Unmanned aerial vehicle cluster dynamic measurement method based on cooperative positioning and multi-source fusion

Abstract

The invention belongs to the technical field of unmanned aerial vehicle surveying and mapping, and particularly relates to an unmanned aerial vehicle cluster dynamic measurement method based on cooperative positioning and multi-source fusion, and the method comprises the steps: constructing a cooperative operation mechanism of a dynamic reference station network and an observation unmanned aerial vehicle cluster, and combining global positioning data provided by RTK and relative distance measurement constraints provided by UWB, and multi-sensor observation information is fused to realize displacement calculation with centimeter-level precision. In addition, the flight path of the unmanned aerial vehicle cluster is optimized through a clustering algorithm and a heuristic optimization algorithm, and the measurement efficiency of the unmanned aerial vehicle cluster is further improved.

Description

Technical Field

[0001] This application belongs to the field of UAV surveying and mapping technology, specifically involving a dynamic measurement method for UAV swarms based on cooperative positioning and multi-source fusion. Background Technology

[0002] Existing displacement measurement methods mainly include traditional surveying techniques (such as total stations and levels), satellite positioning techniques (such as GPS), and measurements using a single UAV equipped with RTK (real-time kinematic) equipment. While these technologies are applied in various scenarios, they all have significant limitations: traditional methods rely on manual operation of equipment such as total stations and levels, requiring the installation of instruments at each measurement point and the collection of data point by point. In complex terrain (such as mountainous areas or disaster zones) or hazardous environments (such as landslides or near high-voltage power lines), manual operation is extremely inefficient and unsafe. While single UAVs equipped with RTK technology achieve centimeter-level positioning by receiving differential signals from ground reference stations, they suffer from limited coverage, poor adaptability to complex environments, and weak anti-interference capabilities in practical applications. Therefore, considering existing technologies, there is an urgent need for a new displacement measurement method with the ability to simultaneously measure over a large area, robustness to complex environments, and real-time dynamic calibration capabilities. Summary of the Invention

[0003] This invention addresses the problems of limited coverage, insufficient positioning accuracy in complex environments, and low measurement efficiency in existing displacement measurement technologies. It proposes a dynamic measurement method for UAV swarms based on cooperative positioning and multi-source fusion. The technical solution is as follows:

[0004] A dynamic measurement method for UAV swarms based on cooperative localization and multi-source fusion includes the following steps:

[0005] S1. By comprehensively considering three factors—satellite common-view analysis, communication signal strength, and spatial distribution uniformity—a multi-objective optimization model is constructed, and an improved particle swarm optimization algorithm is used to solve for the optimal solution for selecting the location of the UAV at the base station.

[0006] S2. The observation of the drone's flight path is transformed into a multiple traveling salesman problem with multiple drones corresponding to multiple survey points;

[0007] S3. After both the base station drone and the observation drone arrive at the designated location, the base station drone broadcasts RTK differential data once every certain period of time, and each observation drone receives the differential signal and calculates its own absolute positioning.

[0008] S4. Observe the target point;

[0009] S5. Location information data fusion.

[0010] Preferably, taking into account satellite common-view analysis, communication signal strength, and spatial distribution uniformity, the objective function is defined as:

[0011] F = w1F1 + w2F2 + w3F3;

[0012] in:

[0013] F1: Satellite common view analysis score; the smaller the value, the better the common view.

[0014] F2: Communication signal strength score; the higher the value, the stronger the signal.

[0015] F3: Spatial distribution uniformity score; the smaller the value, the more uniform the distribution.

[0016] w1, w2, w3: Adaptive weighting coefficients that are dynamically adjusted based on real-time data;

[0017] Satellite common-view analysis score:

[0018]

[0019] Where M is the number of candidate locations, J is the total number of satellites, and j is the number of visible satellites;

[0020] Communication signal strength rating:

[0021]

[0022] P max The maximum communication strength is given by p, where p is the communication strength.

[0023] Spatial distribution uniformity score:

[0024]

[0025] Where L is the average spacing and l is the spacing between candidate positions;

[0026] The weighting coefficients are dynamically adjusted based on real-time data and are defined as follows:

[0027]

[0028] Where k = 1, 2, 3 correspond to F1, F2, F3 respectively; v is the current objective function value, and V is the objective function value.

