Unmanned aerial vehicle cluster control method and system based on artificial intelligence

Abstract

The invention discloses an unmanned aerial vehicle cluster control method and system based on artificial intelligence, and relates to the technical field of unmanned aerial vehicle cluster control, and the method comprises the steps: obtaining global cooperative situation data through a sensor, and carrying out the preprocessing; calculating control output of each unmanned aerial vehicle by using a fuzzy C-means clustering algorithm and an analytic hierarchy process, generating a preliminary flight path, and dynamically adjusting the flight path according to real-time environment feedback through a reinforcement learning algorithm; the unmanned aerial vehicle cluster shares the adjusted flight path and global environment information through a 5G network, analyzes the speed, course angle, height, acceleration gradient and height change rate of each unmanned aerial vehicle by using a distributed control algorithm, and generates cooperative control parameters; according to the invention, by combining the fuzzy C-means clustering algorithm, the analytic hierarchy process, the reinforcement learning algorithm, the distributed control algorithm and the genetic algorithm, efficient cooperative control of the unmanned aerial vehicle cluster in a complex environment is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of UAV swarm control, and particularly to a UAV swarm control method and system based on artificial intelligence. Background Art

[0002] The core of UAV swarm control lies in realizing the cooperative operation of multiple UAVs in complex environments, including path planning, dynamic obstacle avoidance, task allocation, etc. In recent years, the introduction of artificial intelligence technology has provided new solutions for UAV swarm control. In addition, intelligent optimization algorithms such as reinforcement learning and genetic algorithms have also been widely applied to UAV swarm control to meet the requirements of real-time path optimization and cooperative decision-making in dynamic environments.

[0003] Traditional path planning algorithms (such as A* algorithm, Dijkstra algorithm) are difficult to handle the dynamic path adjustment of large-scale UAV swarms in real time in complex environments, resulting in low path planning efficiency and being prone to falling into local optimal solutions. When dealing with real-time environmental feedback in the prior art, it often relies on high-bandwidth and low-latency communication networks. However, in practical applications, the stability and bandwidth limitations of the communication network may lead to information transmission delays or losses, thereby affecting the real-time cooperative control effect of UAV swarms. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a UAV swarm control method based on artificial intelligence to solve the problem that traditional path planning algorithms are difficult to handle the dynamic path adjustment of large-scale UAV swarms in real time in complex environments.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: In the first aspect, the present invention provides a UAV swarm control method based on artificial intelligence, which includes obtaining global cooperative situation data through sensors and performing preprocessing; Calculating the control output of each UAV using the fuzzy C-means clustering algorithm and the analytic hierarchy process, generating a preliminary flight path, and dynamically adjusting the flight path according to real-time environmental feedback through the reinforcement learning algorithm; The UAV swarm shares the adjusted flight path and global environmental information through the 5G network, analyzes the speed, heading angle, altitude, acceleration gradient, and altitude change rate of each UAV using a distributed control algorithm, and generates cooperative control parameters; Optimizing the cooperative control parameters through the genetic algorithm to obtain the optimal flight path, and generating cooperative control instructions through the 5G network and edge computing method and distributing them to the UAV swarm in real time.

[0007] As a preferred solution of the artificial intelligence-based UAV swarm control method described in the present invention, wherein: obtaining the global collaborative situation data through sensors and performing preprocessing, the specific steps are as follows, Collect environmental perception data, UAV state data, and task and collaboration data using sensors to form global collaborative situation data; Through the coordinate transformation algorithm and timestamp synchronization method, unify the global collaborative situation data to the global coordinate system for time synchronization; Use the wavelet transform method to remove high-frequency noise in the global collaborative situation data, use nearest neighbor interpolation to fill in missing values, and perform normalization processing through the Z-score normalization method.

[0008] As a preferred solution of the artificial intelligence-based UAV swarm control method described in the present invention, wherein: calculating the control output of each UAV using the fuzzy C-means clustering algorithm and the analytic hierarchy process to generate a preliminary flight path, the specific steps are as follows, Based on the preprocessed global collaborative situation data, extract path correlation features through principal component analysis for dimensionality reduction and combined with mutual information evaluation; Set the number of cluster centers through the elbow method, set the fuzzy factor based on empirical values, and combine the path correlation features to calculate the membership degree of each UAV to each cluster center; Recalculate the cluster centers based on the membership degrees and perform iterative optimization until the change in the cluster centers is less than the convergence threshold to obtain the cluster area to which each UAV belongs; Construct an objective layer, a criterion layer, and a scheme layer through the analytic hierarchy process, convert task requirements, UAV states, and environmental constraints into quantitative scores, and combine the cluster area to which each UAV belongs to finally generate a preliminary flight path for the UAV swarm.

