CN116700349A - Unmanned aerial vehicle cluster control method, unmanned aerial vehicle cluster control device, unmanned aerial vehicle cluster control equipment and storage medium - Google Patents

Abstract

The application discloses an unmanned aerial vehicle cluster control method, device, equipment and storage medium, wherein the method comprises the following steps: acquiring unmanned aerial vehicle cluster information and attribute information of a high-altitude detection balloon to be inspected; based on unmanned aerial vehicle cluster information and attribute information of a high-altitude detection balloon to be inspected, an initial unmanned aerial vehicle inspection scheme is generated by adopting an unmanned aerial vehicle path planning algorithm; and performing the following iteration steps with the iteration times reaching a preset iteration times threshold as a target to obtain a plurality of unmanned aerial vehicle inspection schemes, wherein the iteration steps comprise: randomly selecting a high-altitude detection balloon from each unmanned aerial vehicle path in the unmanned aerial vehicle inspection scheme to form a new high-altitude detection balloon set; aiming at a new high-altitude detection balloon set, generating a new unmanned aerial vehicle inspection scheme by adopting the unmanned aerial vehicle path planning algorithm; calculating an evaluation value of each unmanned aerial vehicle inspection scheme; and performing unmanned aerial vehicle group control according to the unmanned aerial vehicle inspection scheme with the minimum evaluation value.

Description

A UAV swarm control method, device, equipment and storage medium

technical field

The invention belongs to the technical field of UAV swarm control, and relates to a UAV swarm control method, device, equipment and storage medium, in particular to a convolutional neural network-based optimal path planning UAV swarm control method.

Background technique

An unmanned aerial vehicle is an unmanned aircraft operated by radio remote control equipment and its own program control device, or a device that is completely or intermittently operated autonomously by an on-board computer. With the continuous development of airborne electronic components, the current increasingly complex and changeable mission environment is no longer satisfied with the operation of a single drone. Therefore, multi-agent technology represented by UAV swarms has become a current research hotspot. Faced with increasingly diverse task requirements and highly complex execution environments, the application of agents tends to be cluster collaborative operations.

UAV swarms are easily affected by the environment when flying, so that when UAV swarms are flying in strong winds and air turbulence, the UAV swarms tend to deviate from the flight line, and it is difficult to maintain a stable flight attitude, which makes it easy for the UAV swarm to Out of control, the UAV cluster cannot complete the designated flight mission, reducing the effect of UAV control.

Contents of the invention

Purpose: In order to overcome the deficiencies in the prior art, the present invention provides a UAV cluster control method, device, equipment and storage medium.

Technical solution: In order to solve the above-mentioned technical problems, the technical solution adopted in the present invention is:

In a first aspect, the present invention provides a UAV cluster control method, including:

Obtain the UAV cluster information and the attribute information of the high-altitude sounding balloon that needs to be inspected; wherein the UAV cluster information includes the location of the UAV cluster; the attribute information includes the location of the high-altitude sounding balloon, inspection time, and image capture resolution And the information attributes of the stability of the UAV swarm during flight;

Based on the UAV cluster information and the attribute information of the high-altitude sounding balloons that need to be inspected, the UAV path planning algorithm is used to generate the initial UAV inspection plan;

With the number of iterations reaching the preset threshold of iterations as the goal, perform the following iterative steps to obtain multiple UAV inspection plans, wherein the iterative steps include: from the path of each UAV in the UAV inspection plan Randomly select a high-altitude sounding balloon to form a new high-altitude sounding balloon set; for the new high-altitude sounding balloon set, adopt the UAV path planning algorithm to generate a new UAV inspection plan;

Calculate the evaluation value of each UAV inspection plan;

UAV swarm control is performed according to the UAV inspection plan with the smallest evaluation value.