[0029] Preferably, the improved particle swarm optimization algorithm accelerates the convergence process by dynamically adjusting the inertia weight and learning factor;

[0030] The improvements include:

[0031] Dynamic inertia weight:

[0032]

[0033] Where t is the current iteration number, T is the maximum iteration number, and w max w min These are the maximum and minimum inertia weights, respectively.

[0034] Adaptive learning factor:

[0035]

[0036] Where c1 and c2 are both adaptive learning factors, representing the weights of the particle learning towards its own optimal position and the global optimal position, respectively; O max Maximum fitness value, where 'o' is the fitness value;

[0037] c 1max c 1min Let c1 be the maximum and minimum values;

[0038] c 2max c 2min Let c2 be the maximum and minimum values.

[0039] Preferably, the observation of the drone's flight path is transformed into a multiple traveling salesman problem with multiple drones corresponding to multiple survey points, and the steps are as follows:

[0040] S21. Using a clustering algorithm, the target points are divided into R clusters based on the set of coordinates of the survey points, namely longitude, latitude, elevation, and weight parameters. Then, a heuristic optimization algorithm is used to transform the multiple traveling salesman problem into a single traveling salesman problem for each observation drone, thereby planning the flight route of each observation drone.

[0041] S22. The heuristic optimization algorithm uses an improved genetic algorithm combined with a 2-opt local search strategy to solve the Multiple Traveling Salesman Problem. The following is the specific improved heuristic optimization algorithm:

[0042] S221. Initialize the population: Randomly generate a certain number of initial solutions, each solution representing the flight path of a drone;

[0043] S222. Fitness Evaluation: Calculate the fitness value for each solution. The fitness value is the reciprocal of the objective function.

[0044]

[0045] Where, d i Let N represent the distance of the i-th flight, and N represent the number of flights.

[0046] An improved genetic algorithm combined with a 2-opt local search strategy is used to solve the multiple traveling salesman problem. This algorithm generates new solutions through genetic operations and uses the 2-opt algorithm to locally optimize the solutions.

[0047] S223. Selection operator: Use the roulette wheel selection method to select individuals with higher fitness from the population to enter the parent generation;

[0048] S224. Crossover operator: Performs a crossover operation on the parent individuals using partial matching crossover to generate child individuals;

[0049] S225. Mutation operator: Performs a mutation operation on the offspring individuals, randomly swapping the positions of two individuals;

[0050] S226. Local Search: Apply the 2-opt algorithm to each offspring individual to further optimize the flight path; the 2-opt algorithm reduces the total distance by swapping two edges in the path, as detailed below:

[0051] New path × original path - (a→b) - (c→d) + (a→c) + (b→d);

[0052] Where a, b, c, and d represent four points in the flight path;

[0053] S227. Update the population: Compare the optimized offspring individuals with the parent individuals, retain the individuals with higher fitness, and update the population;

[0054] S228. Termination condition: The algorithm terminates when the maximum number of iterations is reached or the fitness value no longer changes significantly.

[0055] Preferably, after both the base station UAV and the observation UAV arrive at the designated location, the base station UAV broadcasts RTK differential data once per second, including latitude, longitude, elevation, and covariance matrix. Each observation UAV receives the differential signal and calculates its own absolute positioning, i.e., the latitude and longitude information of each UAV. At the same time, each UAV obtains its distance information from each other through UWB and transmits the distance information to each other through the Mesh network to obtain the relative positioning of the UAV cluster, i.e., the distance information of the relative positions of each UAV. The attitude information of the UAV is measured, including pitch angle, roll angle, and yaw angle.

[0056] Preferably, in step S4, the straight-line distance between the UAV and the observed target is measured using the time-of-flight method, which is then used to calculate the actual coordinates of the observed object in conjunction with the UAV's positioning information. When the observed target exhibits displacement, high-frequency sampling distance data is used to extract displacement features through time series analysis, thereby enabling displacement observation of the target.

[0057] Preferably, the observation UAV transmits the absolute positioning information acquired by RTK, the relative positioning information acquired by UWB, and the attitude information measured by IMU back to the base station UAV via a low-power data radio. The base station UAV then transmits this positioning and attitude information back to the ground control station via a high-power data radio. At the ground control station, the UAV's attitude information is converted into position information using an improved strapdown inertial navigation system (SINS) algorithm. This algorithm, based on the traditional SINS dual-integral framework, introduces an adaptive weighted error feedback mechanism based on sensor noise to suppress the divergence of IMU integration error over time.

