A leader filtering method for a layered cooperative navigation system of a UAV cluster

Abstract

The application relates to the technical field of unmanned aerial vehicle navigation, and discloses a long aircraft screening method for an unmanned aerial vehicle cluster hierarchical cooperative navigation system, which comprises the following steps: obtaining state data of each wing aircraft in the system according to a self sensor to determine the azimuth angle and the pitch angle of each wing aircraft relative to each long aircraft, and constructing an observability matrix; performing singular value decomposition on the observability matrix, and determining an observability index according to the maximum singular value and the minimum singular value after the decomposition, which is used for quantifying the observability degree of the long aircraft type of a long aircraft combination used in the unmanned aerial vehicle cluster hierarchical cooperative navigation system to the wing aircraft navigation error contribution; and based on a binary particle swarm optimization algorithm, taking the observability index of the unmanned aerial vehicle cluster as an adaptive function, iteratively searching for a target long aircraft combination, wherein in the iteration, the inertia weight of the algorithm is continuously adjusted according to the relationship between the particle fitness value and the group fitness value, and the individual learning factor and the group learning factor are continuously adjusted according to the iteration number.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) navigation technology, and in particular to a method, system, device, and medium for selecting the lead aircraft in a hierarchical collaborative navigation system for UAV swarms. Background Technology

[0002] With the development of drone technology, large-scale drone swarms are playing an increasingly important role in missions such as reconnaissance, environmental monitoring, and disaster relief. In complex environments where satellite navigation signals are susceptible to interference or rejection, cooperative navigation based on a hierarchical architecture has become a mainstream solution. This architecture typically consists of a lead drone equipped with high-precision navigation equipment and wingmen equipped with low-cost sensors. The wingmen correct their own navigation errors by measuring the relative distance, angle, or speed information between themselves and the lead drone, and using filtering algorithms.

[0003] In practical applications, to ensure system robustness and coverage, the number of lead aircraft is usually greater than the minimum number required for wingman navigation calculations, meaning there is redundancy in the number of lead aircraft. On the one hand, if the wingman simultaneously processes the measurement information of all visible lead aircraft, it will significantly increase the data processing and filtering calculation burden on the airborne computer, severely affecting the real-time performance of the navigation system. On the other hand, different combinations of lead aircraft form different spatial geometric configurations, which have a significant impact on the observability of wingman navigation errors. Therefore, how to quickly select the optimal combination that contributes the most to navigation accuracy from numerous candidate lead aircraft is a key issue in improving the collaborative navigation performance of UAV swarms.

[0004] Existing wingman selection strategies, such as those based on the Minimum Geometric Dilution Precision (MinGDOP) criterion, can theoretically select the node combination with the optimal geometric configuration, thereby achieving the highest navigation accuracy. However, this method typically requires a "combinatorial traversal" search, and when the number of candidate wingmen is large, the search space grows exponentially, resulting in extremely high computational complexity, making it difficult to meet the real-time requirements of large-scale, highly dynamic flight in swarms. While selection methods based on geometric distribution rules, such as the maximum angle separation method and the maximum baseline length method, perform rapid selection through predefined geometric rules, their computational efficiency is superior to those based on the MinGDOP criterion. However, their selection results are often based on meeting specific threshold conditions, and the resulting configuration may not be optimal. Furthermore, during the relative movement of the wingman and wingmen, frequent changes in the selected nodes can easily occur, leading to a decrease in system stability.

[0005] Therefore, existing long-machine screening strategies struggle to balance screening effectiveness and computational efficiency. Summary of the Invention

[0006] The purpose of this invention is to provide a method, system, device and medium for selecting the lead aircraft in a hierarchical cooperative navigation system for unmanned aerial vehicle (UAV) swarms, which can solve the problem of balancing the selection effect and computational efficiency.

[0007] To address the aforementioned technical problems, embodiments of the present invention provide a method for selecting the lead aircraft in a hierarchical cooperative navigation system for unmanned aerial vehicle (UAV) swarms, comprising the following steps: Acquire the status data obtained by each wingman in a hierarchical collaborative navigation system for drone swarms based on its own sensors; Based on the status data of each wingman, determine the azimuth and pitch angles of each wingman relative to each lead aircraft to construct an observability matrix; Singular value decomposition is performed on the observability matrix, and the observability index is determined based on the maximum and minimum singular values ​​among the multiple singular values ​​obtained from the decomposition. The observability index is used to quantify the observability of the contribution of the lead aircraft configuration of the lead aircraft combination to the navigation error of the wingmen in the hierarchical cooperative navigation system of UAV swarm. Based on the binary particle swarm optimization algorithm, the fitness function is used to iteratively search for the target leader combination of UAVs. In the iteration, the inertia weight of the binary particle swarm optimization algorithm is continuously adjusted according to the relationship between the fitness value of the particles and the fitness value of the swarm, and the individual learning factor and the swarm learning factor are continuously adjusted according to the number of iterations.

[0008] Furthermore, the observability index is: ; In the formula, The observability matrix The minimum singular value, For observability matrix The maximum singular value.

[0009] Furthermore, the individual learning factor and the swarm learning factor of the binary particle swarm optimization algorithm are adjusted using the following formula: ; In the formula, and The first Individual learning factor and group learning factor at each iteration and They are respectively and The minimum value, and They are respectively and The maximum value, ,for and normalization progress, This represents the total number of iterations.