In some embodiments, the UAV path planning algorithm includes:

S1. Randomly select an unselected UAV cluster from the UAV collection;

S2. Randomly select an unselected high-altitude sounding balloon from the set of high-altitude sounding balloons that is closest to the location of the selected UAV; if the selected UAV cluster cannot complete the inspection task of the selected high-altitude sounding balloon, restart Select unselected UAVs from the UAV swarm set until the selected UAV swarm can complete the inspection task of the selected high-altitude sounding balloon, and add the position of the selected high-altitude sounding balloon as the inspection point to the selected In the preset path of the UAV cluster;

S3. If there are still unselected high-altitude sounding balloons, execute step S2 in a loop until there are no unselected high-altitude sounding balloons, and output the UAV inspection plan.

Further, in some embodiments, judging whether the selected UAV cluster can complete the inspection task of the selected high-altitude sounding balloon includes:

Whether the selected UAV cluster can complete the inspection task of the selected high-altitude sounding balloon needs to meet the following conditions at the same time:

(1) The position information of the positioning radar carried by the drone cluster is higher than the position longitude and latitude information of the selected high-altitude sounding balloon;

(2) The UAV swarm can reach the selected high-altitude exploration balloon from the current high-altitude exploration balloon and return to a stop.

In some embodiments, the calculation method of the iteration number threshold Δn includes:

Among them, ρ is the preset ratio operator value, N is the number of paths of the UAV cluster, and d _n is the data of the high-altitude sounding balloon included in the path of the nth UAV.

In some embodiments, calculating the evaluation value of each UAV inspection plan includes:

Among them, S is the evaluation value of the UAV inspection plan, N is the number of UAV cluster paths in the UAV inspection plan, and P ⁱ _n is the nth UAV in the UAV inspection plan. The inspection energy consumption of the i-th task to be inspected, _ε1 is the weight of the inspection energy consumption, and T ⁱ _n is the completion of the i-th task to be inspected by the n-th UAV in the UAV inspection scheme The inspection time, ε ₂ is the weight of the inspection time, and I is the number of high-altitude detection balloons.

In some embodiments, the UAV path planning algorithm uses a network unit of a bidirectional long-short-term memory layer and a CNN network unit;

The network unit of the two-way long-short-term memory layer is used to extract the characteristics of the preset path information of the t-sequence UAV according to the location of the UAV cluster and the attribute information of the high-altitude sounding balloon that needs to be inspected;

The CNN network unit is used to determine the optimal path according to the characteristics of the preset path information of the t-sequence UAV.

Further, in some embodiments, the network unit of the two-way long-short-term memory layer is used to extract the preset path of the UAV with the t-sequence according to the location of the UAV cluster and the attribute information of the high-altitude sounding balloon that needs to be patrolled. Characteristics of the information, including:

Given the UAV sequence U={U ₁ , U ₂ ,..,U _NU } and the position information of the task sequence T={T ₁ ,T ₂ ,..,T _NT } to be executed, set The initial position information of the i-th UAV is The position of the jth task point is Then the path distance between each UAV and the mission point /> Expressed as:

In the formula: ||·|| represents the two norms: i=1,..., _NU , j=1,..., _NT ; _NU is the total number of UAVs, and _NT is the task point the total number of

The path distance Γ _m of all UAVs to each mission point is expressed as a matrix of N _U × N _T is the element in it, and m represents the location point of the current path.

Further, in some embodiments, the CNN network unit is used to determine the optimal path according to the characteristics of the preset path information of the t-sequence UAV, including: _JD is the loss function of the preset path, and the loss value The smaller the value, the better the preset path, which is defined as:

In the formula: β is the preset path loss coefficient, which is used to judge the training update coefficient of the evaluation model parameters; β=[μ ₁ ,μ ₂ ]∈R ^1×2 , μ ₁ represents the probability that the sequence is a generated sequence, μ ₂ represents the probability that the sequence is reserved for real, R represents all real numbers, Indicates the path distance between UAV U _i and the jth mission point; /> Indicates whether the UAV U _i executes the preset path distance />

In a second aspect, the present invention provides an unmanned aerial vehicle cluster control device, including a processor and a storage medium;

The storage medium is used to store instructions;

The processor is configured to operate according to the instructions to perform the method according to the first aspect.