[0058] P new (t)=P SINS (t)+λ(t)×ΔP feedbck (t);

[0059] in:

[0060] P SINS (t): Position information obtained by integrating the IMU acceleration twice using the traditional SINS algorithm;

[0061] ΔP feedbck (t): The feedback correction amount obtained by estimating through a sliding window;

[0062] λ(t): Weighting factor, adaptively adjusted according to sensor stability, defined as follows:

[0063] λ(t)=exp(-μ×σ a (t));

[0064] Where σ a (t) represents the standard deviation of acceleration noise within the current sliding window, and μ is an adjustment parameter that enhances the feedback correction when the IMU data is stable.

[0065] Preferably, in step S5, the absolute positioning information acquired by RTK, the relative positioning information acquired by UWB, and the position information calculated by IMU are synchronized in time and coordinate system 1: the time transformation is aligned with the GNSS 1PPS signal as the reference; the coordinate system transformation is aligned with the WGS84 / local coordinate system as the reference, thus obtaining three positioning information sources of time and coordinate system 1; finally, the three positioning information are input into a Kalman filter to obtain more accurate positioning information, and finally combined with the fused high-precision UAV positioning information (X u Y u Z u Using the angle information (θ, φ) and the three-dimensional spatial straight-line distance and slant distance γ between the UAV and the observation target point, the three-dimensional coordinates of the observation target are calculated using the spatial vector projection method. The calculation formula is as follows:

[0066] ΔE = γ·cosφ·sinθ;

[0067] ΔN = γ·cosφ·cosθ;

[0068] ΔU=γ·sinφ;

[0069] X target =X u +ΔE;

[0070] Y target =Y u +ΔN;

[0071] Z target =Z u +ΔU;

[0072] Where ΔE, ΔN, and ΔU are the UAV coordinate offsets, and X... target Y target Z target The absolute coordinates of the target point; azimuth angle θ, elevation angle φ;

[0073] The offset of the target in the UAV coordinate system is calculated using γ, θ, and φ, and then the offset is superimposed on the UAV coordinate system X. u Y u Z u This outputs the absolute coordinates of the target.

[0074] Compared with the prior art, the beneficial effects of this application are as follows:

[0075] By leveraging the collaborative capabilities of drones working along optimized routes, a large-scale, high-efficiency observation method is provided.

[0076] Data fusion was achieved for positioning modules with different characteristics in three application scenarios: UWB, IMU, and RTK, which improved measurement accuracy.

[0077] By leveraging the collaborative efforts of drones, the geometric distribution between base stations and observation devices becomes more controllable, effectively increasing satellite common visibility and spatial uniformity.

[0078] The system robustness is enhanced. Even if a single positioning module fails due to interference, the remaining positioning modules can still operate normally and obtain correct positioning information due to their different anti-interference capabilities.

[0079] The ranging module combines a laser rangefinder with a monocular camera rangefinder, further improving observation accuracy. Attached Figure Description

[0080] Figure 1 This is a diagram illustrating the system hardware configuration of the present invention;

[0081] Figure 2 This is the drone route task allocation diagram of the present invention;

[0082] Figure 3 This is a flowchart of the displacement information acquisition process of the present invention;

[0083] Figure 4 This is a data fusion diagram of the present invention;

[0084] Figure 5 This is a schematic diagram of a wind turbine tower observation example. Detailed Implementation

[0085] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention.

[0086] This invention proposes a dynamic measurement method for UAV swarms based on cooperative localization and multi-source fusion, including the following aspects:

[0087] (1) System hardware composition:

[0088] Base station drone: Equipped with a dual-frequency RTK module (supporting GPS / BeiDou / Galileo), UWB ranging module, IMU (Inertial Measurement Unit), high-power data radio, and edge computing unit.

[0089] Observation UAV: ​​Equipped with RTK module, UWB module, IMU (Inertial Measurement Unit), long-range ranging module (including laser rangefinder and monocular camera rangefinder), low-power data radio, and edge computing unit. Ground control station: Mission planning software and data fusion server.

[0090] (2) Drone swarm route task allocation:

[0091] Base station drone deployment: Based on satellite common-view analysis, communication signal strength and spatial distribution uniformity, base station drones are deployed in locations with few obstacles and in the central area of ​​each observation target point.