[0009] As a preferred solution of the artificial intelligence-based UAV swarm control method described in the present invention, wherein: dynamically adjusting the flight path according to the real-time environment feedback through the reinforcement learning algorithm, the specific steps are as follows, Collect terrain change data and meteorological data to obtain real-time environment feedback; Based on the environmental perception data, UAV state data, and task and collaboration data, define the state space and action space, and design a reward function through a multi-objective optimization method to form a reinforcement learning framework; Based on the real-time environment feedback, dynamically adjust the preliminary flight path through the reinforcement learning framework combined with the proximal policy optimization algorithm.

[0010] As a preferred solution of the UAV swarm control method based on artificial intelligence according to the present invention, wherein: the UAV swarm shares the adjusted flight path and global environment information through a 5G network, and analyzes the speed, heading angle, altitude, acceleration gradient, and altitude change rate of each UAV by using a distributed control algorithm to generate cooperative control parameters. The specific steps are as follows. Through the cubature Kalman filter algorithm, data fusion is performed on the preprocessed global cooperative situation data, and combined with principal component analysis and D-S evidence theory, global environment information is generated. The adjusted flight path and global environment information are encapsulated into a structured data packet, binary encoded using the Protobuf protocol, and compressed into terrain grid data through Huffman coding. Based on the terrain grid data, the TSN proxy node dynamically allocates transmission time slots according to the UAV service type, and performs microsecond-level clock synchronization using the NTP-Precise protocol. Through the visual odometer combined with the Kalman filter method, the speed, heading angle, altitude, acceleration gradient, and altitude change rate of each UAV are extracted to construct a six-dimensional state vector. The six-dimensional state vectors of adjacent UAVs within a radius R are obtained through 5G D2D communication, and the distributed solution is performed using the alternating direction method of multipliers combined with penalty function linearization to obtain the cooperative control parameters of each UAV.

[0011] As a preferred solution of the UAV swarm control method based on artificial intelligence according to the present invention, wherein: the cooperative control parameters are optimized through a genetic algorithm to obtain the optimal flight path. The specific steps are as follows. Using the real number coding method, the cooperative control parameter coding is segmented by time slice, mapped into a multi-dimensional real number gene sequence, and the initial population is generated using the Latin hypercube sampling method. Through tournament selection, simulated binary crossover, and oriented polynomial mutation, the initial population is iteratively optimized until it converges to the maximum number of iterations to obtain the optimal chromosome. The optimal chromosome is analyzed by the spatio-temporal mapping decoding method by time slice to obtain the time series path points, and the feasibility is verified by combining the dynamic model and reinforcement learning, and the optimal flight path is output.

[0012] As a preferred solution of the UAV swarm control method based on artificial intelligence according to the present invention, wherein: the cooperative control instructions are generated through a 5G network and edge computing method and are distributed to the UAV swarm in real time. The specific steps are as follows. Construct a nested JSON structure, use Python to serialize the optimal flight path and the collaborative control parameter encapsulation into a JSON format instruction, and use the AES-128 algorithm to generate a byte key to encrypt the JSON format instruction in CBC mode; Through the 5G network slicing method combined with the TSN proxy node to dynamically allocate transmission time slots, and distribute the encrypted JSON format instructions to the UAV cluster in real time.

[0013] In a second aspect, the present invention provides an unmanned aerial vehicle (UAV) cluster control system based on artificial intelligence, including a data acquisition module, a path planning module, a control parameter module, and an instruction distribution module; The data acquisition module is used to obtain the global collaborative situation data through sensors and perform preprocessing; The path planning module is used to calculate the control output of each UAV by using the fuzzy C-means clustering algorithm and the analytic hierarchy process, generate a preliminary flight path, and dynamically adjust the flight path according to the real-time environment feedback through the reinforcement learning algorithm; The control parameter module is used for the UAV cluster to share the adjusted flight path and the global environment information through the 5G network, analyze the speed, heading angle, altitude, acceleration gradient, and altitude change rate of each UAV by using the distributed control algorithm, and generate collaborative control parameters; The instruction distribution module is used to optimize the collaborative control parameters through the genetic algorithm, obtain the optimal flight path, generate collaborative control instructions through the 5G network and the edge computing method, and distribute them to the UAV cluster in real time.

[0014] In a third aspect, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and: when the computer program is executed by the processor, any step of the method for controlling a UAV cluster based on artificial intelligence as described in the first aspect of the present invention is implemented.

[0015] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and: when the computer program is executed by the processor, any step of the method for controlling a UAV cluster based on artificial intelligence as described in the first aspect of the present invention is implemented.

[0016] The beneficial effects of the present invention are as follows: By combining the fuzzy C-means clustering algorithm, the analytic hierarchy process, the reinforcement learning algorithm, the distributed control algorithm, and the genetic algorithm, the efficient collaborative control of the UAV cluster in a complex environment is realized, and the problems that traditional path planning algorithms are difficult to process the dynamic path adjustment of a large-scale UAV cluster in real time, the lack of optimization of collaborative parameters, and communication delay are solved. The path planning efficiency, global collaborative ability, and task completion degree of the UAV cluster are significantly improved, providing comprehensive technical support for real-time collaborative operations in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following-described drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0018] Figure 1 It is a flowchart of the method for controlling an unmanned aerial vehicle (UAV) cluster based on artificial intelligence in Embodiment 1.

