Intelligent unmanned aerial vehicle flight control method based on 4D flight path

Abstract

The invention discloses an intelligent unmanned aerial vehicle flight control method based on a 4D flight path, and the method comprises the steps: obtaining real-time airspace data for an initial four-dimensional flight path set, carrying out the preprocessing of environment information collected by a sensor through a data compression algorithm, transmitting the information to a calculation center through a satellite-based communication link, and obtaining a compressed real-time airspace data package; according to the corrected track sequence, the track adjustment speed of the track segment is optimized in the satellite-based computing center, a millisecond-level adjustment instruction set is generated for the obstacle avoidance requirement in the complex airspace, and a final track sequence containing the adjustment speed is obtained; and for the control sequence adjusted in real time, uploading actual track data to a computing center, and updating a task database through comparative analysis of the actual track data and the initial four-dimensional track set to obtain a track data set for subsequent task optimization. According to the invention, accurate flight path planning and dynamic optimization of the large-scale unmanned aerial vehicle group are realized, and the flight efficiency and safety in a complex airspace are improved.

Description

Technical Field

[0001] The present invention belongs to the field of information technology, and in particular relates to an intelligent unmanned aerial vehicle flight control method based on 4D trajectory. Background Art

[0002] Unmanned aerial vehicle (UAV) trajectory planning and autonomous flight are important research areas in the aviation field, crucial for improving airspace utilization efficiency and ensuring flight safety. With the rapid expansion of UAV applications, particularly in scenarios such as logistics and emergency rescue in complex airspace, efficient and accurate autonomous flight technology has become a pressing need for industry development. Traditional UAV trajectory planning relies heavily on static calculations or manual intervention by ground control stations, resulting in low computational efficiency, poor real-time performance, and an inability to adapt to dynamic airspace changes. These limitations make it difficult to achieve high-precision spatiotemporal control of UAVs in complex environments, restricting their large-scale commercial application. The core challenges in trajectory planning and autonomous flight stem from the interrelated nature of the following technical factors. First, generating four-dimensional (4D) trajectories for flight plans requires efficient computing power to coordinate the massive daily volume of flight missions. While current computing centers can perform batch calculations, they struggle to quickly generate high-precision trajectories in real time. Second, limited by the data processing capabilities of onboard equipment, the real-time data acquired by UAVs during flight cannot be quickly converted into effective flight information, resulting in delayed trajectory adjustments. Furthermore, dynamic obstacles and weather changes in complex airspaces require UAVs to optimize their trajectories within milliseconds, but existing algorithms lack the precision to accurately synchronize spatiotemporal control. These factors collectively present core technical challenges for trajectory planning and autonomous flight. Therefore, the key challenge is how to achieve efficient four-dimensional trajectory generation, real-time data processing, and dynamic trajectory optimization through satellite-based technology to ensure millisecond-level spatiotemporal synchronization of UAVs in complex airspaces. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to address the deficiencies in the above-mentioned existing technologies and provide an intelligent UAV flight control method based on 4D trajectory, which realizes accurate trajectory planning and dynamic optimization of large-scale UAV swarms, and improves flight efficiency and safety in complex airspace.

[0004] To solve the above technical problems, the present invention adopts a technical solution: an intelligent UAV flight control method based on 4D trajectory, which includes the following steps: Obtain the initial 4D track data of the task assignment and generate 4D track data sets for the massive daily flight tasks; For the 4D trajectory data set, real-time airspace data is obtained from the drone's onboard sensors, and the data is compressed to obtain a compressed real-time airspace data packet; Based on the compressed real-time airspace data packet, the track point sequence in the 4D track data set is optimized to generate an adjusted candidate track set, thereby obtaining a candidate track point sequence containing the optimized path. From the candidate track point sequence, the time-space synchronization constraint algorithm is used to verify the timestamp and position deviation of each track point to obtain the optimized track sequence that meets the time-space synchronization. To optimize the track sequence, real-time weather update data is obtained from the satellite-based meteorological database, and the weather change trend is combined with the track sequence to obtain a corrected track sequence that adapts to the weather dynamics; According to the corrected track sequence, the trajectory adjustment speed of the track segment is optimized using a dynamic programming algorithm to obtain the final track sequence including the adjustment speed; From the final trajectory sequence, a data distribution algorithm is used to transmit the adjustment instruction set to the UAV onboard controller via a satellite-based communication link, obtaining a flight control sequence that can be directly executed by the UAV; According to the flight control sequence, the actual trajectory data after execution is obtained from the drone sensor, and the real-time adjusted control sequence is obtained by analyzing the deviation from the final trajectory sequence; For the control sequence adjusted in real time, the satellite-based data synchronization algorithm is used to upload the actual track data to the satellite-based computing center. By comparing and analyzing it with the initial 4D track set, the mission database is updated to obtain the track data set for subsequent mission optimization.

[0005] The above-mentioned intelligent UAV flight control method based on 4D trajectory, wherein the initial 4D trajectory data of the task assignment is obtained and a 4D trajectory dataset is generated for the massive daily flight tasks, includes: The initial 4D track data of daily flight missions is obtained from the satellite-based computing center and stored as a structured data set containing time, position, and velocity attributes to obtain the initial data set; A distributed computing framework is used to assign tasks to the initial data set. If the data volume exceeds a preset threshold, the task is decomposed into multiple subtasks and assigned to computing nodes to obtain a subtask data set. The subtask data set is processed by the space-time grid partitioning algorithm to generate gridded track points of time, position and speed, and obtain a gridded track point set; For the gridded track point set, the k-means clustering algorithm is used to classify the track points. If the time interval of the track points is less than the preset threshold, they are classified into the same track sequence to obtain a preliminary track sequence set. Based on the preliminary track sequence set, the time, position and speed sequences are completed using the linear interpolation method. If there are missing points in the sequence, they are filled in by interpolation based on the previous and next point data to obtain a complete track sequence set. Through the distributed computing framework, the complete track sequence set is verified in parallel. If the deviation between the time, position and speed data of the sequence and the initial data is less than the preset threshold, it is determined to be a valid sequence and the final track point sequence set is obtained. For the final track point sequence set, a 4D track data structure containing time, position and velocity is generated and stored in the satellite-based computing center to obtain a standardized 4D track data set.