[0092] Observing the drone's route: A clustering algorithm is used to cluster the target points, transforming the multiple traveling salesman problem into multiple single traveling salesman problems. Then, a heuristic optimization algorithm is used to solve the single traveling salesman problems.

[0093] (3) Cooperative positioning and observation point calculation:

[0094] The base station drone calibrates its initial position using PPP (Precise Point Positioning) or by utilizing ground control points.

[0095] Communication links are established between each UAV node through a Mesh network (2.4GHz, with a single hop distance of 1-2km) to transmit positioning and observation information.

[0096] The base station drone broadcasts RTK differential data in real time, and the observation drone receives the data and calculates the absolute coordinates at the centimeter level.

[0097] The UAVs exchange relative distance data at high frequency through the UWB module to obtain the relative coordinates of the UAVs for cooperative positioning, build a distance constraint network, and suppress single-point positioning drift.

[0098] The drone obtains its current attitude angle information through the IMU module.

[0099] The remote ranging module measures the straight-line distance between the survey object and the observation drone.

[0100] (4) Multi-source data fusion:

[0101] The acceleration and angular velocity output by the IMU are processed by the SINS algorithm (straight-through inertial navigation algorithm) to calculate the position, velocity and attitude information of the UAV.

[0102] The time synchronization and coordinate system integration of data from three sources—RTK, UWB, and IMU—are achieved in the data fusion server of the ground control station.

[0103] The absolute position of the UAV obtained by RTK, the relative position between UAVs obtained by UWB, and the position information calculated by IMU are fused by a Kalman filter to obtain a high-precision positioning result for the UAV.

[0104] The high-precision UAV positioning information obtained by data fusion is combined with the straight-line distance parameters from the observation UAV to the survey object to obtain the three-dimensional position coordinates of the survey object.

[0105] A dynamic measurement method for UAV swarms based on cooperative localization and multi-source fusion includes the following steps:

[0106] (1) Deploy base station drones:

[0107] Ground control stations are deployed nearby. Based on the high-precision GNSS receivers and data transmission radios onboard the UAVs, satellite common-view analysis, communication signal strength, and spatial distribution uniformity are analyzed. Three or more base station UAVs are deployed in locations with few obstacles and centrally located among the observation target points. An improved algorithm is used: a base station UAV location selection algorithm based on multi-objective optimization. This algorithm constructs a multi-objective optimization model by comprehensively considering satellite common-view analysis, communication signal strength, and spatial distribution uniformity, and uses an improved particle swarm optimization (PSO) algorithm to solve for the optimal solution. Improvements include an adaptive weight allocation strategy and dynamic adjustment of the search range to improve the algorithm's convergence speed and the global optimality of the solution. Details are as follows:

[0108] 1. Initialization: Input candidate location coordinates, satellite common-view data, communication signal strength data, and region geometry; initialize particle swarm parameters (particle number, inertia weight, learning factor, etc.).

[0109] 2. Calculate the objective function: Calculate the satellite common view score, communication signal strength score, and spatial distribution uniformity score for each candidate location. Based on an adaptive weight allocation strategy, dynamically adjust the weight coefficients to calculate the comprehensive objective function.

[0110] 3. Particle swarm optimization: Update particle velocity and position, dynamically adjust inertia weight and learning factor, and record the optimal solution.

[0111] 4. Termination Condition: The algorithm terminates when the maximum number of iterations is reached or the fitness value no longer changes significantly. Output: Outputs the optimal set of base station UAV locations and a comprehensive score.

[0112] Improved algorithm core formula:

[0113] 1. Objective function

[0114] Taking into account satellite common-view analysis, communication signal strength, and spatial distribution uniformity, the objective function is defined as:

[0115] F = w1F1 + w2F2 + w3F3;

[0116] in:

[0117] F1: Satellite common view analysis score (the smaller the value, the better the common view);

[0118] F2: Communication signal strength score (the higher the value, the stronger the signal);

[0119] F3: Spatial distribution uniformity score (the smaller the value, the more uniform the distribution);

[0120] w1, w2, w3: Adaptive weighting coefficients that are dynamically adjusted based on real-time data.

