A method for generating a UAV swarm repulsion strategy in an interference environment and an electronic device

Abstract

The application provides a UAV swarm driving strategy generation method and electronic equipment in an interference environment, relates to the technical field of UAV prevention and control and automatic control, and the scheme comprises the following steps: constructing a driving station environment model; acquiring detection information of a UAV swarm, and dividing the UAV swarm into a UAV target cluster; acquiring interference items that shield the UAV swarm, and equating the interference items to point clouds composed of discrete voxels; predicting the driving effectiveness of each UAV target cluster under the interference of the voxel point cloud by the driving equipment emitted by each driving station site according to an alternative driving strategy, the resource consumption of each driving station site due to the emission of the driving equipment to drive the UAV swarm, and the driving uncertainty for each UAV target cluster; calculating the reward value of each alternative driving strategy; and taking the alternative driving strategy corresponding to the maximum reward value as the optimal driving strategy for the UAV swarm. The application ensures high driving effectiveness and low resource consumption under the consideration of driving uncertainty.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of unmanned aerial vehicle prevention and control and automatic control technology, in particular, to a method for generating a strategy for driving away a group of unmanned aerial vehicles in an interference environment and an electronic device. BACKGROUND

[0002] With the improvement of the group activity capability of small unmanned aerial vehicles, the group and collaborative application capability is significantly improved, and more and more large-scale activities arrange performances of groups of unmanned aerial vehicles. The performances of groups of unmanned aerial vehicles must be reported, and the performances of unmanned aerial vehicles in controlled airspace or public places without authorization need to be driven away in time. In the prior art, various schemes for driving away and scheduling groups of unmanned aerial vehicles have appeared, mainly including:

[0003] 1. Ground high-power jammer: such a device has a wide range of action, but usually indiscriminately suppresses signals, and has poor selectivity. In a complex electromagnetic environment with a large number of legal unmanned aerial vehicles, it is easy to cause "injury by mistake" and interfere with normal communication and operation.

[0004] 2. Portable jamming gun: the device is relatively flexible to operate, but its design principle determines that it can usually only "one-to-one" target, and the action distance and range are very limited, and it is not powerful enough when facing a large number of unmanned aerial vehicle swarms.

[0005] 3. Net capture gun: this is a physical capture means, but its effective action height and distance are greatly limited, and it is difficult to deal with high-altitude and high-speed unmanned aerial vehicle targets. In addition, the unmanned aerial vehicle captured by the net will lose control and fall, and when used in the air above a densely populated area or important facilities, it may cause the risk of secondary injury or property loss.

[0006] Therefore, a new scheme for driving away and scheduling groups of unmanned aerial vehicles is needed to overcome the defects in the above schemes. SUMMARY

[0007] The technical problem to be solved by the present application is the above-mentioned problems existing in the prior art driving away and scheduling scheme for groups of small unmanned aerial vehicles, and a method for generating a strategy for driving away a group of unmanned aerial vehicles in an interference environment and an electronic device are provided.

[0008] The technical scheme of the present application provides a method for generating a strategy for driving away a group of unmanned aerial vehicles in an interference environment, comprising:

[0009] constructing an environment model of a driving away station, the environment model of the driving away station comprising the positions of S driving away station sites and the inventory of driving away devices in the first driving away station site , ;

[0010] detecting the UAV group entering the protected area to obtain detection information of the UAV group, the detection information including a position set and a speed set of each UAV in the UAV group; dividing the UAV group into UAV target clusters according to the position and speed of each UAV in the UAV group , ;

[0011] obtaining an interference term shielding the UAV group, and equating the interference term to a point cloud composed of discrete voxels;

[0012] constructing a repulsion utility prediction model for predicting repulsion utility of each UAV target cluster under interference of the point cloud of voxels by repulsion devices emitted by each repulsion station site according to an alternative repulsion strategy ;

[0013] setting resource constraint allocation and optimization conditions for predicting resource consumption of each repulsion station site for repelling the UAV group by emitting repulsion devices according to the alternative repulsion strategy ;

[0014] constructing a repulsion uncertainty prediction model for predicting repulsion uncertainty of each UAV target cluster according to the alternative repulsion strategy, the repulsion uncertainty being caused by observation uncertainty of the point cloud of voxels to the UAV target cluster ;

[0015] constructing a repulsion strategy reward and punishment model for calculating a reward value of each alternative repulsion strategy : ; , and is a weight coefficient

[0016] taking the alternative repulsion strategy corresponding to the maximum reward value as the optimal repulsion strategy for the UAV group.