[0019] Figure 2 It is a schematic diagram of the UAV cluster control system based on artificial intelligence in Embodiment 1.

[0020] Figure 3 It is a schematic diagram of the process for generating and optimizing cooperative control parameters in Embodiment 1.

[0021] Figure 4 It is a schematic diagram of the process for path planning and dynamic adjustment of the UAV cluster in Embodiment 1. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will provide a detailed description of the specific embodiments of the present invention with reference to the drawings of the specification.

[0023] In the following description, many specific details are set forth to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0024] Secondly, the so-called "one embodiment" or "embodiment" herein refers to specific features, structures, or characteristics that can be included in at least one implementation manner of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that is mutually exclusive with other embodiments.

[0025] Embodiment 1, referring to Figures 1 to 4 , is the first embodiment of the present invention. This embodiment provides a method for controlling an unmanned aerial vehicle (UAV) cluster based on artificial intelligence, including the following steps: S1. Obtain the global cooperative situation data through sensors and perform preprocessing.

[0026] Collect environmental perception data, UAV state data, and task and cooperation data by using sensors to form global cooperative situation data.

[0027] Through the coordinate transformation algorithm and timestamp synchronization method, the global collaborative situation data is unified to the global coordinate system for time synchronization.

[0028] It should be noted that, first, the global collaborative situation data is transformed to the unified global coordinate system by using the coordinate transformation algorithm (such as Euler angle or quaternion transformation); second, the global collaborative situation data is time-aligned through the timestamp synchronization method (such as interpolation or nearest neighbor matching) to ensure that all global collaborative situation data is consistent under the same time reference; finally, the spatio-temporal unity of the global collaborative situation data in the global coordinate system is achieved.

[0029] The wavelet transform method is used to remove the high-frequency noise in the global collaborative situation data, the nearest neighbor interpolation is adopted to fill in the missing values, and the normalization process is carried out by the Z-score normalization method.

[0030] It should be noted that, first, the appropriate wavelet basis function and decomposition level are selected to perform wavelet decomposition on the global collaborative situation data; second, the noise in the high-frequency components is filtered out through threshold processing, and the effective low-frequency information is retained; then, the nearest neighbor interpolation method is used to fill in the missing values in the global collaborative situation data to ensure the continuity of the global collaborative situation data; finally, the global collaborative situation data is normalized by the Z-score normalization method to convert the data into a distribution with a mean of 0 and a standard deviation of 1, ensuring analysis under the same scale.

[0031] S2. Use the fuzzy C-means clustering algorithm and the analytic hierarchy process to calculate the control output of each UAV and generate a preliminary flight path.

[0032] Based on the preprocessed global collaborative situation data, the path correlation features are extracted by principal component analysis for dimensionality reduction and combined with mutual information evaluation.

[0033] It should be noted that, first, principal component analysis (PCA) is used to reduce the dimensionality of the high-dimensional preprocessed global collaborative situation data, retaining the main components to reduce the computational complexity; second, the relevance of each principal component to the path planning objective is evaluated through mutual information, and the low-dimensional features strongly related to the path planning are screened out, and finally the path correlation features are obtained.

[0034] The number of clustering centers is set by the elbow method, the fuzzy factor is set based on the empirical value, and combined with the path correlation features, the membership degree of each UAV to each clustering center is calculated.

[0035] Specifically as follows: For the path association feature, calculate the sum of squared errors (SSE) of clustering under different numbers of cluster centers (ranging from 1 to a preset maximum value), plot the curve of SSE varying with the number of cluster centers, observe the inflection point (i.e., the elbow) of the curve, and select the number of cluster centers corresponding to the inflection point as the optimal value; secondly, set the fuzzy factor based on empirical values (usually from 1.5 to 2.5) to control the fuzziness of clustering; then, combined with the path association feature, use the fuzzy C-means clustering algorithm to calculate the membership degree of each UAV to each cluster center; Calculate the membership degree of each UAV to each cluster center, and the expression is: ; where, is the membership degree of the th UAV belonging to the th cluster center, is the path association feature of the th UAV, is the th cluster center, is the th cluster center, is the fuzzy factor, is the index variable of the cluster center, is the number of cluster centers; It should be noted that represents the current cluster center index (the th cluster center), represents the index for traversing all cluster centers (from 1 to ).

[0036] Recalculate the cluster centers based on the membership degrees and perform iterative optimization until the change in the cluster centers is less than the convergence threshold to obtain the cluster regions to which each UAV belongs.