[0121] 2. Scoring Function

[0122]

[0123] Where M represents the number of candidate locations, J represents the total number of satellites, and j represents the number of visible satellites.

[0124]

[0125] Where P is the maximum communication strength and p is the communication strength.

[0126]

[0127] Where L is the average spacing and l is the spacing between candidate positions.

[0128] 3. Adaptive weight allocation

[0129] The weighting coefficients are dynamically adjusted based on real-time data and are defined as follows:

[0130]

[0131] Where k = 1, 2, 3 correspond to F1, F2, F3 respectively, v is the current objective function value, and V is the objective function value.

[0132] 4. Improved Particle Swarm Optimization (PSO)

[0133] The particle swarm optimization algorithm accelerates the convergence process by dynamically adjusting the inertia weights and learning factors. Improvements include:

[0134] Dynamic inertia weight:

[0135]

[0136] Where t is the current iteration number, T is the maximum iteration number, and w max w min These are the maximum and minimum inertia weights, respectively.

[0137]

[0138] Where c1 and c2 are adaptive learning factors, representing the weights of the particle learning towards its own optimal position and the global optimal position, respectively. max Maximum fitness value, where 'o' is the fitness value;

[0139] c 1max c 1min The maximum and minimum values ​​of c1;

[0140] c 2max c 2min The maximum and minimum values ​​of c2;

[0141] Base station drones consist of, for example Figure 1 As shown, after the base station drone reaches the designated location, it performs precise single-point positioning via PPP. The positioning accuracy is converged by hovering for 15 minutes. At the same time, the elevation deviation is corrected by the ground control station.

[0142] (2) Observing the drone's route

[0143] The composition of the observation drone is as follows Figure 1 As shown, the flight path of the observation drones falls under the Multiple Traveling Salesman Problem (MTSP) involving multiple drones and multiple survey points. First, a clustering algorithm is used to divide the target points into R clusters based on the coordinate set (longitude, latitude, and elevation) and weight parameters (terrain complexity and priority). Then, a heuristic optimization algorithm transforms the MTSP into a Single Traveling Salesman Problem (TSP) for each observation drone, thus planning the flight path for each drone. The heuristic optimization algorithm employs a modified Genetic Algorithm (GA) combined with a 2-opt local search strategy to solve the MTSP.

[0144] The following are specific improvements to the heuristic optimization algorithm:

[0145] 1. Initialize the population

[0146] A certain number of initial solutions are randomly generated, and each solution represents the flight path of a drone.

[0147] 2. Fitness assessment

[0148] Calculate the fitness value for each solution, which is the reciprocal of the objective function (total flight distance or time).

[0149]

[0150] Where, d i Let N represent the distance of the i-th flight, and N represent the number of flights.

[0151] An improved genetic algorithm combined with a 2-opt local search strategy is proposed for solving the Multiple Traveling Salesman Problem (MTSP). This algorithm generates new solutions through genetic operations and uses the 2-opt algorithm to locally optimize the solutions, thereby improving both solution quality and computational efficiency.

[0152] 3. Selecting Operators

[0153] Using the roulette wheel selection method, individuals with higher fitness are selected from the population to enter the parent generation.

[0154] 4. Crossover operator

[0155] Partial matching crossover (PMX) is used to crossover the parent individuals to generate offspring individuals.

[0156] 5. Mutation Operator

[0157] Perform a mutation operation on the offspring individuals, randomly swapping the positions of two points.

[0158] 6. Local search (2-opt)

[0159] The 2-opt algorithm is applied to each offspring individual to further optimize the flight path. The 2-opt algorithm reduces the total distance by swapping two edges in the path. The specific operation is as follows:

[0160] New path = Original path - (a→b) - (c→d) + (a→c) + (b→d)

[0161] Here, a, b, c, and d represent four points in the path.

[0162] 7. Update the population

[0163] The optimized offspring individuals are compared with the parent individuals, and individuals with higher fitness are retained to update the population.

[0164] 8. Termination Conditions

[0165] The algorithm terminates when the maximum number of iterations is reached or the fitness value no longer changes significantly.

[0166] This yields the optimal flight path for a single UAV, and finally, flight tasks are assigned to the observation UAVs via the ground control station.