[0017] Preferably, in the method for generating a repulsion strategy for a UAV group in an interference environment, the UAV group entering a protected area is detected to obtain detection information of the UAV group, the detection information including a position set and a speed set of each UAV in the UAV group; the UAV group is divided into UAV target clusters according to the position and speed of each UAV in the UAV group , , wherein

[0018] the position set of each UAV is represented as , the i-th UAV position is represented as positions of the unmanned aerial vehicles, representing a total number of unmanned aerial vehicles in the unmanned aerial vehicle group;

[0019] The speed set is represented as: representing the speed of the i-th unmanned aerial vehicle;

[0020] After multi-scale clustering of the unmanned aerial vehicle group according to the spatial proximity and speed similarity principle, a plurality of unmanned aerial vehicle target clusters are obtained;

[0021] wherein the cluster center position , the speed mean and the intra-cluster dispersion of each unmanned aerial vehicle target cluster are calculated as follows:

[0022]

[0023]

[0024]

[0025] representing the number of unmanned aerial vehicles in the cluster.

[0026] Preferably, the unmanned aerial vehicle group repelling strategy generation method in the interference environment, the repelling utility prediction model is constructed for predicting the repelling utility of each unmanned aerial vehicle target cluster under the interference of the voxel point cloud by the repelling device emitted by each repelling station site according to the alternative repelling strategy , the repelling utility prediction model is represented as:

[0027]

[0028] wherein, represents the position of the selected repelling station site in the alternative repelling strategy, represents the emission time of the repelling device in the alternative repelling strategy, is a geometric coverage factor, representing the spatio-temporal interaction coverage ratio of the interference signal coverage area with a radius of to the single unmanned aerial vehicle in the unmanned aerial vehicle target cluster within a time window , and represents the repelling success probability of the alternative repelling strategy to the unmanned aerial vehicle target cluster ;

[0029] and are calculated as follows: ​​​​​​​​

[0030] ;

[0031] ;

[0032] Represents the time t. The location of the drone;

[0033] ;

[0034] ;

[0035] ;

[0036] ;

[0037] in, Let be the probability that the i-th drone is successfully driven away. For drone target clusters The average probability of a single drone being successfully driven away from the target. An electromagnetic model representing a voxel point cloud; The electromagnetic interference field representing a voxel point cloud. The minimum strength required to repel interference signals emitted by the device. Indicates the occlusion factor. express The corresponding set of deportation equipment within the deportation station site. This represents the emission intensity factor of the repelling equipment. Represents the distance decay function. This represents the cumulative attenuation of the repelling device along the propagation path. Represents any point on the path Electromagnetic attenuation coefficient, This represents the set of voxels traversed by the deflection device along its path from the transmitter to the receiver. Voxel representation Electromagnetic attenuation coefficient, This indicates that the path is in voxels The arc length within.

[0038] Preferably, in the method for generating a drone swarm drive-away strategy under interference conditions, the masking factor... It was calculated in the following way:

[0039] Voxel acquisition Occupancy probability ;

[0040] Based on occupancy probability The masking factor is obtained from the mapping relationship within the UAV target cluster. .

[0041] Preferably, the method for generating a drone swarm repelling strategy in an interference environment, the setting of resource constraint allocation and optimization conditions is used to predict the resource consumption of each repelling station site caused by launching the repelling device to repel the drone swarm according to the candidate repelling strategy In the method:

[0042] The resource constraint allocation and optimization conditions include:

[0043] The total resource constraint condition is: , The number of repelling devices launched by the repelling station site at the site position to the drone target cluster ;

[0044] The total inventory constraint condition is: , The total inventory upper limit is represented by

[0045] Preferably, the method for generating a drone swarm repelling strategy in an interference environment, the construction of a repelling uncertainty prediction model is used to predict the repelling uncertainty of each drone target cluster according to the candidate repelling strategy , the repelling uncertainty is caused by the observation uncertainty of the voxel point cloud to the drone target cluster, including:

[0046] Obtain the observation probability of the drone in the drone target cluster that is in the voxel when the detector is at the position ;

[0047] ;

[0048] Wherein, is the detector efficiency scale, is the spatial attenuation factor, , The center position of the voxel is represented by

[0049] The repelling uncertainty of the drone target cluster is proportional to its observation probability.

[0050] Preferably, the method for generating a drone swarm repelling strategy in an interference environment, the repelling device is a flying vehicle equipped with an interference module, and the interference module emits an electromagnetic interference signal with a radius of R.

[0051] In a second aspect, the present application provides a computer readable storage medium, wherein the storage medium stores program information, and a computer reads the program information to execute the steps of the UAV swarm driving strategy generation method in the interference environment according to any one of the first aspect.