In a third aspect, the present invention provides a device, comprising:

memory;

processor;

as well as

Computer program;

Wherein, the computer program is stored in the memory and is configured to be executed by the processor to implement the method described in the first aspect above.

In a fourth aspect, the present invention provides a storage medium on which a computer program is stored, and when the computer program is executed by a processor, the method described in the first aspect is implemented.

Beneficial effects: the drone cluster control method, device, equipment and storage medium provided by the present invention have the following advantages: the cluster information of drones determined by using high-precision positioning radar and the attribute information of high-altitude sounding balloons that need to be inspected , use the UAV path planning algorithm to generate the UAV inspection plan, and according to the evaluation value of each UAV inspection plan, find the UAV inspection plan with the smallest evaluation value as the optimal solution, and realize the UAV inspection plan. The cluster flies smoothly and optimizes the flight path in real time;

Furthermore, it can effectively solve the problem that the UAV cluster deviates from the flight line and it is difficult to maintain a stable flight attitude when the UAV cluster encounters strong wind and air turbulence.

Description of drawings

1 is a schematic flow diagram of a method according to an embodiment of the present invention;

Fig. 2 is a schematic diagram of practical application according to an embodiment of the present invention.

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings and embodiments. The following examples are only used to illustrate the technical solution of the present invention more clearly, but not to limit the protection scope of the present invention.

In the description of the present invention, several means more than one, and multiple means more than two. Greater than, less than, exceeding, etc. are understood as not including the original number, and above, below, within, etc. are understood as including the original number. If the description of the first and second is only for the purpose of distinguishing the technical features, it cannot be understood as indicating or implying the relative importance or implicitly indicating the number of the indicated technical features or implicitly indicating the order of the indicated technical features relation.

In the description of the present invention, reference to the terms "one embodiment," "some embodiments," "exemplary embodiments," "examples," "specific examples," or "some examples" is intended to mean that the embodiments are A specific feature, structure, material, or characteristic described by or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

Example 1

In a first aspect, this embodiment provides a method for controlling a UAV swarm, including:

Obtain the UAV cluster information and the attribute information of the high-altitude sounding balloon that needs to be inspected; wherein the UAV cluster information includes the location of the UAV cluster; the attribute information includes the location of the high-altitude sounding balloon, inspection time, and image capture resolution And the information attributes of the stability of the UAV swarm during flight;

Based on the UAV cluster information and the attribute information of the high-altitude sounding balloons that need to be inspected, the UAV path planning algorithm is used to generate the initial UAV inspection plan;

With the number of iterations reaching the preset threshold of iterations as the goal, perform the following iterative steps to obtain multiple UAV inspection plans, wherein the iterative steps include: from the path of each UAV in the UAV inspection plan Randomly select a high-altitude sounding balloon to form a new high-altitude sounding balloon set; for the new high-altitude sounding balloon set, adopt the UAV path planning algorithm to generate a new UAV inspection plan;

Calculate the evaluation value of each UAV inspection plan;

UAV swarm control is performed according to the UAV inspection plan with the smallest evaluation value.

In some embodiments, the UAV path planning algorithm includes:

S1. Randomly select an unselected UAV cluster from the UAV collection;

S2. Randomly select an unselected high-altitude sounding balloon from the set of high-altitude sounding balloons that is closest to the location of the selected UAV; if the selected UAV cluster cannot complete the inspection task of the selected high-altitude sounding balloon, restart Select unselected UAVs from the UAV swarm set until the selected UAV swarm can complete the inspection task of the selected high-altitude sounding balloon, and add the position of the selected high-altitude sounding balloon as the inspection point to the selected In the preset path of the UAV cluster;

S3. If there are still unselected high-altitude sounding balloons, execute step S2 in a loop until there are no unselected high-altitude sounding balloons, and output the UAV inspection plan.