[0010] Furthermore, in the binary particle swarm optimization algorithm, the velocity in each dimension of the velocity vector of each particle is updated using the following formula: ; In the formula, For inertial weights, and A random number in the range [0, 1]. For particles The optimal position of an individual The current global optimal solution for the population. For the first m The binary position vector of each particle.

[0011] Furthermore, obtaining the target long machine combination through the iterative search also includes: Based on the target lead aircraft combination, determine the measurement data of each wingman in the hierarchical cooperative navigation system of the UAV swarm; By combining the measurement and status data of each wingman, hierarchical collaborative navigation of the drone swarm is achieved.

[0012] Furthermore, the acquisition of status data obtained by each wingman in the hierarchical cooperative navigation system of the UAV swarm based on its own sensors includes: In each navigation cycle of the hierarchical collaborative navigation system for drone swarms, acquire the status data obtained by each wingman based on its own sensors; After obtaining the target long-machine combination through iterative search, the process further includes: Based on the status data of each wingman in the hierarchical cooperative navigation system of the UAV swarm in each navigation cycle, the leader selection result of the hierarchical cooperative navigation system of the UAV swarm in each navigation cycle is obtained. Based on the leader selection results of the UAV swarm hierarchical cooperative navigation system in each navigation cycle, hierarchical cooperative navigation of the UAV swarm is carried out in each navigation cycle.

[0013] Embodiments of the present invention also provide a leader selection system for a hierarchical cooperative navigation system for unmanned aerial vehicle (UAV) swarms, comprising: The data acquisition module is used to acquire the status data obtained by each wingman in the hierarchical collaborative navigation system of the UAV swarm based on its own sensors; The matrix construction module is used to determine the azimuth and pitch angles of each wingman relative to each lead aircraft based on the status data of each wingman, in order to construct an observability matrix; The index determination module is used to perform singular value decomposition on the observability matrix and determine the observability index based on the maximum and minimum singular values ​​among the multiple singular values ​​obtained from the decomposition. The observability index is used to quantify the observability of the contribution of the lead aircraft configuration of the lead aircraft combination to the navigation error of the wingmen in the hierarchical cooperative navigation system of UAV swarm. The leader selection module is used to iteratively search for target leader combinations based on the binary particle swarm optimization algorithm, using the observability index of the UAV swarm as the fitness function. In the iteration, the inertia weight of the binary particle swarm optimization algorithm is continuously adjusted according to the relationship between the fitness value of the particles and the fitness value of the swarm, and the individual learning factor and the swarm learning factor are continuously adjusted according to the number of iterations.

[0014] Embodiments of the present invention also provide a computer device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the lead aircraft selection method of the above-described hierarchical cooperative navigation system for unmanned aerial vehicle swarms.

[0015] Embodiments of the present invention also provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the lead aircraft selection method of the above-described hierarchical cooperative navigation system for unmanned aerial vehicle swarms.

[0016] The leader selection method for a hierarchical cooperative navigation system for unmanned aerial vehicle (UAV) swarms provided by this invention has at least the following beneficial effects: By performing singular value decomposition on the observability matrix constructed from the azimuth and pitch angles of each wingman relative to each lead aircraft, an observability index is defined. This index quantifies the observability of the contribution of the lead aircraft configuration of the hierarchical cooperative navigation system to the navigation error of the wingmen. The higher the observability, the stronger the ability to correct the navigation error of the wingmen, and the higher the final navigation accuracy. Therefore, using this index as the fitness function of the binary particle swarm optimization algorithm can give the algorithm a clear mathematical direction during the search, thereby accurately locking the lead aircraft combination that contributes the most to the navigation accuracy, i.e., the optimal lead aircraft selection result. At the same time, through the iterative update mechanism of this algorithm, it can quickly approach the global optimum, resulting in higher computational efficiency.

[0017] In addition, this invention improves the traditional binary particle swarm optimization algorithm. Its inertia weight is continuously adjusted according to the relationship between the fitness value of the particles and the fitness value of the swarm. The individual learning factor and the swarm learning factor are continuously adjusted according to the number of iterations. This adaptive inertia weight and adaptive learning factor suppress the jump of the final selection result by improving the search and convergence ability of the algorithm, so that the selected long structure configuration is highly consistent with the theoretical optimal configuration, thus ensuring the stability of the navigation system. Attached Figure Description

[0018] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings:

