponding coordinate axis direction on the virtual aerodynamic interference direction; Converting the intensity information in the air pressure change rate vector into an intensity coefficient; Multiplying the intensity coefficient by the interference weight coefficient in each coordinate axis direction to obtain a weighted coefficient in each coordinate axis direction; Based on the weighting coefficients, weighted calculations are performed on the components of the directions of the coordinate axes to obtain weighted direction components; synthesizing the weighted directional components to obtain a virtual aerodynamic interference direction; Calculating, by a rule generation unit in a second rule setting module, an angular deviation between the virtual aerodynamic interference direction and a preset environmental disturbance action range reference direction; According to the angle deviation, the preset environmental disturbance action range is rotated and adjusted so that the central axis of the adjusted environmental disturbance action range is consistent with the virtual aerodynamic force interference direction, thereby obtaining boundary parameters of the adjusted environmental disturbance action range; The adjusted boundary parameters of the environmental disturbance effect range are integrated to generate a second boundary constraint rule for characterizing the environmental disturbance effect range.

5. The distributed digital twin architecture cluster UAV adversarial training collaborative optimization method according to claim 1 is characterized in that: The method of correcting the control parameters of each individual UAV based on the abnormal fluctuation amplitude of the power system state data and the spatial gradient change information of the fluid dynamic environment parameters includes: Extract the magnitude and frequency of abnormal fluctuations from the power system status data; Normalizing the magnitude of the abnormal fluctuation amplitude to obtain a corresponding basic value, converting the occurrence frequency into a correction coefficient according to a preset ratio, and generating a fluctuation characteristic value by multiplying the basic value by the correction coefficient; Analyzing the changes of the fluid dynamic environment parameters at different spatial gradients to obtain spatial gradient change information; Associating the fluctuation characteristic value with the spatial gradient change information to form a comprehensive information set; Based on the comprehensive information set, determining the control parameter adjustment direction and initial adjustment value of each individual drone; According to the preset control parameter adjustment restriction conditions, the initial adjustment value is adjusted to obtain a target adjustment value that meets the restriction conditions; According to the target adjustment value, the control parameters of each individual drone are modified according to the control parameter adjustment direction to obtain the modified control parameters of each individual drone.

6. The distributed digital twin architecture cluster UAV adversarial training collaborative optimization method according to claim 5 is characterized in that: The associating the fluctuation characteristic value with the spatial gradient change information to form a comprehensive information set includes: For each individual drone, the corresponding fluctuation characteristic value is matched with the value at the same position in the spatial gradient change information according to the spatial gradient position of the individual drone, and the corresponding relationship between each pair of matching data is recorded; Counting the occurrence frequency of the corresponding relationship of each pair of matching data in the corresponding drone individuals, and marking the corresponding relationship with an occurrence frequency greater than or equal to a preset frequency threshold as a strong association; Group the strongly correlated matching data according to the individual identifiers of each drone to obtain data groups; Counting the extreme values ​​and median values ​​of the fluctuation characteristic values ​​in the data group to form a characteristic interval, and simultaneously calculating the upper and lower limits and the mean value of the spatial gradient change information to form a numerical interval; A comprehensive information set is formed based on the individual identifiers of the drones, the characteristic intervals, and the numerical intervals.

7. The distributed digital twin architecture cluster UAV adversarial training collaborative optimization method according to claim 1 is characterized in that: The modified control parameters of the multiple UAV individuals exchanged based on the distributed nodes and the actual response results are used to collaboratively generate an anti-interference swarm flight strategy, including: Based on distributed nodes, each drone is controlled to send its actual position information, revised control parameters and actual response results to other drones in the cluster at preset time intervals, and at the same time receive the actual position information, revised control parameters and actual response results sent by other drones to form a cluster dataset containing the actual position information, revised control parameters and actual response results of all drones; Extracting the adjustment amplitude and adjustment frequency of the corrected control parameters of each individual drone from the cluster data set; Comparing the deviation between the adjustment amplitude and the adjustment frequency and the actual response result, and determining the maneuvering requirements of each individual UAV according to the deviation, wherein different maneuvering requirements correspond to different urgency levels; Extract the actual position information of each drone individual from the cluster data set, and calculate the distance and relative position change rate between adjacent drone individuals; Determining a cluster formation constraint based on the distances and relative position change rates between the adjacent individual drones, wherein the distances between the adjacent individual drones are inversely proportional to the strength of the corresponding cluster formation constraint; Based on the urgency of each individual drone's maneuvering needs and the strength of the cluster formation constraints, the adjustment plans of all individual drones are integrated into an anti-interference swarm flight strategy.

8. A distributed digital twin architecture cluster drone adversarial training collaborative optimization system, characterized by: include: An acquisition module is used to obtain the power system status data and fluid dynamic environment parameters of multiple individual drones in the cluster; A correction module, configured to correct the control parameters of each individual UAV based on the abnormal fluctuation amplitude of the power system state data and the spatial gradient change information of the fluid dynamic environment parameters; A construction module for constructing adversarial training scenarios based on the fluctuation characteristics of power system state data and the spatial gradient distribution of fluid dynamic environment parameters through the digital twin model of the distributed digital twin architecture; A recording module is used to load the corresponding corrected control parameters into each of the drone individuals and record the actual response results of the drone individuals in the confrontation training scenario; An adjustment module is configured to compare the deviation between the actual response result and the expected behavior of the digital twin model, and when the deviation exceeds a dynamic fault tolerance threshold, trigger a fault tolerance adjustment mechanism to reconstruct the correction logic of the control parameters; A generation module is used to collaboratively generate an anti-interference swarm flight strategy based on the corrected control parameters of multiple drone individuals exchanged by distributed nodes and the actual response results.

9. A computing device, characterized in that It includes a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a cluster drone adversarial training collaborative optimization method with a distributed digital twin architecture as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that A computer program is stored, and when the computer program is executed by a computer, the method for collaborative optimization of cluster drone adversarial training of a distributed digital twin architecture as described in any one of claims 1 to 7 is implemented.

Images