Further, in some embodiments, judging whether the selected UAV cluster can complete the inspection task of the selected high-altitude sounding balloon includes:

Whether the selected UAV cluster can complete the inspection task of the selected high-altitude sounding balloon needs to meet the following conditions at the same time:

(1) The position information of the positioning radar carried by the drone cluster is higher than the position longitude and latitude information of the selected high-altitude sounding balloon;

(2) The UAV swarm can reach the selected high-altitude exploration balloon from the current high-altitude exploration balloon and return to a stop.

In some embodiments, the calculation method of the iteration number threshold Δn includes:

Among them, ρ is the preset ratio operator value, N is the number of paths of the UAV cluster, and d _n is the data of the high-altitude sounding balloon included in the path of the nth UAV.

In some embodiments, calculating the evaluation value of each UAV inspection plan includes:

Among them, S is the evaluation value of the UAV inspection plan, N is the number of UAV cluster paths in the UAV inspection plan, is the inspection energy consumption of the nth UAV in the UAV inspection scheme to complete the i-th inspection task, _ε1 is the weight of the inspection energy consumption, /> In the drone inspection scheme, the nth drone completes the inspection time of the i-th inspection task to be inspected, _ε2 is the weight of the inspection time, and I is the number of high-altitude sounding balloons.

Optionally, the quality of the learning rate can directly affect the final accuracy value of the model, and the multi-classification cross-entropy function is used to represent the loss value of the model. By setting the rounds of training, the learning rate value is halved every 20 rounds to reduce the learning rate, and through continuous optimization of the learning rate, the loss value of the model reaches the minimum and has recently entered a steady state. Adam's update rule is as follows: Assuming ε = 0, the effective descent step in time step t and parameter space is There are two supremums for the effective descending step size: that is, at /> In the case, the supremum of the effective step size satisfies /> and other cases where | _Δt |≤α is satisfied. The magnitude of the effective step size of each time step in the parameter space is approximately limited by the step size factor α, i.e.

An epoch process has only one iteration and one update data. Loss Function function formula:

Use the Loss function to update the formula of the weight parameter:

The prediction result error evaluation standard adopts the mean absolute percentage error (Mean Absolute Percent Error, MAPE), that is, the formula:

In the formula is the predicted value; y _i is the real value; n is the data sample size.

A network with a Bidirectional Long Short-Term Memory (BiLSTM) layer is optionally designed for fast estimation of the best path found. In order to generate realistic paths for training learning-based networks, BiLSTM provides extremely fast path planning for real-time drone applications. It is valuable for real-time autonomous UAV applications in complex and large environments because it provides new methods with fast path planning capabilities.

The output feature vector I _H of the information feature extraction network is used as the hidden state input of the sequence generation network, so that I _H can provide constraints for each generated sequence. Then set the initialization input x ₁ =N _U +1, and the network calculation is as follows (6).

x _t+1 =O _t

Where: W _z , W _r are network parameters, σ(·) is sigmoid. The entire calculation process of formula (6) can be expressed as a BiLSTM unit, the input data at time t is set to x _t , and the hidden state input is Where x _t is the result calculated by the BiLSTM unit at the previous moment, /> Hidden state output calculated by the BiLSTM unit at the previous moment. put x _t , /> Substituting into formula (6) to calculate the hidden state output at this moment /> contains all the reserved UAV preset path information before time t. Then the output O _t at time t is obtained.

Optionally, the features extracted by multiple BiLSTM units with the preset path information of the t-sequence UAV are input into the CNN network for optimal analysis of the preset path.

Optionally train a CNN model to fit the drone's behavioral policy. Moreover, since the dataset contains global environment information and optimal paths, the trained CNN model can estimate the distribution of environment information. The estimated information can be viewed as implementing local environmental information and target locations to infer more appropriate behaviors, thereby effectively mitigating first-time conflicts. And use the optimal path planning algorithm to calculate their optimal drone path. Then extract the state and position of the environmental information within the detection range at different times during the path planning process, and train the CNN model. When the state of a new scene is fed into the training model, the predicted behavior of the drone is treated as a guide to the heading direction in the case of real-time path planning. Various scenarios are first collected and their optimal UAV paths are calculated using an optimal path planning algorithm. Then extract the state and position of the environmental information within the detection range at different times during the path planning process, and train the CNN model. When the state of a new scene is fed into the training model, the predicted behavior of the drone is treated as a guide to the heading direction in the case of real-time path planning. Various scenarios are first collected and their optimal UAV paths are calculated using an optimal path planning algorithm. Then extract the state and position of the environmental information within the detection range at different times during the path planning process, and train the CNN model.