[0052] In a third aspect, the present application provides a computer program product, comprising computer programs / instructions, which, when executed by a processor, implement the steps of the UAV swarm driving strategy generation method in the interference environment according to any one of the first aspect.

[0053] In a fourth aspect, the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the UAV swarm driving strategy generation method in the interference environment according to any one of the first aspect.

[0054] The above technical solutions provided by the present application have the following technical effects compared with the prior art:

[0055] The UAV swarm driving strategy generation method in the interference environment and the electronic device provided by the present application equate the interference items shielding the UAV swarm to point clouds composed of discrete voxels for simulation, thereby being capable of simulating the scene in which the UAV swarm is shielded or electromagnetically interfered in the actual situation. In addition, the present application calculates the reward value of the alternative driving strategy from three aspects of the driving utility, resource consumption, and driving uncertainty of the UAV target cluster, takes the alternative driving strategy corresponding to the maximum reward value as the optimal driving strategy for the UAV swarm, ensures a high driving utility and low resource consumption on the premise of considering the driving uncertainty, and the present application is suitable for various non-compliant performances of the UAV swarm and related virtual confrontation deduction scenarios. BRIEF DESCRIPTION OF DRAWINGS

[0056] Figure 1 A flowchart of the UAV swarm driving strategy generation method in the interference environment according to an embodiment of the present application;

[0057] Figure 2 A schematic diagram of equating the interference items shielding the UAV swarm to point clouds composed of discrete voxels and the detection range of the detector according to an embodiment of the present application;

[0058] Figure 3 A schematic diagram of the UAV swarm and the coverage range of the driving signal emitted by the driving device according to an embodiment of the present application;

[0059] Figure 4 A schematic diagram of the driving effect of the optimal driving strategy on the UAV swarm according to an embodiment of the present application;

[0060] Figure 5A schematic diagram of a hardware connection relationship of an electronic device for performing the method for generating a UAV swarm repelling strategy in an interference environment according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0061] The specific embodiments of the present application will be further described below with reference to the accompanying drawings.

[0062] It is easy to understand that, according to the technical solution of the present application, a person skilled in the art can replace various structural modes and implementation modes without changing the essential spirit of the present application. Therefore, the following specific embodiments and drawings are only exemplary descriptions of the technical solution of the present application, and should not be regarded as the whole or as a limitation or restriction on the technical solution of the application.

[0063] The present embodiment provides a method for generating a UAV swarm repelling strategy in an interference environment, which is applied to a control host configured with an operating system, such as Figure 1 As shown in the figure, the method comprises:

[0064] S100: constructing a repelling station environment model, which comprises the position of each repelling station site and the first repelling device inventory in each repelling station site , .

[0065] The design drawing of the repelling station site records the geographic position coordinate information of the repelling station site, so that the position of the repelling station site can be determined. Repelling devices are placed in each repelling station site, which can be unmanned aerial vehicles with electromagnetic interference functional components and the like.

[0066] S200: detecting a UAV swarm entering a protected area to obtain detection information of the UAV swarm, which comprises a single UAV position set and a speed set in the UAV swarm; and dividing the UAV swarm into a UAV target cluster , .

[0067] In the present application, a detector is installed in the repelling station site or other appropriate position, which can monitor the UAV swarm entering the protected area, and then control the repelling device to emit and schedule the UAV swarm after detecting the UAV swarm. In this step, the position and speed of each UAV in the UAV swarm are detected in real time, which can be realized by a detection radar, and the existing detection radar has the function of monitoring the position and speed of the UAV in real time.

[0068] When the UAV group is divided into clusters, the existing clustering algorithm can be used for implementation, such as the K-means clustering algorithm. Each UAV is taken as a target, and the positions of all UAVs in the UAV group are taken as objects to be divided into clusters. A plurality of target positions are selected as initial clustering centers in advance. The distance between each target position and each clustering center is calculated. The targets are grouped according to the nearest distance principle to obtain a plurality of clusters. Then, the mean value of the target positions in each cluster is taken as a new clustering center, and the above grouping step is re-executed to obtain a new division cluster result. The above steps are repeated for multiple times until the division cluster result meets the iteration end condition, and the final J UAV target clusters are obtained.

[0069] S300: Obtain an interference term shielding the UAV group, and equate the interference term to a point cloud composed of discrete voxels.

[0070] As shown in Figure 2 a schematic diagram of the discrete voxels after the interference term is equivalent. The schematic diagram of the detector detection range is also given. The point cloud space formed by the discrete voxels is used to describe the shielding / fog area and other interference terms in the present scheme, and the actual flight environment of the UAV group is reproduced.