[0037] It should be noted that according to the current membership degrees and path association features, check whether the change amount of the cluster centers is less than a preset convergence threshold (such as 10-5); if the condition is not met, repeat the process of calculating the membership degree of each UAV to each cluster center until the change amount of the cluster centers is less than the convergence threshold; finally, according to the final membership degree distribution, divide each UAV into the cluster region with the highest membership degree; It should also be noted that first, initially set a relatively small convergence threshold (such as 10-5 or 10-6) to ensure that the change in the cluster centers is small enough. Through verification, observe the convergence speed and clustering effect of the fuzzy C-means clustering algorithm under different convergence thresholds, find a balance between accuracy and computational cost, and finally, combined with empirical values, determine a convergence threshold that can ensure both clustering quality and avoid excessive iteration.

[0038] Construct the target layer, criterion layer and solution layer through the Analytic Hierarchy Process (AHP), convert the task requirements, UAV states and environmental constraints into quantitative scores, and combine with the clustering regions to which each UAV belongs to finally generate the preliminary flight paths of the UAV swarm.

[0039] It should be noted that through the AHP, construct a hierarchical decision-making model structure of the target layer, criterion layer and solution layer, clarify that the overall goal is to "generate the preliminary flight paths of the UAV swarm", and decompose the key factors affecting path planning into criteria such as task requirements (such as task priority, target location), UAV state data (such as power, speed, load), and environmental constraints (such as terrain, meteorology, obstacles). At the same time, generate possible flight path solutions for each UAV as alternative solutions to form a complete AHP framework. Then, convert the task requirements, UAV states and environmental constraints into quantitative scores, construct a judgment matrix through the pairwise comparison method, calculate the weights of each criterion and ensure that the consistency test passes; combine with the clustering regions to which each UAV belongs, comprehensively evaluate the task requirements, UAV states and environmental constraints of each clustering region, select the optimal path solution for each UAV, and finally generate the preliminary flight paths of the UAV swarm. It should also be noted that the conversion into quantitative scores is as follows: Define the scoring criteria for specific indicators under each criterion. For example, the task priority is divided into high, medium and low, corresponding to scores of 3, 2, and 1 respectively; the UAV power is scored as a percentage, with full power being 100 points, and 10 points deducted for every 10% reduction; the terrain complexity is scored according to the obstacle density, with no obstacles being 100 points, and 20 points deducted for each additional obstacle. Secondly, through empirical values, specific scores are given to each indicator. Finally, combine the weights of each indicator (determined by the AHP) to calculate the comprehensive score, and unify the conversion of task requirements, UAV states and environmental constraints into comparable quantitative scores.

[0040] S3. Dynamically adjust the flight paths according to the real-time environmental feedback through the reinforcement learning algorithm.

[0041] Collect terrain change data and meteorological data to obtain real-time environmental feedback.

[0042] Based on the environmental perception data, UAV state data, and task and collaboration data, define the state space and action space, and design the reward function through the multi-objective optimization method to form a reinforcement learning framework.

[0043] It should be noted that, first, environmental perception data (such as terrain changes, meteorological information), UAV status data (such as battery power, speed, load), and mission and cooperation data (such as mission objectives, cooperation strategies) are integrated into a multi-dimensional state vector to define the state space; second, the possible actions of the UAV (such as acceleration, turning, altitude adjustment) are defined as the action space; then, a reward function is defined through a multi-objective optimization method, comprehensively considering multiple objectives such as mission completion, energy consumption, and obstacle avoidance effect, and corresponding weights are assigned to different objectives to ensure that the reward function can comprehensively reflect the advantages and disadvantages of the UAV's actions; finally, the state space, action space, and reward function are combined to construct a complete reinforcement learning framework.

[0044] Based on real-time environmental feedback, the initial flight path is dynamically adjusted through the reinforcement learning framework combined with the proximal policy optimization algorithm.

[0045] It should be noted that, first, environmental perception data (such as terrain changes, meteorological information) is obtained through multi-source sensors, integrated with UAV status data (such as battery power, speed, load) and mission data (such as target location, mission priority) into the current state, and input into the reinforcement learning framework; second, the proximal policy optimization (PPO) algorithm is used to generate an optimized action policy (such as adjusting speed, heading angle, altitude) according to the current state and the reward function (comprehensively considering objectives such as mission completion, energy consumption, and obstacle avoidance effect); then, the optimized policy is executed to update the UAV's flight path, and at the same time, new environmental feedback data is continuously collected to iteratively optimize the policy until the mission is completed or a preset condition is reached, and finally, the dynamic adjustment of the UAV swarm flight path is realized.

[0046] S4. The UAV swarm shares the adjusted flight path and global environmental information through the 5G network, and uses a distributed control algorithm to analyze the speed, heading angle, altitude, acceleration gradient, and altitude change rate of each UAV to generate cooperative control parameters.