[0019] Figure 1 A flowchart illustrating the lead aircraft selection method for a hierarchical collaborative navigation system for unmanned aerial vehicle (UAV) swarms provided by this invention. Figure 2 This invention provides a schematic diagram of an optimal long-arm configuration for three lead aircraft. Figure 3 This invention provides a schematic diagram of an optimal long-arm configuration for four lead aircraft. Figure 4 A flowchart of the BPSO algorithm provided for this invention; Figure 5 A schematic diagram of an algorithm update provided by the present invention; Figure 6 A long-machine screening strategy framework diagram provided by the present invention; Figure 7 A schematic diagram of a drone formation flight trajectory provided by the present invention; Figure 8 This invention provides a schematic diagram of the main flight process of an unmanned aerial vehicle (UAV) swarm. Figure 9 This invention provides a schematic diagram illustrating the trend of observability variation with configuration for four algorithms. Figure 10 A schematic diagram of the long-machine screening results of the MinGDOP method provided by the present invention; Figure 11 A schematic diagram of a geometric layout filtering result provided by the present invention; Figure 12 A schematic diagram of the screening results based on the observability index provided by the present invention; Figure 13 A schematic diagram comparing the screening time of several methods provided by this invention; Figure 14 This invention provides a schematic diagram comparing the root mean square error of wingman positions under four screening strategies. Figure 15 This is a schematic diagram comparing the positional ARMSE under four screening strategies provided by the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0021] The technical solutions provided by the various embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0022] One embodiment of the present invention relates to a method for selecting the lead aircraft in a hierarchical cooperative navigation system for unmanned aerial vehicle (UAV) swarms. The specific process of the method for selecting the lead aircraft in this embodiment can be as follows: Figure 1 As shown, it includes: Step 101: Obtain the status data of each wingman in the hierarchical cooperative navigation system of the UAV swarm based on its own sensors; Step 102: Based on the status data of each wingman, determine the azimuth and pitch angles of each wingman relative to each lead aircraft to construct an observability matrix; Step 103: Perform singular value decomposition on the observability matrix, and determine the observability index based on the maximum and minimum singular values ​​among the multiple singular values ​​obtained from the decomposition; wherein, the observability index is used to quantify the observability of the contribution of the lead aircraft configuration of the lead aircraft combination used in the hierarchical cooperative navigation system of UAV swarm to the navigation error of the wingman. Step 104: Based on the binary particle swarm optimization algorithm, the target leader combination is obtained by iterative search using the observability index of the UAV swarm as the fitness function; wherein, in the iteration, the inertia weight of the binary particle swarm optimization algorithm is continuously adjusted according to the relationship between the fitness value of the particles and the fitness value of the swarm, and the individual learning factor and the swarm learning factor are continuously adjusted according to the number of iterations.

[0023] The following is a detailed description of the implementation details of the leader selection method of the hierarchical cooperative navigation system for UAV swarms in this embodiment. The following content is only for the convenience of understanding and is not necessary for implementing this solution.

[0024] In steps 101 to 103, the observability index of the hierarchical cooperative navigation system of the UAV swarm is constructed based on the position error equations calculated by the inertial navigation systems of the lead aircraft and the wingman.

[0025] First, we analyze the impact of long-frame configurations and their dynamic changes on cooperative navigation accuracy: To establish a cluster-based collaborative navigation model, the following coordinate system is used: inertial coordinate system ( (system), carrier coordinate system ( Earth coordinate system (system) and navigation coordinate system ( Tie), The system is usually chosen as the East-North-Sky geographic coordinate system ( Furthermore, due to rotational errors, the calculated navigation coordinate system differs from the actual coordinate system. There are differences in the system, which are defined as follows: The system calculates the high-precision layer length based on the position output of the inertial navigation system. With low-precision layer wingmen The distance between them is:

[0026] ; in, and The first High-precision layer long machine and the first The position of the low-precision wingman in the Earth coordinate system, and: ; in, and These are the inertial navigation position errors for the lead aircraft and the wingman, respectively.

[0027] Apply the above formula to the actual position and Taylor expansion, ignoring higher-order terms, yields: ; in, For high-precision layer long machine With low-precision layer wingmen The distance between them Represents the actual distance value. , , , and These are the azimuth and elevation angles, respectively.

[0028] In addition, the distance between the lead aircraft and the wingman, obtained from relative ranging, is as follows: ; in, This refers to the noise of the ranging sensor.

[0029] By further subtracting from the previous equation, we can obtain the pure distance error: ; When the number of lead aircraft is three or more, if the wingmen With the lead aircraft Components of the relative direction vector between , and If linearly independent, then the wingman Inertial navigation error It can form three or more linearly independent equations. Observable; if wingman With the lead aircraft Components of the relative direction vector between , and Linear correlation, at this time the wingman Inertial navigation error It is impossible to construct three or more linearly independent equations, therefore Unable to be estimated and corrected, resulting in wingmen The navigation accuracy has decreased.

[0030] Then, an observability index for configuration optimization is established: When the number of high-precision layer master machines in the cluster is When the aircraft is in flight, the relationship between the inertial navigation position error and the range error of the low-precision wingman can be rewritten as: ; in, This indicates the low-precision wingman currently being calibrated. The first one that provides reference information The comprehensive distance error index between high-precision long machines.

[0031] Furthermore, the observability matrix can be represented as ; In the formula, and The first The corresponding wingman The azimuth and pitch angles of the lead aircraft.

[0032] Define observability index This indicator quantifies the observability of the current long-configuration aircraft's navigation error. The closer the ratio is to 1, the better the configuration and the higher the navigation accuracy. Its calculation formula is: ; In the formula, For observability matrix The minimum singular value, For observability matrix The maximum singular value.

[0033] The above formula shows that the observability index of the system is related to the relative azimuth and pitch angles between the lead and wingmen, and can be used to... As an evaluation model for cooperative navigation accuracy, by solving for the relative azimuth and pitch angles that maximize the model, the long-form configuration at which cooperative navigation accuracy is optimal can be obtained. Figure 2 and Figure 3 The optimal lead configurations for three and four lead aircraft are given respectively.

[0034] In step 104, a long-machine screening optimization framework based on binary particle swarm optimization (BPSO) is established. This framework includes a defined observability index. As the fitness function of the particle swarm optimization algorithm, the goal of the algorithm is to find the particle position that maximizes the fitness function value while satisfying the constraint of the number of long machines.