[0071] S400: Construct a repulsion utility prediction model for predicting the repulsion utility of each UAV target cluster under the interference of the voxel point cloud by the repulsion equipment emitted by each repulsion station site according to the alternative repulsion strategy .

[0072] The advantage item for repelling the UAV group is obtained in this step.

[0073] S500: Set resource constraint allocation and optimization conditions for predicting the resource consumption of each repulsion station site for emitting the UAV group by emitting the repulsion equipment according to the alternative repulsion strategy .

[0074] The resource consumption obtained in this step is the consumption item for repelling the UAV group, for example, how many repulsion equipment are emitted. The single consumption of each repulsion equipment is known, and the resource consumption can be obtained by directly multiplying the single consumption by the number of emitted equipment.

[0075] S600: Construct a repulsion uncertainty prediction model for predicting the repulsion uncertainty of each UAV target cluster according to the alternative repulsion strategy . The observation uncertainty caused by the shielding of the voxel point cloud to the UAV target cluster is used to obtain the repulsion uncertainty.

[0076] The uncertainty in the process of repelling the UAV group is obtained in this step, that is, the possibility that the UAV group cannot be normally repelled, which is a consumption item.

[0077] S700: Construct a drive strategy reward and punishment model for calculating the reward value of each candidate drive strategy : ; 、 and is a weight coefficient.

[0078] The drive strategy reward and punishment model for calculating the reward value in this step is the principle of weighted summation, which adopts positive addition for the advantage item and negative subtraction for the consumption item.

[0079] In the specific implementation process, many candidate drive strategies can be set, and in each candidate drive strategy, it is specified that each drive station site should launch a drive device to which unmanned aerial vehicle target cluster, and the launch time of the drive device, the electromagnetic interference range and electromagnetic interference intensity of the electromagnetic interference module carried by each drive device, etc. After calculating the reward value obtained by driving the unmanned aerial vehicle group according to each candidate drive strategy in this step, it is obvious that the higher the reward value, the greater the advantage of the candidate drive strategy.

[0080] S800: Take the candidate drive strategy corresponding to the maximum reward value as the best drive strategy for the unmanned aerial vehicle group.

[0081] In this step, the candidate drive strategy corresponding to the maximum reward value is directly selected as the best drive strategy for the unmanned aerial vehicle group.

[0082] The above scheme provided by the embodiment of the present application equates the interference item of the shielded unmanned aerial vehicle group to a point cloud composed of discrete voxels for simulation, thereby being able to simulate the scene in which the unmanned aerial vehicle group is shielded or subjected to electromagnetic interference in actual situations. In addition, the reward value of the candidate drive strategy is calculated from three aspects of the drive effectiveness, resource consumption and drive uncertainty of the unmanned aerial vehicle target cluster in the scheme of the present application, and the candidate drive strategy corresponding to the maximum reward value is taken as the best drive strategy for the unmanned aerial vehicle group, which can ensure high drive effectiveness and low resource consumption on the premise of considering drive uncertainty. The scheme provided by the present application is applicable to various illegal performances and virtual confrontation deduction scenarios of unmanned aerial vehicle groups.

[0083] Preferably, in the scheme of the present application, the unmanned aerial vehicle group entering the protected area is detected in step S200 to obtain detection information of the unmanned aerial vehicle group, and the detection information includes a single unmanned aerial vehicle position set and a speed set in the unmanned aerial vehicle group; the unmanned aerial vehicle group is divided into unmanned aerial vehicle target clusters , in which:

[0084] The single unmanned aerial vehicle position set is represented as: , represents the number of drones in the drone group, represents the position of each drone in the drone group, represents the total number of drones in the drone group;

[0085] The speed set is represented as: ;

[0086] After multi-scale clustering each drone in the drone group according to the principle of spatial proximity and speed similarity, a plurality of drone target clusters are obtained.

[0087] The cluster center position, the speed average and the intra-cluster dispersion of each drone target cluster are calculated as follows:

[0088] ;

[0089] ;

[0090] ;

[0091] represents the number of drones in the drone group.

[0092] In the scheme, the drone group is divided into a plurality of drone target clusters, which can reduce the target state dimension, and in specific implementation, the space occupied by each drone target cluster is adapted to the coverage space of the electromagnetic interference module of the repelling device.