[0167] (3) Unmanned Aerial Vehicle Cooperative Positioning

[0168] After both the base station UAV and the observation UAV arrive at the designated location, the base station UAV broadcasts RTK differential data (including latitude, longitude, elevation, and covariance matrix) once per second. Each observation UAV receives the differential signal and calculates its own absolute positioning (i.e., the latitude and longitude information of each UAV). At the same time, each UAV (including the base station UAV and the observation UAV) obtains distance information from each other through UWB and transmits distance information to each other through the Mesh network to obtain the relative positioning of the UAV cluster (i.e., the distance information of the relative positions of each UAV). Finally, the IMU module measures the attitude information of the UAV (including pitch angle, roll angle, and yaw angle). (4) The ranging module observes the target point.

[0169] The ranging module consists of a laser rangefinder and a monocular camera rangefinder. The laser rangefinder has two functions: first, it measures the straight-line distance between the UAV and the observed target using the time-of-flight method, which is then combined with the UAV's positioning information to calculate the actual coordinates of the observed object; second, when the observed target exhibits displacement, it samples distance data at a high frequency (40Hz) and extracts displacement features (such as amplitude and frequency) through time-series analysis, thereby achieving displacement observation of the target. The monocular camera rangefinder uses a monocular vision displacement monitoring method independently developed by our team. Through an original oblique photography enhancement strategy and the TransUPerNet background segmentation model, combined with a multi-parameter displacement transformation model (MDTM), it jointly corrects the pitch angle θ, yaw angle β, and principal ray angle η, achieving multi-target sub-millimeter level synchronous displacement measurement (measured accuracy ±0.12mm, monitoring distance 150m). The core technology has been verified through domestically produced hardware. In addition, the monocular camera rangefinder also has the function of observing the appearance of the observed target to determine whether the target has obvious damage.

[0170] (5) Location information data fusion

[0171] The observation UAV transmits absolute positioning information acquired by RTK, relative positioning information acquired by UWB, and attitude information measured by IMU back to the base station UAV via a low-power data radio. The base station UAV then transmits this positioning and attitude information back to the ground control station via a high-power data radio. At the ground control station, the UAV's attitude information is converted into position information using an improved strapdown inertial navigation system (SINS) algorithm. This algorithm, based on the traditional SINS dual-integral framework, introduces an adaptive weighted error feedback mechanism based on sensor noise to suppress the divergence of IMU integration error over time. The core formula of the improved algorithm is as follows:

[0172] P new (t)=P SINS (t)+λ(t)×ΔP feedbck (t)

[0173] in:

[0174] P SINS (t): Position information obtained by the traditional SINS algorithm (obtained by integrating the IMU acceleration twice);

[0175] ΔP feedbck (t): The feedback correction amount obtained by estimating through a sliding window;

[0176] λ(t): Weighting factor, adaptively adjusted according to sensor stability, defined as follows:

[0177] λ(t)=exp(-μ×σ a (t));

[0178] Where σ a (t) represents the standard deviation of acceleration noise within the current sliding window, and μ is the adjustment parameter. This factor enhances the feedback correction when IMU data is stable and automatically weakens the feedback when there is significant vibration or unstable data, effectively improving the accuracy and robustness of the location information. Subsequently, the absolute positioning information acquired by RTK, the relative positioning information acquired by UWB, and the location information calculated by IMU are synchronized in time and coordinate system one: time transformation is aligned with the GNSS1PPS signal as the reference; coordinate system transformation is aligned with the WGS84 / local coordinate system as the reference, thus obtaining three sources of positioning information in terms of time and coordinate system one. Finally, the three types of positioning information are input into a Kalman filter to obtain more accurate positioning information, such as... Figure 4 As shown. Finally, the fused high-precision UAV positioning information (X) is combined. u Y u Z u The three-dimensional coordinates of the observed target are calculated using the spatial vector projection method, along with angle information (azimuth θ, elevation φ) and the three-dimensional spatial straight-line distance and slant distance γ between the UAV and the observation target point measured by the ranging module.

[0179] The core formula is as follows:

[0180] ΔE = γ·cosφ·sinθ;

[0181] ΔN = γ·cosφ·cosθ;

[0182] ΔU=γ·sinφ;

[0183] X target =X u +ΔE;

[0184] Y target =Y u +ΔN;

[0185] Z target =Z u +ΔU;

[0186] Where ΔE, ΔN, and ΔU are the UAV coordinate offsets, and X... target Y target Z target absolute coordinates of the target point

[0187] The specific steps are as follows: calculate the target's offset in the UAV coordinate system using γ, θ, and φ, and then superimpose the offset onto the UAV coordinate system X. u Y u Z u This outputs the absolute coordinates of the target.