[0035] To improve search efficiency and avoid getting trapped in local optima, the standard particle swarm optimization algorithm is improved as follows: An adaptive inertia weight is introduced, which is dynamically adjusted based on the relationship between the particle's current fitness and the swarm's average fitness. When a particle's fitness is lower than the average, the weight is increased to enhance global exploration capabilities; when a particle's fitness is higher than the average, the weight is decreased to focus on local search capabilities.

[0036] Introducing an adaptive learning factor and , respectively represented as: ; In the formula, and They are respectively and The minimum value, and They are respectively and The maximum value; in, and The changes can be represented by the normalized progress as follows: ; in, This represents the total number of iterations.

[0037] Individual learning factors The learning factor decreases linearly with the number of iterations, initially focusing on exploration based on individual experience. The rate increases linearly with the number of iterations, and in later stages it tends to approach the global optimum. This is achieved by normalizing the progress parameter. By controlling the changes in both, the population can be guided to quickly converge from a dispersed search to the global optimum.

[0038] The improved particle swarm optimization algorithm is used for iterative search. The search stops when the maximum number of iterations or the fitness threshold is reached, and the combination of the long machine numbers corresponding to the globally optimal particle is output.

[0039] For a detailed implementation of the Binary Particle Swarm Optimization (BPSO) algorithm, please refer to [link to relevant documentation]. Figure 4 ,as follows: (1) Initialize the particle swarm size Maximum number of iterations Inertial weight Learning factor , .

[0040] Randomly initialize a binary position vector for each particle: ; in, For the first m The binary position vector of each particle.

[0041] Initialize the velocity vector: ; in, For the first m The initial velocity vector of each particle.

[0042] Calculate the fitness of all particles based on the objective function, and initialize the optimal value for each particle. and global optimal .

[0043] (2) For each particle, such as Figure 5 As shown, the velocity in each dimension of the velocity vector is updated in the following way: ; In the formula, Particle velocity inertia weight, and These are the weights of individual and group experiences, respectively. and A random number in the range [0, 1]. For particles The optimal position of an individual The current global optimal solution for the group.

[0044] (3) The position is updated by mapping the velocity to a probability using the Sigmoid function: ; In the formula, The Sigmoid function maps the velocity to [0, 1].

[0045] Then, the position is updated based on probability: ; In the formula, It is a random number, in the range [0, 1].

[0046] (4) Update the individual optimum and the global optimum. If the current fitness value is better than the individual optimum, then update... If a particle's fitness value is better than the global optimum, then update. .

[0047] (5) Reach the maximum number of iterations Stop when the fitness value meets the requirements.

[0048] To address the problem that particle swarm optimization (PSO) algorithms are prone to getting stuck in local optima during the optimization process in practical applications, this embodiment designs an improved PSO optimization algorithm based on adaptively adjusted inertia weights and learning factors. This enhances the algorithm's global search capability and convergence performance, thereby more effectively meeting the requirements for dynamic lead selection in cooperative navigation.

[0049] Inertia weight This represents the ability of a particle to maintain its previous state of motion. Introducing adaptive inertia weights can achieve an effective balance between global exploration and local search. In the initial stage, when the population distribution is relatively discrete, , At this point, the global search capability improves, allowing for the exploration of more long-machine combinations and preventing getting trapped in local optima; in the later stages, when the population distribution converges, , At this point, the local search capability is improved, accelerating the convergence speed. and These are the minimum and maximum values ​​of the inertia weight, respectively. For the first The average fitness of all particles at the next iteration For particles No. The fitness value at the next iteration For the first The minimum fitness of all particles at the next iteration.

[0050] Individual learning weight Group learning weights It is a key parameter that determines the performance of the algorithm. Characterizes the particle's independent exploration ability, that is, the degree to which it regresses to its individual extreme value; This characterizes the global search capability, i.e., the tendency to converge towards the group's extreme value. To balance global and local searches, a dynamic adjustment strategy is adopted: initially, to expand the search range and maintain particle diversity, a "larger" approach is used. smaller The configuration is set to prevent premature convergence; in the mid-term, to filter and gradually approach the optimal solution, the value is gradually reduced. and increase To achieve a smooth transition; in the later stages, to ensure the stability and accuracy of convergence, a "smaller" approach is adopted. Larger The parameter settings, which evolve with the iteration process, can effectively reduce the oscillation of particles near the optimal solution and guide the population to converge quickly towards the global optimum.

[0051] In some embodiments, after obtaining the target lead aircraft combination through iterative search using the above embodiments, the measurement data of each wingman in the hierarchical cooperative navigation system of the UAV swarm is determined based on the target lead aircraft combination; the hierarchical cooperative navigation of the UAV swarm is then performed by combining the measurement data and status data of each wingman. That is, the wingman only receives navigation information sent by the selected optimal lead aircraft combination, combines it with its own inertial navigation data, and uses a Kalman filter to perform cooperative navigation calculations to correct its own navigation errors.

[0052] In some embodiments, within each navigation cycle of the hierarchical cooperative navigation system for UAV swarms, this embodiment acquires the status data obtained by each wingman based on its own sensors. After iteratively searching to obtain the target leader combination, the leader selection result of the hierarchical cooperative navigation system for each navigation cycle is obtained based on the status data of each wingman in each navigation cycle. Finally, based on the leader selection result of the hierarchical cooperative navigation system for each navigation cycle, hierarchical cooperative navigation of the UAV swarm is performed in each navigation cycle.