[0047] Through the cubature Kalman filter algorithm, data fusion is performed on the preprocessed global cooperative situation data, and combined with principal component analysis and D-S evidence theory to generate global environmental information.

[0048] It should be noted that the system state is predicted through non-linear state equations and observation equations, and the state prediction value and error covariance are calculated using Cubature Points. Then, combined with the observed values, the state estimation and covariance matrix are updated to achieve the unification and noise elimination of multi-source data in space and time. The covariance matrix of the data is calculated, the eigenvalues and eigenvectors are extracted, the principal component with the largest contribution rate is selected, and the original data is projected into a low-dimensional space to retain the main features and reduce redundant information. A Frame of Discernment is defined, and a Basic Probability Assignment (BPA) function is assigned to each evidence source. The BPA functions of multi-source evidence are fused through the Dempster combination rule, the confidence intervals and uncertainty values of each hypothesis are calculated, and an uncertainty model is constructed based on these results. Finally, the fused state estimation, dimension-reduced features, and uncertainty model are integrated to generate high-precision and low-uncertainty global environmental information; It should also be noted that the uncertainty model is constructed through the D-S evidence theory, which is used to quantify the confidence intervals and uncertainties of each hypothesis in the multi-source data fusion process. It is calculated through the Basic Probability Assignment (BPA) function and the Dempster combination rule, and finally generates high-precision and low-uncertainty global environmental information.

[0049] The adjusted flight path and the global environmental information are encapsulated into a structured data packet, binary encoded using the Protobuf protocol, and compressed into terrain grid data through Huffman coding.

[0050] It should be noted that, first, the adjusted flight path and the global environmental information are encapsulated in a predefined structured format (such as JSON or a custom data structure) to ensure the integrity and logical relevance of the fields. Second, the structured data packet is serialized into a binary format using the Protobuf protocol. The data structure is described by a predefined proto file, and the fields are converted into a compact binary stream using the efficient encoding mechanism of Protobuf, significantly reducing the volume. Then, Huffman coding is performed on the binary data. By statistically analyzing the frequency distribution of the data, an optimal prefix code table is constructed, with high-frequency data represented by short codes and low-frequency data represented by long codes, further compressing the data volume. Finally, the compressed terrain grid data is generated.

[0051] Based on the terrain grid data, the TSN proxy node dynamically allocates transmission time slots according to the UAV service type, and the NTP-Precise protocol is used for microsecond-level clock synchronization.

[0052] It should be noted that, first, the TSN proxy node dynamically allocates transmission time slots according to the priorities and bandwidth requirements of UAV service types (such as real-time control, data transmission, environmental monitoring), and allocates high-priority services (such as real-time control) to fixed time slots through a Time-Aware Shaper, and low-priority services (such as data transmission) to the remaining time slots to ensure efficient transmission of services; second, the NTP-Precise protocol is used for microsecond-level clock synchronization. Through the master-slave clock architecture, the master node sends timestamp packets, and the slave node receives and calculates the clock deviation, smooths the clock error using a filter, and achieves microsecond-level precision synchronization through frequency adjustment and phase compensation.

[0053] By combining visual odometry with the Kalman filter method, the speed, heading angle, altitude, acceleration gradient, and altitude change rate of each UAV are extracted to construct a six-dimensional state vector.

[0054] It should be noted that, first, continuous frame images are extracted from the cameras or visual sensors carried by the UAVs using visual odometry, and the relative displacement and attitude changes of the UAVs are calculated through feature point matching and motion estimation to initially obtain speed, heading angle, and altitude information; second, these observed values are input into the Kalman filter, and through two steps of state prediction (based on the motion model) and state update (combining the observed values), the speed, heading angle, altitude, acceleration gradient, and altitude change rate of the UAVs are optimized and estimated, noise is eliminated, and the accuracy is improved; finally, these optimized state variables are integrated into a six-dimensional state vector (including speed, heading angle, altitude, acceleration gradient, altitude change rate, and timestamp).

[0055] The six-dimensional state vectors of adjacent UAVs within a radius R are obtained through 5G D2D communication, and the alternating direction method of multipliers is combined with penalty function linearization for distributed solution to obtain the cooperative control parameters of each UAV.

[0056] It should be noted that each UAV exchanges six-dimensional state vectors (including speed, heading angle, altitude, acceleration gradient, altitude change rate, and timestamp) with adjacent UAVs in real time within a radius R through 5G D2D communication to construct local network state information, and models the cooperative control problem as a distributed optimization problem. The goal is to minimize the overall energy consumption and path deviation of the UAV cluster while satisfying constraints such as obstacle avoidance and task completion. The alternating direction method of multipliers (ADMM) is used to decompose the global optimization problem into multiple sub-problems. Each UAV independently solves the sub-problems based on local state information, introduces Lagrange multipliers and penalty function linearization to handle the constraints, updates the local solution and Lagrange multipliers in each iteration, and exchanges intermediate results with adjacent UAVs through communication to gradually approach the global optimal solution. Each UAV generates cooperative control parameters (such as speed, heading angle, altitude adjustment amount) according to the solution results.