[0188] The above description is only one embodiment of the present invention. All equivalent changes and modifications made within the scope of the claims of the present invention shall be covered by the present invention.

Claims

1. A dynamic measurement method for UAV swarms based on cooperative localization and multi-source fusion, characterized in that, Includes the following steps: S1. By comprehensively considering three factors—satellite common-view analysis, communication signal strength, and spatial distribution uniformity—a multi-objective optimization model is constructed, and an improved particle swarm optimization algorithm is used to solve for the optimal solution for selecting the location of the UAV at the base station. S2. The observation of the drone's flight path is transformed into a multiple traveling salesman problem with multiple drones corresponding to multiple survey points; S3. After both the base station drone and the observation drone arrive at the designated location, the base station drone broadcasts RTK differential data once every certain period of time, and each observation drone receives the differential signal and calculates its own absolute positioning. S4. Observe the target point; S5. Location information data fusion.

2. The method for dynamic measurement of UAV swarms based on cooperative localization and multi-source fusion according to claim 1, characterized in that, Taking into account satellite common-view analysis, communication signal strength, and spatial distribution uniformity, the objective function is defined as: F = w1F1 + w2F2 + w3F3; in: F1: Satellite common view analysis score; the smaller the value, the better the common view. F2: Communication signal strength score; the higher the value, the stronger the signal. F3: Spatial distribution uniformity score; the smaller the value, the more uniform the distribution. w1, w2, w3: Adaptive weighting coefficients that are dynamically adjusted based on real-time data; Satellite common-view analysis score: Where M is the number of candidate locations, J is the total number of satellites, and j is the number of visible satellites; Communication signal strength rating: P max The maximum communication strength is given by p, where p is the communication strength. Spatial distribution uniformity score: Where L is the average spacing and l is the spacing between candidate positions; The weighting coefficients are dynamically adjusted based on real-time data and are defined as follows: Where k = 1, 2, 3 correspond to F1, F2, F3 respectively; v is the current objective function value, and V is the objective function value.

3. The method for dynamic measurement of UAV swarms based on cooperative positioning and multi-source fusion according to claim 2, characterized in that, The improved particle swarm optimization algorithm accelerates the convergence process by dynamically adjusting the inertia weight and learning factor; Dynamic inertia weight: Where t is the current iteration number, T is the maximum iteration number, and w max w min These are the maximum and minimum inertia weights, respectively; Adaptive learning factor: Where c1 and c2 are both adaptive learning factors, representing the weights of the particle learning towards its own optimal position and the global optimal position, respectively; O max Maximum fitness value, where 'o' is the fitness value; c 1max c 1min Let c1 be the maximum and minimum values; c 2max c 2min Let c2 be the maximum and minimum values.

4. The method for dynamic measurement of UAV swarms based on cooperative localization and multi-source fusion according to claim 2, characterized in that, The observation of the drone's flight path is transformed into a multiple traveling salesman problem involving multiple drones and multiple survey points. The steps are as follows: S21. Using a clustering algorithm, the target points are divided into R clusters based on the set of coordinates of the survey points, namely longitude, latitude, elevation, and weight parameters. Then, a heuristic optimization algorithm is used to transform the multiple traveling salesman problem into a single traveling salesman problem for each observation drone, thereby planning the flight route of each observation drone. S22. The heuristic optimization algorithm uses an improved genetic algorithm combined with a 2-opt local search strategy to solve the Multiple Traveling Salesman Problem. The following is the specific improved heuristic optimization algorithm: S221. Initialize the population: Randomly generate a certain number of initial solutions, each solution representing the flight path of a drone; S222. Fitness Evaluation: Calculate the fitness value for each solution. The fitness value is the reciprocal of the objective function. Where, d i Let N represent the distance of the i-th flight, and N represent the number of flights. An improved genetic algorithm combined with a 2-opt local search strategy is used to solve the multiple traveling salesman problem. This algorithm generates new solutions through genetic operations and uses the 2-opt algorithm to locally optimize the solutions. S223. Selection operator: Use the roulette wheel selection method to select individuals with higher fitness from the population to enter the parent generation; S224. Crossover operator: Performs a crossover operation on the parent individuals using partial matching crossover to generate child individuals; S225. Mutation operator: Performs a mutation operation on the offspring individuals, randomly swapping the positions of two individuals; S226. Local Search: Apply the 2-opt algorithm to each offspring individual to further optimize the flight path; the 2-opt algorithm reduces the total distance by swapping two edges in the path, as detailed below: New path = original path - (a→b) - (c→d) + (a→c) + (b→d); Where a, b, c, and d represent four points in the flight path; S227. Update the population: Compare the optimized offspring individuals with the parent individuals, retain the individuals with higher fitness, and update the population; S228. Termination condition: The algorithm terminates when the maximum number of iterations is reached or the fitness value no longer changes significantly.