In some embodiments, the UAV path planning algorithm uses a network unit of a bidirectional long-short-term memory layer and a CNN network unit;

The network unit of the two-way long-short-term memory layer is used to extract the characteristics of the preset path information of the t-sequence UAV according to the location of the UAV cluster and the attribute information of the high-altitude sounding balloon that needs to be inspected;

The CNN network unit is used to determine the optimal path according to the characteristics of the preset path information of the t-sequence UAV.

Further, in some embodiments, the network unit of the two-way long-short-term memory layer is used to extract the preset path of the UAV with the t-sequence according to the location of the UAV cluster and the attribute information of the high-altitude sounding balloon that needs to be patrolled. Characteristics of the information, including:

Given the UAV sequence U={U ₁ , U ₂ ,..,U _NU } and the position information of the task sequence T={T ₁ ,T ₂ ,..,T _NT } to be executed, set The initial position information of the i-th UAV is The position of the jth task point is Then the path distance between each UAV and the mission point /> Expressed as:

In the formula: ||·|| represents the two norms: i=1,..., _NU , j=1,..., _NT ; _NU is the total number of UAVs, and _NT is the task point the total number of

The path distance Γ _m of all UAVs to each mission point is expressed as a matrix of N _U × N _T is the element in it, and m represents the location point of the current path.

Further, in some embodiments, the CNN network unit is used to determine the optimal path according to the characteristics of the preset path information of the t-sequence UAV, including: _JD is the loss function of the preset path, and the loss value The smaller the value, the better the preset path, which is defined as:

In the formula: β is the preset path loss coefficient, which is used to judge the training update coefficient of the evaluation model parameters; β=[μ ₁ ,μ ₂ ]∈R ^1×2 , μ ₁ represents the probability that the sequence is a generated sequence, μ ₂ represents the probability that the sequence is reserved for real, R represents all real numbers, Indicates the path distance between UAV U _i and the jth mission point; /> Indicates whether the UAV U _i executes the preset path distance />

Example 2

In the second aspect, based on Embodiment 1, this embodiment provides a UAV cluster control device, including a processor and a storage medium;

The storage medium is used to store instructions;

The processor is configured to operate according to the instructions to execute the method according to Embodiment 1.

In some embodiments, an unmanned aerial vehicle swarm control system includes the above-mentioned unmanned aerial vehicle swarm control device.

Further, the UAV swarm control system also includes: a high-altitude weather detection balloon, a ground positioning radar, and an inertial measurement unit;

The high-altitude weather detection balloon can monitor the wind speed and wind direction around the UAV cluster; the ground positioning radar can observe the flight position of the UAV cluster through the ground control center in real time. Timely detection; Inertial Measurement Unit (Inertial Measurement Unit), which measures the horizontal state, speed, and position information of the UAV during flight. When the UAV cluster deviates from the path, it can be controlled and corrected in time through the ground control center;

The ground console uses the control device to preset the UAV cluster to carry out flight missions. When the UAV cluster encounters strong winds during flight, the high-altitude weather detection balloon will transmit weather information such as wind speed, wind direction, temperature and humidity to the control device. The control device provides a reference. At the same time, when the transmitted wind force is greater than the maximum bearing range of the drone cluster, the control device will generate an alarm message to remind the staff to recall the drone cluster and stop the flight mission, effectively preventing the drone cluster In the case of accidental loss caused by forced flight, when the wind force sensed by the drone cluster has no effect on the flight of the drone, the staff can adjust the attitude, speed, and heading of the drone cluster through the ground control panel so that it can adapt Smooth flight with wind power.