[0053] This invention designs a leader selection strategy, embeds the observability index into a particle swarm optimization framework, and then achieves rapid acquisition of the optimal leader configuration through dynamic search. Specifically, the leader selection strategy framework consists of two parts: a wingman navigation solution process and a leader configuration selection process.

[0054] The long structure selection process includes initialization (initial location, initial population, and initial algorithm parameters), observability index calculation, and search strategy update, as detailed below: (1) Initialization. Randomly generate several high-precision layer-long machines. Forming a drone swarm and assigning it numbers; from Selecting from long-range aircraft shelf The candidate lead aircraft is formed Various combination methods are used, and population size is selected from them. Each combination forms a particle, constituting the initial population. One particle is ,in The drone's serial number and initial speed are... .

[0055] (2) Calculate the observability of the UAV combination corresponding to each particle, i.e., the fitness function. and average fitness value of particles The particle corresponding to the maximum observability is set as the global optimum. Each initial particle is set to the initial individual optimal value. .

[0056] (3) Update according to the steps of the aforementioned binary particle swarm algorithm, and terminate the iteration after reaching the maximum number of iterations.

[0057] The wingman navigation solution process includes steps such as navigation parameter initialization, time update, measurement update, and navigation parameter output. Among them, the time update is performed according to the standard Unscented Kalman Filter (UKF) algorithm.

[0058] The specific state model is established as follows: Based on the error equation of the Miniature Inertial Measurement Unit (MIMU), the state equation of the cooperative navigation system is established.

[0059] To establish a cluster-based collaborative navigation model, the following coordinate system is used: inertial coordinate system ( (system), carrier coordinate system ( Earth coordinate system (system) and navigation coordinate system ( Tie), The system is usually chosen as the East-North-Sky geographic coordinate system ( Furthermore, due to rotational errors, the calculated navigation coordinate system differs from the actual coordinate system. There are differences in the system, which are defined as follows: System. The wingman MIMU error equation is selected as the state equation for cluster cooperative navigation, where the state vector is:

[0060] ; In the formula, , and These represent the attitude, velocity, and position errors of the MIMU, respectively. and These represent the constant drift of the gyroscope and accelerometer, respectively.

[0061] Therefore, the MIMU attitude error equation can be written as: ; In the formula, and From System and Tie The transformation matrix of the system; This represents the transformation matrix from angular velocity to Euler angles; yes System relative to The angular velocity of the system, That is the corresponding error.

[0062] The further differential equations for the MIMU velocity and position errors can be written as: ; In the formula, It is the specific force measured by the accelerometer. It is the Earth's angular velocity. It is the angular velocity relative to the Earth. and They are respectively and The error, A and B Both are coefficient matrices. and These represent the velocity error and position error of the MIMU, respectively.

[0063] MIMU error also includes the output errors of the gyroscope and accelerometer, which can be described as: ; In the formula, and It corresponds to white noise.

[0064] The state model of the wingman's cooperative navigation system can be obtained and further discretized as follows: ; In the formula, yes The state vector at any given time; It is a discrete nonlinear function describing the state model. yes Zero-mean Gaussian white noise at time 10.

[0065] For measurement updates, it is assumed that the number of long machines obtained from the long machine type screening process is: The resulting measurement model is: ; In the formula, Indicates wingman exist The total observation vector after time-time concatenation Indicates wingman With the Specific observation information between the lead aircraft Represents the basic observation matrix, This indicates the measurement error relative to the sensor itself. Indicates the first Navigation status error of the lead aircraft. and The calculation formulas are as follows: ; ; In the formula, This represents the direction cosine matrix from the Earth coordinate system to the navigation coordinate system. This represents the Jacobian matrix related to the position error. This represents the direction cosine matrix from the body coordinate system to the navigation coordinate system. The projection matrix representing the velocity measurement error in the body coordinate system. It is The first column is obtained by swapping the second and third columns in sequence.

[0066] After obtaining the measurement model, the measurement update can be performed according to the standard UKF algorithm. After obtaining the state estimation result, the navigation parameters at the current time are output.

[0067] Therefore, the leader selection method of the hierarchical cooperative navigation system for UAV swarms in this embodiment can be achieved through... Figure 6 The process shown includes the following steps: (1) Set the initial state estimate of the filter and its error covariance matrix .

[0068] (2) Time update, calculate state one-step prediction and its error covariance matrix .

[0069] (3) Initialize the population and initialize the positions of the particles. and speed .

[0070] (4) Calculate the observability (fitness value) of the UAV combination corresponding to each particle. .

[0071] (5) Update the search strategy and update the inertia weight. , and .

[0072] (6) Calculate the fitness value corresponding to the new position. And update the individual optimal position of the particle. and the global optimal position .

[0073] (7) Repeat steps 104 to 106. When the maximum number of iterations is reached, output the global optimal position of the particle, i.e. the optimal long structure.

[0074] (8) Based on the obtained long structure configuration, perform measurement updates to obtain the measurement model and state estimation. and its error covariance matrix .

[0075] (9) Repeat (2) to (8) to perform the next moment filtering solution until the entire navigation process is completed.