[0093] Further preferably, in the above scheme, the repelling utility prediction model constructed in step S400 is used to predict the repelling utility of each drone target cluster under the interference of the repelling device of each repelling station site in the alternative repelling strategy , and the repelling utility prediction model is represented as:

[0094] ;

[0095] wherein, represents the selected repelling station site position in the alternative repelling strategy, represents the repelling device emission time in the alternative repelling strategy, is a geometric coverage factor, representing the spatio-temporal interaction coverage ratio of the interference signal coverage area with a radius of to a single drone in the drone target cluster within a time window , represents the repelling utility of the drone target cluster under the alternative repelling strategy​​​​ The probability of a successful expulsion;

[0096] Calculated in the following way and :

[0097] ;

[0098] ;

[0099] Represents the time t. The location of the drone;

[0100] ;

[0101] ;

[0102] ;

[0103] ;

[0104] in, Let be the probability that the i-th drone is successfully driven away. For drone target clusters The average probability of all drones being successfully driven away. An electromagnetic model representing a voxel point cloud; The electromagnetic interference field representing a voxel point cloud. The minimum strength required to repel interference signals emitted by the device. Indicates the occlusion factor. express The corresponding set of deportation equipment within the deportation station site. This indicates the emission intensity factor of the repelling equipment. Represents the distance decay function. This represents the cumulative attenuation of the repelling device along the propagation path. Represents any point on the path Electromagnetic attenuation coefficient, This represents the set of voxels traversed by the deflection device along its path from the transmitter to the receiver. Voxel representation Electromagnetic attenuation coefficient, This indicates that the path is in voxels The arc length within.

[0105] In the above scheme, the calculation of the distance attenuation function is assumed to be based on... r When it is a variable, it can be represented as: Specifically, in the above formula, r is replaced with... Perform the calculation. Also the degree of attenuation of the expression, for example, set at any point on the path Indicates how bad the environment is around this point, such as thick fog, wall blocking, air wave-absorbing, how much electromagnetic interference signal loses every meter. The greater the value, the more the loss of electromagnetic signal. A small piece of path length, an infinitesimal distance. It is the loss contribution of a small piece of path. After integration operation, it is to add up all the small pieces of loss to get the total loss. If the electromagnetic interference signal is emitted at the same time as the emission of the device, the emission point refers to the starting point or emission station of the device, which can be replaced by the station site. If the electromagnetic interference signal is emitted in the air, the emission point is the position of the electromagnetic interference signal. The receiving point refers to the end point of the electromagnetic interference signal, which can be represented as the position of a certain unmanned aerial vehicle in the unmanned aerial vehicle target cluster in the scheme, that is, the electromagnetic interference signal must reach the position to play a repelling role. The path refers to the travel route of the electromagnetic interference signal, that is, the whole route of the electromagnetic interference signal from the emission point to the receiving point, which is not a simple straight line because the environment may have obstacles (such as buildings, fog areas or terrain), and the electromagnetic interference signal will be weakened. The interference of the environment is simulated by using the voxel point cloud, and the path passes through these voxels. If there is attenuation in the voxel, the strength of the electromagnetic interference signal will decrease, and thus the total loss of energy on the path is calculated by using the integral term.

[0106] Further preferably, the shielding factor in the above scheme is calculated by: obtaining the occupation probability of the voxel ; obtaining the shielding factor according to the mapping relationship of the occupation probability in the unmanned aerial vehicle target cluster. Each voxel can be regarded as a three-dimensional cube with a set value of edge length and a center position of the cube. The spatial coordinate system of the voxel is as shown in Figure 2 , and the position of each unmanned aerial vehicle in each unmanned aerial vehicle target cluster is in its own coordinate system. The mapping relationship of the two coordinate systems can be determined, and thus the relationship between the voxel position and the position of each unmanned aerial vehicle in the unmanned aerial vehicle target cluster can be determined. If the position coordinates covered by the voxel and the position coordinates occupied by the unmanned aerial vehicle can overlap, it means that the voxel can cover part of the unmanned aerial vehicle, that is, there is shielding. The occupation probability can be obtained in combination with the interference strength (such as the interference degree of thick fog and interference medium). If the degree of shielding is smaller, the value of the occupation probability is smaller, and if the degree of shielding is larger, the value of the occupation probability is larger.

[0107] ​​Preferably, in the above scheme, the setting of the resource constraint allocation and optimization condition in step S500 is used to predict the resource consumption of each of the repelling station sites caused by launching the repelling equipment to repel the UAV swarm according to the candidate repelling strategy In the above scheme, the setting of the resource constraint allocation and optimization condition in step S500 is used to predict the resource consumption of each of the repelling station sites caused by launching the repelling equipment to repel the UAV swarm according to the candidate repelling strategy

[0108] The resource constraint allocation and optimization condition includes a total resource constraint condition: , represents the position of the station The number of repelling equipment launched by the repelling station site to the UAV target cluster ;

[0109] The total inventory constraint condition: , represents the upper limit of the total inventory.