[0057] S5. Optimize the cooperative control parameters through a genetic algorithm to obtain the optimal flight path.

[0058] Adopt a real - number coding method. Segment the coding of the cooperative control parameters by time slices and map them into a multi - dimensional real - valued gene sequence. Then use the Latin hypercube sampling method to generate the initial population.

[0059] It should be noted that the cooperative control parameters (such as speed, heading angle, altitude adjustment amount, etc.) are segmented by time slices, and each time slice corresponds to a UAV control parameter vector , where is the UAV control parameter vector corresponding to the th time slice, is the speed change amount in the th time slice (the unit is generally m / s), is the heading angle change amount within the th time slice (the unit is generally degrees or radians), is the altitude change amount within the th time slice (the unit is generally meters), is the acceleration within the th time slice (the unit is generally m / s²), is the vertical altitude change rate in the th time slice (the unit is generally m / s), is the transpose symbol; arranged in chronological order as a multi - dimensional real - valued gene sequence , where is the multi - dimensional real - valued gene sequence, is the UAV control parameter vector corresponding to the first time slice, is the UAV control parameter vector corresponding to the second time slice, is the UAV control parameter vector corresponding to the th time slice, is the total number of time slices; secondly, use the Latin hypercube sampling method to generate the initial population. Uniformly divide the intervals within the domain of each dimension parameter to ensure that each interval is sampled only once, generate uniformly distributed sample points, and randomly combine these sample points into multi - dimensional gene sequences to form the initial population.

[0060] Iteratively optimize the initial population through tournament selection, simulated binary crossover, and directional polynomial mutation until convergence to the maximum number of iterations to obtain the optimal chromosome.

[0061] Specifically as follows: The tournament selection method is used to screen high-quality individuals from the initial population. Several individuals (such as 2 or 3) are randomly selected for comparison, and the individual with the highest fitness is selected as the parental individual to enter the next generation. The selected parental individuals are subjected to simulated binary crossover (SBX), and through the control of the crossover probability, offspring individuals are generated. The expression is as follows: ; ; where is the gene value of the first offspring individual, is the gene value of the second offspring individual, is the gene value of the first parental individual, is the gene value of the second parental individual, is the crossover distribution index; The offspring individuals are subjected to orientation polynomial mutation, and through the control of the mutation probability, mutant individuals are generated. The expression is as follows: ; where is the new gene value of the mutant offspring individual, is the upper limit of the gene value, is the lower limit of the gene value, is the mutation distribution index, is the gene value of the offspring individual; Repeat the above process until the maximum number of iterations is reached, and select the chromosome with the highest fitness as the optimal solution.

[0062] The optimal chromosome is parsed according to time slices through the space-time mapping decoding method to obtain the time series path points, and the feasibility is verified by combining the dynamic model and reinforcement learning, and the optimal flight path is output.

[0063] Specifically as follows: The optimal chromosome is parsed by time slices. Each time slice corresponds to a control parameter vector, and the control parameters are mapped to a set of time series path points through integral operation , where is the set of time series path points, is the first time series path point, is the second time series path point, is the th time series path point, is the length of the time series; verify the feasibility of the waypoints in combination with the UAV dynamics model (such as the six-degree-of-freedom model) to ensure that the path meets the speed, acceleration, and attitude constraints. For example, check whether the speed is within the maximum speed range and whether the heading angle is within the steering angle limit; then, further optimize the waypoints using the reinforcement learning framework and dynamically adjust the waypoints to improve performance; finally, output the verified and optimized optimal flight path.

[0064] S6. Generate collaborative control instructions through the 5G network and edge computing methods and distribute them to the UAV cluster in real time.

[0065] Construct a nested JSON structure, serialize the optimal flight path and collaborative control parameters into JSON format instructions using Python, and generate a byte key using the AES-128 algorithm to encrypt the JSON format instructions in CBC mode.

[0066] It should be noted that a nested JSON structure is constructed, the optimal flight path and collaborative control parameters (such as speed, heading angle, altitude adjustment amount, etc.) are encapsulated into a nested dictionary, the nested dictionary is serialized into a JSON format string using Python, a 16-byte random key is generated, the JSON string is encrypted using the cryptography library of Python with AES-128, in CBC mode (Cipher Block Chaining), the JSON string is padded to a multiple of 16 bytes, a random initialization vector is generated, and the data is encrypted in blocks using the key and the random initialization vector, and finally the encrypted byte stream is output.

[0067] Distribute the encrypted JSON format instructions to the UAV cluster in real time through the 5G network slicing method in combination with the TSN proxy node to dynamically allocate transmission time slots.