[0076] The following is a simulation verification embodiment of the long-machine screening method of the present invention: like Figure 7 As shown, to ensure the dynamic change of the spatial topology of the UAV swarm, the simulation experiment set each UAV to execute a series of representative maneuver commands, covering climb, descent, turning, and speed changes. The initial state settings of the wingman are as follows: geographical coordinates are 108.91°E, 34.245°N, flight altitude is 200 m; initial attitude (roll, pitch, yaw angle) and initial three-axis velocity are all initialized to zero. To better simulate the cooperative flight characteristics under complex mission scenarios, the lead UAV adopts diverse maneuver strategies to construct a spatial configuration that evolves over time. The entire simulation experiment lasts approximately 1800 s. For a detailed description of the entire swarm flight process and specific flight state parameter configurations, please refer to [references to be inserted here]. Figure 8 Compared with Table 1:

[0077] Table 1 1 Accelerate level flight 10 0 0 0 0, 0.8, 0 2 Uniform level flight 100 0 0 0 0, 0, 0 3 Pull up 20 1 0 0 0, 0, 1.571 4 level 20 -1 0 0 0, 0, -1.571 5 Uniform level flight 500 0 0 0 0, 0, 0 6 Turn left 90° 0.8 0 -5.0 1.0 -2.618, 0, 0 7 Turning phase 45 0 0 1.0 -2.618, 0, 0 8 level 1 0 10 1.0 -2.618, 0, 0 9 Accelerate level flight 3 0 0 -0 -0.8, 0, 0 10 Uniform level flight 50 0 0 0 0, 0, 0 11 Turn left 90° 0.8 0 -5.0 1.0 -2.618, 0, 0 12 Turning phase 45 0 0 1.0 -2.618, 0, 0 13 level 1 0 10 1.0 -2.618, 0, 0 14 Uniform level flight 150 0 0 0 0, 0, 0 15 Dive 20 -1 0 -1.0 0, 0, -1.571 16 level 20 1 0 1.0 0, 0, 1.571 17 Uniform level flight 400 0 0 0 0, 0, 0 18 Deceleration and level flight 10 0 0 0 0, -0.8, 0 19 Uniform level flight 1000 0 0 0 0, 0, 0 The lead aircraft employs a high-precision INS / differential GNSS integrated navigation system, while the wingman uses a low-precision MIMU in conjunction with a relative navigation sensor for cooperative navigation. The sensor parameters required for the wingman's integrated navigation are shown in Table 2.

[0078] Table 2 #imgpt267# Figure 9 The observability variations over time for four lead selection methods are presented. Randomly selected observability fluctuates drastically and has a low mean, failing to meet the observability requirements of cooperative navigation. The selection method based on the minimum geometric precision factor criterion achieves optimal and stable performance through exhaustive search, and can be considered the theoretically optimal solution; however, its computational time is too long, making it unsuitable for large-scale or real-time applications. While the geometric layout method is simple to implement, it exhibits limitations under complex time-varying configurations. In contrast, the particle swarm optimization-based selection method proposed in this invention can quickly search for highly observable configurations, achieving performance almost identical to the minimum geometric precision factor criterion method while significantly reducing computational cost.

[0079] To further explore the similarities and differences among different screening strategies, long-run screening data within a time window of 1200s to 1500s were selected for comparative analysis. Figure 10 The filtering output of MinGDOP is shown, and the specific results for each time period are detailed in Table 3. Given the traversal characteristics of MinGDOP, these results can be considered as the theoretically optimal configuration. Figure 11 This reveals the performance characteristics of geometric layout selection. Because this method follows the maximum angle separation principle, requiring only that the GDOP index reach a specific threshold, it cannot guarantee that the selected node set will reach the optimal selection state. This directly leads to a large number of jump points in the results, exhibiting frequent fluctuations, redundant switching, and discontinuities on the time axis. This instability of the configuration may pose a threat to cooperative navigation, for example, by causing frequent changes in communication links or measurement sources, thereby weakening the robustness of the system. Figure 12 The screening effect of the proposed method is demonstrated. Thanks to the efficient global search capability of the particle swarm optimization algorithm in binary space, this method significantly suppresses result jumps, and the screened configurations are very close to the results of the MinGDOP method.

[0080] Table 3 1200-1220 L2, L5, L7 1390-1450 L2, L5, L7, L10 1220-1300 L4, L5, L8 1450-1480 L1, L3, L5 1300-1350 L2, L4, L5, L9 1480-1500 L1, L5, L9, L10 1350-1390 L3, L6, L7, L7 Figure 13 The data visually presents the computational performance of the MinGDOP method, the geometric layout screening method, and the method proposed in this invention. The data shows that the MinGDOP method, due to its traversal approach, has the heaviest computational load, with a single screening cycle taking up to 57.659 ms. The geometric layout screening method simplifies the process based on the maximum separation angle rule, reducing its time to 27.01 ms. In contrast, the method proposed in this invention, leveraging the intelligent iterative optimization mechanism of the particle swarm optimization algorithm, further compresses the time to 15.51 ms. Quantitative comparisons demonstrate that this method's computational efficiency is significantly better than the former two, reducing its time by 73.1% compared to the MinGDOP method and by 42.58% compared to the geometric layout screening method.