[0110] In actual deployment, the number of repelling equipment is limited, so the number of repelling equipment launched is constrained by the inventory number of a single repelling station site, and the number of repelling equipment of all repelling station sites is also necessarily less than the upper limit of the total inventory.

[0111] Further, the constructing of the repelling uncertainty prediction model in step S600 is used to predict the repelling uncertainty of each UAV target cluster according to the candidate repelling strategy The repelling uncertainty is caused by the observation uncertainty of the voxel point cloud to the UAV target cluster, and includes:

[0112] The observation probability of the UAV in the voxel in the UAV target cluster by the detector at the position ;

[0113] ;

[0114] wherein, is the detector efficiency scale, is the spatial attenuation factor, , represents the center position of the voxel The repelling uncertainty of the UAV target cluster is proportional to its observation probability.

[0115] In specific applications, when the UAV swarm is repelled, the detector will detect the positions of these UAVs, but the environment is complex, there are fog areas, building obstructions or electromagnetic interference, resulting in some UAVs can be accurately detected, at this time The value of is high, close to 1, and some UAVs have low detection accuracy, at this time The value of the observation probability of each UAV is low, close to 0. is the observation credibility probability of the position of a single UAV. In order to deal with the observation uncertainty caused by detector noise and shielding, an observation weight is preferably introduced in the present solution: , The role of is the weight fraction of this data, High data has more influence on the result, Low data has less influence or is ignored. For example, if a UAV is in an open area, High (such as 0.9), then High. The system will give more recognition to this position data, and pull the center of the UAV target cluster towards it. If another UAV is in a fog area or interference area, Low (such as 0.2), then Low (such as =0.2). The system will not recognize or ignore the value, avoiding the distortion of the entire UAV target cluster calculation by fuzzy data. Based on the same principle, the uncertainty cost (higher when the observation probability is low and the shielding is high) is also proportional to the observation probability.

[0116] Further preferably, the repelling device is a flying vehicle carrying an interference module that emits an electromagnetic interference signal with a radius of R. The repelling effect of the UAV group using the algorithm of the present application is shown in Figure 3 and Figure 4 , Figure 3 It shows that when the single repelling device is at the center of repelling, the radius R of the electromagnetic interference signal it emits is 200 meters, and it can repel five UAVs, i.e. the UAV represented by the green origin in the figure. Figure 4 After each repelling station site emits a repelling device, the repelling effect of different interference module-carrying UAVs on the invading UAV group is respectively shown.

[0117] Preferably, the repelling strategy reward and punishment model in the above-mentioned solution of the present application is obtained after learning and training using a deep learning algorithm. In the specific training process, the reward value calculated in a single training is the immediate reward. Based on this immediate reward , the advantage function is further calculated, which quantifies the advantage value of the current selected repelling strategy compared to the average level, thereby further guiding the optimization of the repelling strategy.

[0118] ;

[0119] In the formula, represents the comprehensive advantage value of the current repelling strategy relative to the average level at time . A positive value indicates that the current repelling strategy is better than the average level, and a negative value indicates that the current repelling strategy is worse than the average level. In the above formula, denotes the immediate temporal difference error, which measures the deviation of the current value prediction, and is equal to the actual reward value (i.e., the immediate score) plus the future estimated reward value and then subtracts the current estimated reward value . The greater the deviation, the less accurate the estimate, and adjustments are needed. Here, denotes the value function, and denotes the input state and the output value of the state . Specifically, is constructed through the immediate temporal difference error , which is directly dependent on the value: , which reflects is the basis for calculation, as the immediate score, to assist in evaluating the deviation of the value prediction, thereby deriving long-term advantages.

[0120] In the above formula: denotes the discount factor, i.e., the discount rate of future reward values, the greater the discount. denotes the smoothing factor, i.e., the weight of controlling short-term deviation compared to long-term deviation. The higher the value, the more emphasis on long-term deviation, the lower the value, the more emphasis on short-term deviation.

[0121] In addition, in order to provide more driving strategy advantage comparison, the embodiment also provides a clipping objective function:

[0122] ;

[0123] wherein, , represents the repelling strategy, and represents the input is the current state (such as the position of each UAV in the UAV group, the information of the UAV target group), and the output is the probability of repelling strategy execution action (such as how many repelling devices are launched and which UAV target group). is a parameter, and is adjusted to improve the repelling strategy. i.e., the probability ratio of the new repelling strategy to the old repelling strategy, is the clipping coefficient (0.1-0.3). >1, then the new repelling strategy is more biased towards action ; <1, then the new repelling strategy is more biased away from action. min refers to the smaller of two values, one is the original If higher, encourage new evict policy), the other is the value of the limiting ) to ) after the clip function cut , is the average expectation.