[0068] It should be noted that the 5G network slicing technology is used to create a dedicated slice for the UAV cluster, allocate high-priority and low-latency network resources to ensure the real-time and reliable transmission of instructions. The TSN proxy node dynamically allocates transmission time slots according to the UAV service type (such as real-time control, data transmission), and distributes the encrypted JSON format instructions to fixed time slots through the Time-Aware Shaper to ensure the priority transmission of high-priority instructions; then, the encrypted instructions are sent to the UAV cluster through the 5G slice network.

[0069] This embodiment also provides an AI-based UAV cluster control system, including: a data acquisition module, a path planning module, a control parameter module, and an instruction distribution module; The data acquisition module is used to obtain the full-domain collaborative situation data through sensors and perform preprocessing; A path planning module, which is used to calculate the control output of each UAV by using the fuzzy C-means clustering algorithm and the analytic hierarchy process, generate a preliminary flight path, and dynamically adjust the flight path according to the real-time environmental feedback through the reinforcement learning algorithm; A control parameter module, which is used for the UAV cluster to share the adjusted flight path and the global environmental information through the 5G network, analyze the speed, heading angle, altitude, acceleration gradient and altitude change rate of each UAV by using the distributed control algorithm, and generate cooperative control parameters; An instruction distribution module, which is used to optimize the cooperative control parameters through the genetic algorithm, obtain the optimal flight path, generate cooperative control instructions through the 5G network and the edge computing method, and distribute them to the UAV cluster in real time.

[0070] This embodiment also provides a computer device, which is applicable to the situation of the UAV cluster control method based on artificial intelligence, and includes: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the UAV cluster control method based on artificial intelligence proposed in the above embodiment.

[0071] This computer device can be a terminal. The computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of this computer device is used to provide computing and control capabilities. The memory of this computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of this computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a carrier network, NFC (near field communication) or other technologies. The display screen of this computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of this computer device can be a touch layer covered on the display screen, or a button, a trackball or a touchpad provided on the housing of the computer device, or an external keyboard, a touchpad or a mouse, etc.

[0072] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the method for realizing the control of an unmanned aerial vehicle (UAV) cluster based on artificial intelligence as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM for short), electrically erasable programmable read-only memory (EEPROM for short), erasable programmable read-only memory (EPROM for short), programmable read-only memory (PROM for short), read-only memory (ROM for short), magnetic memory, flash memory, a magnetic disk or an optical disc.

[0073] In summary, the present invention combines the fuzzy C-means clustering algorithm, the analytic hierarchy process, the reinforcement learning algorithm, the distributed control algorithm, and the genetic algorithm to achieve the efficient cooperative control of UAV clusters in complex environments, solve the problems that traditional path planning algorithms are difficult to handle the dynamic path adjustment of large-scale UAV clusters in real time, have insufficient cooperative parameter optimization, and communication delays, significantly improve the path planning efficiency, global cooperative ability, and task completion rate of UAV clusters, and provide comprehensive technical support for real-time cooperative operations in complex environments.

[0074] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. An artificial intelligence-based method for controlling a drone swarm, characterized in that: including, acquiring the whole-domain collaborative situation data through sensors and performing preprocessing; calculating the control outputs of each unmanned aerial vehicle (UAV) using the fuzzy C-means clustering algorithm and the analytic hierarchy process (AHP), generating a preliminary flight path, and dynamically adjusting the flight path according to real-time environmental feedback through the reinforcement learning algorithm; the UAV swarm shares the adjusted flight path and the global environmental information through the 5G network, analyzes the speed, heading angle, altitude, acceleration gradient, and altitude change rate of each UAV using the distributed control algorithm, and generates collaborative control parameters; optimizing the collaborative control parameters through the genetic algorithm to obtain the optimal flight path, generating collaborative control instructions through the 5G network and edge computing methods, and distributing them to the UAV swarm in real time.

2. The method for controlling a drone swarm based on artificial intelligence according to claim 1, wherein: The step of acquiring the whole-domain collaborative situation data through sensors and performing preprocessing is as follows. Collecting environmental perception data, UAV state data, and task and collaboration data using sensors to form the whole-domain collaborative situation data; Unifying the whole-domain collaborative situation data to the global coordinate system for time synchronization through the coordinate transformation algorithm and the timestamp synchronization method; Removing high-frequency noise in the whole-domain collaborative situation data using the wavelet transform method, filling in missing values using the nearest neighbor interpolation method, and performing normalization processing through the Z-score normalization method.

3. The method for controlling a drone swarm based on artificial intelligence according to claim 1, characterized in that: The step of calculating the control outputs of each UAV using the fuzzy C-means clustering algorithm and the AHP to generate a preliminary flight path is as follows. Based on the preprocessed whole-domain collaborative situation data, reducing the dimension through principal component analysis and combining mutual information evaluation to extract path correlation features; Setting the number of clustering centers through the elbow method, setting the fuzzy factor based on empirical values, and combining the path correlation features to calculate the membership degree of each UAV to each clustering center; Recalculating the clustering centers based on the membership degree and performing iterative optimization until the change in the clustering centers is less than the convergence threshold to obtain the clustering regions to which each UAV belongs; Constructing the goal layer, criterion layer, and scheme layer through the AHP, converting the task requirements, UAV states, and environmental constraints into quantitative scores, and combining the clustering regions to which each UAV belongs to finally generate the preliminary flight path of the UAV swarm.