[0081] Figure 14 and Figure 15 Histograms of the root mean square error (RMSE) and average root mean square error (ARMSE) distributions of wingmen under four selection strategies are presented. The data shows that the errors in random selection along the latitude (0.356m), longitude (0.172m), and altitude (0.215m) directions are significantly higher than other methods, at 0.356m, 0.172m, and 0.215m respectively. In contrast, the MinGDOP method, geometric layout selection, and the method proposed in this invention control the errors at lower levels, with longitude errors of 0.114m, 0.148m, and 0.138m respectively. Among these, the method proposed in this invention, leveraging the optimization capabilities of the particle swarm optimization algorithm, achieves accuracy very close to that of the MinGDOP method and significantly outperforms geometric layout selection.

[0082] The simulation results show that the proposed method significantly optimizes computational efficiency while maintaining navigation accuracy close to the theoretical optimum (MinGDOP method). Its computation time is only 26.9% of that of the MinGDOP method and lower than that of geometric layout selection (57.42%). This indicates that by introducing a particle swarm optimization algorithm to avoid complex full combinatorial traversal, the computational burden can be significantly reduced without sacrificing performance, making it very suitable for large-scale clusters and applications with high real-time requirements.

[0083] The steps of the various methods described above are only for clarity. In practice, they can be combined into one step or some steps can be split into multiple steps. As long as they include the same logical relationship, they are all within the protection scope of this invention. Adding insignificant modifications or introducing insignificant designs to the algorithm or process, without changing the core design of the algorithm and process, are also within the protection scope of this invention.

[0084] Another embodiment of the present invention relates to a lead aircraft selection system for a hierarchical cooperative navigation system for unmanned aerial vehicle (UAV) swarms. The implementation details of the lead aircraft selection system in this embodiment are described below. The following details are provided for ease of understanding and are not essential for implementing this solution. The lead aircraft selection system in this embodiment includes: The data acquisition module is used to acquire the status data obtained by each wingman in the hierarchical collaborative navigation system of the UAV swarm based on its own sensors; The matrix construction module is used to determine the azimuth and pitch angles of each wingman relative to each lead aircraft based on the status data of each wingman, in order to construct an observability matrix; The index determination module is used to perform singular value decomposition on the observability matrix and determine the observability index based on the maximum and minimum singular values ​​among the multiple singular values ​​obtained from the decomposition. The observability index is used to quantify the observability of the contribution of the lead aircraft configuration of the lead aircraft combination to the navigation error of the wingmen in the hierarchical cooperative navigation system of UAV swarm. The leader selection module is used to iteratively search for target leader combinations based on the binary particle swarm optimization algorithm, using the observability index of the UAV swarm as the fitness function. In the iteration, the inertia weight of the binary particle swarm optimization algorithm is continuously adjusted according to the relationship between the fitness value of the particles and the fitness value of the swarm, and the individual learning factor and the swarm learning factor are continuously adjusted according to the number of iterations.

[0085] It is not difficult to see that this embodiment is a system embodiment corresponding to the above method embodiments, and this embodiment can be implemented in conjunction with the above method embodiments. The relevant technical details and technical effects mentioned in the above embodiments are still valid in this embodiment, and will not be repeated here to reduce repetition. Accordingly, the relevant technical details mentioned in this embodiment can also be applied to the above embodiments.

[0086] It is worth mentioning that all modules involved in this embodiment are logical modules. In practical applications, a logical unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. Furthermore, to highlight the innovative aspects of this invention, this embodiment does not introduce units that are not closely related to solving the technical problem proposed by this invention; however, this does not mean that other units are absent from this embodiment.

[0087] Another embodiment of the present invention relates to a computer device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the leader selection method of the hierarchical cooperative navigation system for UAV swarms in the above embodiments.

[0088] The memory and processor are connected via a bus, which can include any number of interconnecting buses and bridges, connecting various circuits of one or more processors and memories. The bus can also connect various other circuits, such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and will not be described further herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be a single element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by the processor is transmitted over the wireless medium via an antenna, which further receives data and transmits it to the processor.

[0089] The processor manages the bus and general processing, and also provides various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory is used to store data used by the processor during operation.

[0090] Another embodiment of the present invention relates to a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the method embodiments described above.

[0091] That is, those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0092] Those skilled in the art will understand that the above embodiments are specific embodiments for implementing the present invention, and in practical applications, various changes can be made to them in form and detail without departing from the spirit and scope of the present invention.

Claims

1. A method for selecting the lead drone in a hierarchical cooperative navigation system for unmanned aerial vehicle (UAV) swarms, characterized in that, The method includes: Acquire the status data obtained by each wingman in a hierarchical collaborative navigation system for drone swarms based on its own sensors; Based on the status data of each wingman, determine the azimuth and pitch angles of each wingman relative to each lead aircraft to construct an observability matrix; Singular value decomposition is performed on the observability matrix, and the observability index is determined based on the maximum and minimum singular values ​​among the multiple singular values ​​obtained from the decomposition. The observability index is used to quantify the observability of the contribution of the lead aircraft configuration of the lead aircraft combination to the navigation error of the wingmen in the hierarchical cooperative navigation system of UAV swarm. Based on the binary particle swarm optimization algorithm, the fitness function is used as the observability index of the UAV swarm, and the target leader combination is obtained through iterative search. In the iteration, the inertia weight of the binary particle swarm optimization algorithm is continuously adjusted according to the relationship between the fitness value of the particles and the fitness value of the swarm, and the individual learning factor and the swarm learning factor are continuously adjusted according to the number of iterations. The process involves determining the azimuth and pitch angles of each wingman relative to each lead aircraft based on the status data of each wingman, in order to construct an observability matrix, including: Based on the positions of the wingman and the lead aircraft in the Earth coordinate system and the distance between them calculated through the position output by the inertial navigation system, the inertial navigation position error equations for the lead aircraft and the wingman are constructed. After Taylor expansion of the inertial navigation position error equation at the actual position and ignoring higher-order terms, the difference between the equation and the distance equation between the lead aircraft and the wingman obtained from relative ranging is calculated to obtain the pure distance error equation. By combining the pure distance error equations between each lead aircraft and each wingman, as well as the azimuth and pitch angles of each wingman relative to each lead aircraft, an observability matrix is ​​constructed. The observability index is: ； In the formula, The observability matrix The minimum singular value, The observability matrix The maximum singular value.