[0124] The value function loss is calculated in combination with entropy regularization:

[0125]

[0126] Where represents the parameters of the evict policy corresponding to the optimization and value . represents the maximum clipping target, is the weight, adjusting the importance of the components in the function. represents the error square of the value estimate, is the more accurate target value. represents the entropy regularization, is the policy entropy, measuring the randomness of the evict decision. That is, the above function calculates the total score = evict policy score + estimate accuracy score + exploration reward score, based on the selection of the target with the highest total score, to prevent the evict policy from being stuck in a local optimum.

[0127] Above, the calculation of is used to implement advantage estimation, the calculation formula is used for policy optimization, and the calculation of is used for joint optimization (policy-value-entropy). In implementation, the above calculation processes are sequentially linked and the calculated quantities are fed back to the parameters of the policy network and the value network of the deep learning algorithm through gradient update, thereby completing parameter learning and policy improvement. In each training process, based on the immediate reward calculated in a single round, a policy optimization loop can be further constructed to iteratively train and optimize the parameters involved in the evict utility prediction model, the evict uncertainty prediction model, and the evict policy reward and punishment model, so that the selected evict policy tends to be better in the long term or multiple scenarios.

[0128] The embodiments of the present application also provide an electronic device, such as Figure 5As shown, the electronic device includes at least one processor 51 and at least one memory 52, at least one of the memories 52 stores program information, and at least one of the processors 51 reads the program information and executes the interference environment UAV group repulsion strategy generation method described in any of the above method embodiments. The device can also include an input device 53 and an output device 54. The processor 51, the memory 52, the input device 53 and the output device 54 can be communicatively connected. The memory 52 as a kind of non-volatile computer readable storage medium, it can be used to store non-volatile software programs, non-volatile computer executable programs and modules. The processor 51 executes various functions and data processing by running the non-volatile software programs, instructions and modules stored in the memory 52, that is, realizes the interference environment UAV group repulsion strategy generation method provided in any of the above schemes. The memory 52 can include a program storage area and a data storage area, wherein the program storage area can store the operating system and the application program required by at least one function; the data storage area can store the data created according to the use of the interference environment UAV group repulsion strategy generation method and the like. In addition, the memory 52 can include a high-speed random access memory, and can also include a non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory 52 can optionally include a memory remotely arranged with respect to the processor 51, and these remote memories can be connected to the device executing the interference environment UAV group repulsion strategy generation method through a network. Examples of the above network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network and a combination thereof. The input device 53 can receive the input user clicks and generate signal inputs related to the user settings and function controls of the interference environment UAV group repulsion strategy generation method. The output device 54 can include a display device such as a display screen. When the one or more modules are stored in the memory 52 and executed by the one or more processors 51, the interference environment UAV group repulsion strategy generation method in any of the above method embodiments is executed.

[0129] According to the needs, the above technical solutions can be combined to achieve the best technical effect.

[0130] The above is only the principles and preferred embodiments of the present application. It should be noted that for those skilled in the art, on the basis of the principles of the present application, a number of other variants can also be made, which should be regarded as the protection scope of the present application.

Claims

1. A method for generating a strategy to drive away unmanned aerial vehicle (UAV) swarms under interference conditions, characterized in that, include: Construct an environment model of the expulsion stations, the environment model of the expulsion stations including the locations of S expulsion station sites. and the Inventory of deportation equipment at each deportation station site , ; The system detects drone swarms entering the protected area, acquiring swarm detection information, including the set of individual drone positions and speeds within the swarm. Based on the positions and speeds of the individual drones, the swarm is then divided into... A cluster of drone targets , ; Obtain the interference terms that obscure the UAV swarm, and equate the interference terms to a point cloud composed of discrete voxels; A deterrence effectiveness prediction model is constructed to predict the deterrence effectiveness of deterrence equipment launched by each deterrence station according to the alternative deterrence strategies against each UAV target cluster under voxel point cloud interference. ; Resource constraints and optimization conditions are set to predict the resource consumption of each of the alternative driving-off strategies when the drone swarm is driven away by driving-off equipment. ; Construct a model to predict the uncertainty of driving away each UAV target cluster according to the alternative driving away strategies. The uncertainty of the expulsion is obtained based on the observation uncertainty caused by the occlusion of the UAV target cluster by the voxel point cloud; Construct a reward and punishment model for expulsion strategies to calculate the reward value for each alternative expulsion strategy. : ; , and These are the weighting coefficients; The alternative expulsion strategy corresponding to the maximum reward value is selected as the optimal expulsion strategy for the drone swarm.