5. The method for dynamic measurement of UAV swarms based on cooperative localization and multi-source fusion according to claim 1, characterized in that, After both the base station drone and the observation drone arrive at their designated locations, the base station drone broadcasts RTK differential data once per second, including latitude, longitude, elevation, and covariance matrix. Each observation drone receives the differential signal and calculates its own absolute positioning, i.e., the latitude and longitude information of each drone. At the same time, the drones obtain each other's distance information through UWB and transmit distance information to each other through the Mesh network to obtain the relative positioning of the drone cluster, i.e., the distance information of the relative positions of each drone. The drone attitude information is measured, including pitch angle, roll angle, and yaw angle.

6. The method for dynamic measurement of UAV swarms based on cooperative localization and multi-source fusion according to claim 1, characterized in that, Step S4 measures the straight-line distance between the UAV and the observed target using the time-of-flight method, which is then used to calculate the actual coordinates of the observed object in conjunction with the UAV's positioning information. When the observed target exhibits displacement, high-frequency sampling distance data is used to extract displacement features through time series analysis, thereby enabling the observation of the target's displacement.

7. The method for dynamic measurement of UAV swarms based on cooperative localization and multi-source fusion according to claim 1, characterized in that, The UAV attitude information is converted into position information using an improved strapdown inertial navigation system (SINS) algorithm. This algorithm introduces an adaptive weighted error feedback mechanism based on sensor noise on the basis of the traditional SINS dual-integral framework to suppress the problem of IMU integration error diverging over time. P new (t)=P SINS (t)+λ(t)×ΔP feedbck (t); in: P SINS (t): Position information obtained by integrating the IMU acceleration twice using the traditional SINS algorithm; ΔP feedbck (t): The feedback correction amount obtained by estimating through a sliding window; λ(t): Weighting factor, adaptively adjusted according to sensor stability, defined as follows: λ(t)=exp(-μ×σ a (t)); Where σ a (t) represents the standard deviation of acceleration noise within the current sliding window, and μ is an adjustment parameter that enhances the feedback correction when the IMU data is stable.

8. The method for dynamic measurement of UAV swarms based on cooperative localization and multi-source fusion according to claim 1, characterized in that, In step S5, the absolute positioning information acquired by RTK, the relative positioning information acquired by UWB, and the position information calculated by IMU are synchronized in time and coordinate system 1: time transformation is aligned with GNSS 1PPS signal as the reference; coordinate system transformation is aligned with WGS84 / local coordinate system as the reference, thus obtaining three positioning information sources of time and coordinate system 1; finally, the three positioning information are input into a Kalman filter to obtain more accurate positioning information, and finally combined with the fused high-precision UAV positioning information (X u Y u Z u Using the angle information (θ, φ) and the three-dimensional spatial straight-line distance and slant distance γ between the UAV and the observation target point, the three-dimensional coordinates of the observation target are calculated using the spatial vector projection method. The calculation formula is as follows: ΔE = γ·cosφ·sinθ; ΔN = γ·cosφ·cosθ; ΔU=γ·sinφ; X target =X u +ΔE; Y target =Y u +ΔN; WITH target =Z u +ΔU; Where ΔE, ΔN, and ΔU are the UAV coordinate offsets, and X... target Y target Z target The absolute coordinates of the target point; azimuth angle θ, elevation angle φ; The offset of the target in the UAV coordinate system is calculated using γ, θ, and φ, and then the offset is superimposed on the UAV coordinate system X. u Y u Z u This outputs the absolute coordinates of the target.

Images