2. The method for selecting the lead drone in a hierarchical cooperative navigation system for unmanned aerial vehicle (UAV) swarms according to claim 1, characterized in that, The individual learning factor and swarm learning factor of the binary particle swarm optimization algorithm are adjusted using the following formula: ； In the formula, and The first Individual learning factor and group learning factor at each iteration and They are respectively and The minimum value, and They are respectively and The maximum value, ,for and normalization progress, This represents the total number of iterations.

3. The method for selecting the lead drone in a hierarchical cooperative navigation system for unmanned aerial vehicle (UAV) swarms according to claim 2, characterized in that, In the binary particle swarm optimization algorithm, the velocity in each dimension of the velocity vector of each particle is updated using the following formula: ； In the formula, For inertial weights, and A random number in the range [0, 1]. For particles The optimal position of an individual The current global optimal solution for the population. For the first In the next iteration, the particle For the m The binary position vector of the steerable aircraft. For the first t In the next iteration, the particle For the m The velocity vector of the long-range aircraft.

4. The method for selecting the lead drone in a hierarchical cooperative navigation system for unmanned aerial vehicle (UAV) swarms according to claim 1, characterized in that, After obtaining the target long-machine combination through iterative search, the method further includes: Based on the target lead aircraft combination, determine the measurement data of each wingman in the hierarchical cooperative navigation system of the UAV swarm; By combining the measurement and status data of each wingman, hierarchical collaborative navigation of the drone swarm is achieved.

5. The method for selecting the lead drone in a hierarchical cooperative navigation system for unmanned aerial vehicle (UAV) swarms according to claim 4, characterized in that, The acquisition of status data obtained by each wingman in the hierarchical cooperative navigation system of the UAV swarm based on its own sensors includes: In each navigation cycle of the hierarchical collaborative navigation system for drone swarms, acquire the status data obtained by each wingman based on its own sensors; After obtaining the target long-machine combination through iterative search, the method further includes: Based on the status data of each wingman in the hierarchical cooperative navigation system of the UAV swarm in each navigation cycle, the leader selection result of the hierarchical cooperative navigation system of the UAV swarm in each navigation cycle is obtained. Based on the leader selection results of the UAV swarm hierarchical cooperative navigation system in each navigation cycle, hierarchical cooperative navigation of the UAV swarm is carried out in each navigation cycle.

6. A leader selection system for a hierarchical cooperative navigation system for unmanned aerial vehicle (UAV) swarms, characterized in that, The system includes: The data acquisition module is used to acquire the status data obtained by each wingman in the hierarchical collaborative navigation system of the UAV swarm based on its own sensors; The matrix construction module is used to determine the azimuth and pitch angles of each wingman relative to each lead aircraft based on the status data of each wingman, in order to construct an observability matrix; The index determination module is used to perform singular value decomposition on the observability matrix and determine the observability index based on the maximum and minimum singular values ​​among the multiple singular values ​​obtained from the decomposition. The observability index is used to quantify the observability of the contribution of the lead aircraft configuration of the lead aircraft combination to the navigation error of the wingmen in the hierarchical cooperative navigation system of UAV swarm. The leader selection module is used to iteratively search for target leader combinations based on the binary particle swarm optimization algorithm, using the observability index of the UAV swarm as the fitness function. In the iteration, the inertia weight of the binary particle swarm optimization algorithm is continuously adjusted according to the relationship between the fitness value of the particles and the fitness value of the swarm, and the individual learning factor and the swarm learning factor are continuously adjusted according to the number of iterations. The matrix construction module is used to construct the inertial navigation position error equations for the wingman and wingman based on their positions in the Earth coordinate system and the distance between them calculated from the position output by the inertial navigation system. After Taylor expansion of the inertial navigation position error equation at the actual position and ignoring higher-order terms, the difference between the equation and the distance equation between the lead aircraft and the wingman obtained from relative ranging is calculated to obtain the pure distance error equation. By combining the pure distance error equations between each lead aircraft and each wingman, as well as the azimuth and pitch angles of each wingman relative to each lead aircraft, an observability matrix is ​​constructed. The observability index is: ； In the formula, The observability matrix The minimum singular value, The observability matrix The maximum singular value.

7. A computer device, characterized in that, include: At least one processor; And a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the leader selection method of the hierarchical cooperative navigation system for unmanned aerial vehicle swarms as described in any one of claims 1 to 5.

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the leader selection method of the hierarchical cooperative navigation system for unmanned aerial vehicle swarms as described in any one of claims 1 to 5.

Images