At the same time, when the UAV cluster is out of the control of the control device during flight, the UAV automatically turns on the positioning information transmission, and the warning light is automatically triggered, and a sound and light alarm is issued, which is convenient for the staff to find it; when the UAV cluster is abnormal When falling, the airbag can be triggered to reduce the damage of the drone and improve the flight safety of the drone cluster.

Example 3

In the third aspect, based on Embodiment 1, this embodiment provides a device, including:

memory;

processor;

as well as

Computer program;

Wherein, the computer program is stored in the memory and is configured to be executed by the processor to implement the method described in Embodiment 1.

Example 4

In a fourth aspect, based on Embodiment 1, this embodiment provides a storage medium on which a computer program is stored, and when the computer program is executed by a processor, the method described in Embodiment 1 is implemented.

Those skilled in the art should understand that the embodiments of the present application may be provided as methods, systems, or computer program products. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.

The present application is described with reference to flowcharts and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present application. It should be understood that each procedure and/or block in the flowchart and/or block diagram, and a combination of procedures and/or blocks in the flowchart and/or block diagram can be realized by computer program instructions. These computer program instructions may be provided to a general purpose computer, special purpose computer, embedded processor, or processor of other programmable data processing equipment to produce a machine such that the instructions executed by the processor of the computer or other programmable data processing equipment produce a An apparatus for realizing the functions specified in one or more procedures of the flowchart and/or one or more blocks of the block diagram.

These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing apparatus to operate in a specific manner, such that the instructions stored in the computer-readable memory produce an article of manufacture comprising instruction means, the instructions The device realizes the function specified in one or more procedures of the flowchart and/or one or more blocks of the block diagram.

These computer program instructions can also be loaded onto a computer or other programmable data processing device, causing a series of operational steps to be performed on the computer or other programmable device to produce a computer-implemented process, thereby The instructions provide steps for implementing the functions specified in the flow chart or blocks of the flowchart and/or the block or blocks of the block diagrams.

The above is only a preferred embodiment of the present invention, it should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, some improvements and modifications can also be made, and these improvements and modifications are also possible. It should be regarded as the protection scope of the present invention.

Claims (10)