4. The method for controlling a drone swarm based on artificial intelligence according to claim 1, characterized in that: The step of dynamically adjusting the flight path according to real-time environmental feedback through the reinforcement learning algorithm is as follows. Collecting terrain change data and meteorological data to obtain real-time environmental feedback; Defining the state space and action space based on the environmental perception data, UAV state data, and task and collaboration data, and designing the reward function through the multi-objective optimization method to form the reinforcement learning framework; Based on the real-time environmental feedback, dynamically adjusting the preliminary flight path through the reinforcement learning framework combined with the proximal policy optimization algorithm.

5. The method for controlling a drone swarm based on artificial intelligence according to claim 4, wherein: The step of the UAV swarm sharing the adjusted flight path and the global environmental information through the 5G network, analyzing the speed, heading angle, altitude, acceleration gradient, and altitude change rate of each UAV using the distributed control algorithm, and generating collaborative control parameters is as follows. Performing data fusion on the preprocessed whole-domain collaborative situation data through the cubature Kalman filter algorithm, and generating the global environmental information by combining principal component analysis and D-S evidence theory. Encapsulate the adjusted flight path and global environmental information into a structured data packet, perform binary encoding using the Protobuf protocol, and compress it into terrain grid data through Huffman coding; Based on the terrain grid data, the TSN proxy node dynamically allocates transmission time slots according to the UAV service type, and uses the NTP-Precise protocol for microsecond-level clock synchronization; Extract the speed, heading angle, altitude, acceleration gradient, and altitude change rate of each UAV through visual odometry combined with the Kalman filtering method, and construct a six-dimensional state vector; Obtain the six-dimensional state vectors of adjacent UAVs within a radius R through 5G D2D communication, and use the alternating direction multiplier method combined with penalty function linearization for distributed solution to obtain the cooperative control parameters of each UAV.

6. The method for controlling a drone swarm based on artificial intelligence according to claim 5, wherein: Optimize the cooperative control parameters through the genetic algorithm to obtain the optimal flight path. The specific steps are as follows: Adopt the real number coding method, segment the encoding of the cooperative control parameters by time slice, map them into a multi-dimensional real number gene sequence, and use the Latin hypercube sampling method to generate the initial population; Through tournament selection, simulated binary crossover, and oriented polynomial mutation, iterate and optimize the initial population until it converges to the maximum number of iterations to obtain the optimal chromosome; Parse the optimal chromosome by time slice through the space-time mapping decoding method to obtain the time series path points, and verify the feasibility by combining the dynamic model and reinforcement learning, and output the optimal flight path.

7. The method for controlling an unmanned aerial vehicle cluster based on artificial intelligence according to claim 6, wherein: Generate cooperative control instructions through the 5G network and edge computing method and distribute them to the UAV cluster in real time. The specific steps are as follows: Construct a nested JSON structure, use Python to serialize the optimal flight path and cooperative control parameter encapsulation into JSON format instructions, and use the AES-128 algorithm to generate a byte key to encrypt the JSON format instructions in CBC mode; Through the 5G network slicing method combined with the TSN proxy node to dynamically allocate transmission time slots, and distribute the encrypted JSON format instructions to the UAV cluster in real time.

8. An artificial intelligence-based unmanned aerial vehicle (UAV) swarm control system, based on the artificial intelligence-based UAV swarm control method according to any one of claims 1 to 7, characterized in that: Including a data acquisition module, a path planning module, a control parameter module, and an instruction distribution module; The data acquisition module is used to obtain the global cooperative situation data through sensors and perform preprocessing; The path planning module is used to calculate the control output of each UAV using the fuzzy C-means clustering algorithm and the analytic hierarchy process, generate a preliminary flight path, and dynamically adjust the flight path according to the real-time environment feedback through the reinforcement learning algorithm; The control parameter module is used for the UAV cluster to share the adjusted flight path and global environmental information through the 5G network, analyze the speed, heading angle, altitude, acceleration gradient, and altitude change rate of each UAV using the distributed control algorithm, and generate cooperative control parameters; The instruction distribution module is used to optimize the cooperative control parameters through the genetic algorithm to obtain the optimal flight path, generate cooperative control instructions through the 5G network and edge computing method, and distribute them to the UAV cluster in real time.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the method for controlling an unmanned aerial vehicle cluster based on artificial intelligence according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, it implements the steps of the artificial intelligence-based UAV swarm control method according to any one of claims 1 to 7.

Images