2. The method for generating a drone swarm drive-away strategy under interference conditions according to claim 1, characterized in that, The process involves detecting the swarm of drones entering the protected area and acquiring detection information, including the set of positions and speeds of individual drones within the swarm. Based on the positions and speeds of the individual drones, the swarm is then divided into... A cluster of drone targets , middle: The set of locations of a single UAV is represented as follows: , Indicates the first The location of the drone This indicates the total number of drones in the drone swarm; The set of velocities is represented as: , Indicates the first The speed of the drone; After performing multi-scale clustering on the UAV swarm based on the principles of spatial proximity and velocity similarity, J UAV target clusters are obtained; Among them, the cluster center position of each UAV target cluster average speed and cluster divergence Calculated separately as follows: ； ； ； in, This indicates the number of drones in the cluster.

3. The method for generating a drone swarm drive-away strategy under interference conditions according to claim 2, characterized in that, The constructed deterrence effectiveness prediction model is used to predict the deterrence effectiveness of deterrence equipment launched by each deterrence station according to the alternative deterrence strategies against each UAV target cluster under voxel point cloud interference. The expulsion utility prediction model is expressed as follows: ； in, This indicates the location of the selected expulsion station among the alternative expulsion strategies. This indicates the launch time of the decoy device in the alternative decoy strategy. The geometric coverage factor represents the coverage within the time window. Inside, radius is The area covered by the interference signal and the drone target cluster The spatiotemporal interaction coverage ratio of a single drone in China. This indicates that the alternative expulsion strategies are applicable to the drone target swarm. The probability of a successful expulsion; Calculated in the following way and : ； ； Represents the time t. The location of the drone; ； ； ； ； in, Let be the probability that the i-th drone is successfully driven away. For drone target clusters The average probability of a single drone being successfully driven away from the target. An electromagnetic model representing a voxel point cloud; The electromagnetic interference field representing a voxel point cloud. The minimum strength required to repel interference signals emitted by the device. Indicates the occlusion factor. express The corresponding set of deportation equipment within the deportation station site. This indicates the emission intensity factor of the repelling equipment. Represents the distance decay function. This represents the cumulative attenuation of the repelling device along the propagation path. Represents any point on the path Electromagnetic attenuation coefficient, This represents the set of voxels traversed by the deflection device along its path from the transmitter to the receiver. Voxel representation Electromagnetic attenuation coefficient, This indicates that the path is in voxels The arc length within.

4. The method for generating a drone swarm drive-away strategy under interference conditions according to claim 3, characterized in that, Masking factor It is calculated in the following way: Voxel acquisition Occupancy probability ; Based on occupancy probability The masking factor is obtained from the mapping relationship within the UAV target cluster. .

5. The method for generating a drone swarm drive-away strategy under interference conditions according to claim 4, characterized in that, The resource constraint allocation and optimization conditions are set to predict the resource consumption of each of the drive-off stations in order to drive away the drone swarm by launching drive-off equipment according to the alternative drive-off strategies. middle: The resource constraint allocation and optimization conditions include: Overall resource constraints: , The location of the site is indicated as The decoy station targets the drone swarm. The number of decoy devices launched; Overall inventory constraints: , This indicates the upper limit of total inventory.

6. The method for generating a drone swarm drive-away strategy under interference conditions according to claim 4, characterized in that, The constructed uncertainty prediction model is used to predict the uncertainty of driving away each UAV target cluster according to the alternative driving away strategies. The uncertainty in driving away the target swarm is obtained based on the observation uncertainty caused by the occlusion of the UAV target swarm by the voxel point cloud, including: Obtain the position of the detector At that time, the target cluster of the unmanned aerial vehicles was located in the voxel Observation probability of drones within: ； in, For detector efficiency scale The spatial decay factor, , The voxel represents The central position; Uncertainty in driving away the drone target swarm It is directly proportional to its observation probability.

7. The method for generating a drone swarm drive-away strategy under interference conditions according to any one of claims 1-6, characterized in that: The driving-off device is an aircraft equipped with an interference module, which emits an electromagnetic interference signal with a radius of R.

8. A computer-readable storage medium, characterized in that, The storage medium stores program information, and after the computer reads the program information, it executes the steps of the method for generating a drone swarm drive-away strategy under interference environment as described in any one of claims 1-7.

9. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instruction is executed by the processor, it implements the steps of the method for generating a drone swarm drive-away strategy under interference conditions as described in any one of claims 1-7.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method for generating a drone swarm drive-away strategy under interference conditions as described in any one of claims 1-7.

Images