1. A method for unmanned aerial vehicle cluster control, is characterized in that, described method comprises: Obtain the UAV cluster information and the attribute information of the high-altitude sounding balloon that needs to be inspected; wherein the UAV cluster information includes the location of the UAV cluster; the attribute information includes the location of the high-altitude sounding balloon, inspection time, and image capture resolution And the information attributes of the stability of the UAV swarm during flight; Based on the UAV cluster information and the attribute information of the high-altitude sounding balloons that need to be inspected, the UAV path planning algorithm is used to generate the initial UAV inspection plan; With the number of iterations reaching the preset threshold of iterations as the goal, perform the following iterative steps to obtain multiple UAV inspection plans, wherein the iterative steps include: from the path of each UAV in the UAV inspection plan Randomly select a high-altitude sounding balloon to form a new high-altitude sounding balloon set; for the new high-altitude sounding balloon set, adopt the UAV path planning algorithm to generate a new UAV inspection plan; Calculate the evaluation value of each UAV inspection plan; UAV swarm control is performed according to the UAV inspection plan with the smallest evaluation value. 2. The unmanned aerial vehicle swarm control method according to claim 1, is characterized in that, described unmanned aerial vehicle path planning algorithm comprises: S1. Randomly select an unselected UAV cluster from the UAV collection; S2. Randomly select an unselected high-altitude sounding balloon from the set of high-altitude sounding balloons that is closest to the location of the selected UAV; if the selected UAV cluster cannot complete the inspection task of the selected high-altitude sounding balloon, restart Select unselected UAVs from the UAV swarm set until the selected UAV swarm can complete the inspection task of the selected high-altitude sounding balloon, and add the position of the selected high-altitude sounding balloon as the inspection point to the selected In the preset path of the UAV cluster; S3. If there are still unselected high-altitude sounding balloons, execute step S2 in a loop until there are no unselected high-altitude sounding balloons, and output the UAV inspection plan. 3. the unmanned aerial vehicle swarm control method according to claim 2, is characterized in that, whether the unmanned aerial vehicle swarm that judges choosing can finish the inspection task of the high-altitude sounding balloon of choosing, comprises: Whether the selected UAV cluster can complete the inspection task of the selected high-altitude sounding balloon needs to meet the following conditions at the same time: (1) The position information of the positioning radar carried by the drone cluster is higher than the position longitude and latitude information of the selected high-altitude sounding balloon; (2) The UAV swarm can reach the selected high-altitude exploration balloon from the current high-altitude exploration balloon and return to a stop. 4. The unmanned aerial vehicle swarm control method according to claim 1, is characterized in that, the computing method of described iteration times threshold Δn comprises: Among them, ρ is the preset ratio operator value, N is the number of paths of the UAV cluster, and d _n is the data of the high-altitude sounding balloon included in the path of the nth UAV. 5. The UAV swarm control method according to claim 1, wherein calculating the evaluation value of each UAV inspection scheme includes: Among them, S is the evaluation value of the UAV inspection plan, N is the number of UAV cluster paths in the UAV inspection plan, is the inspection energy consumption of the nth UAV in the UAV inspection scheme to complete the i-th inspection task, _ε1 is the weight of the inspection energy consumption, /> In the drone inspection scheme, the nth drone completes the inspection time of the i-th inspection task to be inspected, _ε2 is the weight of the inspection time, and I is the number of high-altitude sounding balloons. 6. according to claim 1 or 2 described unmanned aerial vehicle cluster control method, it is characterized in that, described unmanned aerial vehicle path planning algorithm adopts the network unit of two-way long short-term memory layer and CNN network unit; The network unit of the two-way long-short-term memory layer is used to extract the characteristics of the preset path information of the t-sequence UAV according to the location of the UAV cluster and the attribute information of the high-altitude sounding balloon that needs to be inspected; The CNN network unit is used to determine the optimal path according to the characteristics of the preset path information of the t-sequence UAV. 7. The UAV swarm control method according to claim 1, wherein the network unit of the two-way long-short-term memory layer is used for according to the UAV swarm position and the attribute information of the high-altitude sounding balloon that needs to be inspected Extract the features with the preset path information of the t-sequence UAV, including: Given the UAV sequence U={U ₁ , U ₂ ,..,U _NU } and the position information of the task sequence T={T ₁ ,T ₂ ,..,T _NT } to be executed, set The initial position information of the i-th UAV is The position of the jth task point is /> Then the path distance between each UAV and the mission point /> Expressed as: In the formula: ||·|| represents the two norms: i=1,..., _NU , j=1,..., _NT ; _NU is the total number of UAVs, and _NT is the task point the total number of The path distance Γ _m of all UAVs to each mission point is expressed as a matrix of N _U × N _T is the element in it, and m represents the location point of the current path. 8. The unmanned aerial vehicle swarm control method according to claim 7, is characterized in that, described CNN network unit, is used for according to the feature that has t sequence unmanned aerial vehicle preset path information, determines optimum path, comprises: J _D is the loss function of the preset path, the smaller the loss value is, the better the preset path is, defined as: In the formula: β is the preset path loss coefficient, which is used to judge the training update coefficient of the evaluation model parameters; β=[μ ₁ ,μ ₂ ]∈R ^{1 ×2} , μ ₁ represents the probability that the sequence is a generated sequence, μ ₂ represents the probability that the sequence is reserved for real, R represents all real numbers, Indicates the path distance between UAV U _i and the jth mission point; /> Indicates whether the UAV U _i executes the preset path distance /> 9. An unmanned aerial vehicle swarm control device, is characterized in that, comprises processor and storage medium; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the method according to any one of claims 1-8. 10. A storage medium, wherein a computer program is stored thereon, and when the computer program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.