[0006] The above-mentioned intelligent UAV flight control method based on 4D trajectory, wherein the real-time airspace data is obtained from the UAV airborne sensor for the 4D trajectory data set, and the data is compressed to obtain a compressed real-time airspace data packet, includes: For the 4D trajectory dataset, real-time airspace data is obtained from the drone's onboard sensors, and the environmental information is collected using a preset sampling frequency to obtain sensor data; For sensor data, a data compression algorithm is used to pre-process the environmental information. If the data redundancy exceeds the preset threshold, lossless compression is performed to obtain a compressed data packet. The compressed data packets are transmitted to the computing center via the satellite-based communication link. If the transmission bandwidth is lower than the preset threshold, the compression ratio is adjusted to obtain a stable data transmission flow. The data transmission stream is received at the computing center, and the data packet is restored using a decompression algorithm to obtain restored spatial domain data; For the restored spatial data, a clustering algorithm is used to extract features from the spatial data. If the dimension of the feature vector exceeds the preset threshold, dimensionality reduction is performed to obtain feature data. According to the feature data, the airspace data is classified using the preset matching rules to obtain the classified airspace data; For the classified airspace data, real-time airspace data packets are generated and stored in the preset database of the satellite-based computing center.

[0007] The above-mentioned intelligent UAV flight control method based on 4D track, wherein the track point sequence in the 4D track data set is optimized according to the compressed real-time airspace data packet to generate an adjusted candidate track set, and obtain a candidate track point sequence containing the optimized path, includes: By decompressing the real-time airspace data packet, the initial data set containing the track point sequence and environmental information is obtained to obtain structured airspace data; Using a deep reinforcement learning algorithm, we perform preliminary optimization of the track point sequence based on structured airspace data at the satellite-based computing center to obtain a preliminarily optimized track point sequence. Based on the dynamic obstacle data acquired in real time, if a conflict between an obstacle and the initially optimized track point sequence is detected, the track point coordinates are adjusted to obtain a conflict-avoiding track point sequence. By integrating weather change data, if the weather conditions exceed the preset threshold, the conflict avoidance trackpoint sequence is adjusted twice to obtain a trackpoint sequence that adapts to weather changes; A genetic algorithm is used to perform global optimization on the trackpoint sequence that adapts to weather changes, generating a set of candidate trackpoint sequences containing multiple optimized paths. By comparing the cost function values of each path in the candidate track point sequence set, the track point sequence with the lowest cost function value is determined to obtain the final optimized track point sequence; For the final optimized track point sequence, a track point sequence data packet that meets the airspace constraints is generated, and a candidate track point sequence containing the optimized path is obtained and output to the satellite-based computing center.

[0008] The above-mentioned intelligent UAV flight control method based on 4D trajectory adopts a spatiotemporal synchronization constraint algorithm from a candidate track point sequence to verify the timestamp and position deviation of each track point to obtain an optimized track sequence that satisfies spatiotemporal synchronization, including: Obtain initial track data from the candidate track point sequence, use a preset millisecond-level time window to determine the timestamp and position coordinates of each track point, and obtain the initial track point set; For the initial track point set, a spatiotemporal synchronization constraint algorithm is used to check the timestamp and position deviation of each track point. If the deviation exceeds the preset threshold, it is marked as an abnormal track point to obtain an abnormal point set; Obtain the marked track points from the outlier point set, use the local interpolation algorithm to recalculate the coordinates and timestamps of the outlier track points, and obtain the corrected local track segment; The corrected local track segment is used to replace the corresponding abnormal track points in the initial track point set to generate an updated track point set; For the updated track point set, a smoothing filtering algorithm is used to eliminate the sudden changes between track points and obtain a smooth track sequence; From the smoothed track sequence, the timestamps and position deviations of adjacent track points are checked. If the deviation exceeds the preset threshold, the local interpolation calculation is repeated to obtain the final optimized track sequence. Through the final optimized track sequence, track data that meets the time and space synchronization constraints are generated, and the optimized track sequence that meets the time and space synchronization is obtained.

[0009] The above-mentioned intelligent UAV flight control method based on 4D trajectory, wherein the method optimizes the trajectory sequence, obtains real-time weather update data from a satellite-based meteorological database, combines the weather change trend with the trajectory sequence, and obtains a corrected trajectory sequence that adapts to the weather dynamics, including: Obtain real-time meteorological data from satellite-based meteorological databases, parse the data format using standard protocols, and obtain structured meteorological data sets; The structured meteorological dataset is analyzed through a prediction model, and the long short-term memory network (LSTM) is used to calculate the time series features to obtain the meteorological change trend; If the fluctuation amplitude of the meteorological change trend exceeds the preset threshold, the sliding window method is used to smooth the change trend to obtain a smooth change trend; If the fluctuation range of the meteorological change trend is lower than the threshold, the change trend is directly output; According to the smooth change trend, a weighted fusion algorithm is used to calculate the matching weight of the track sequence and the change trend, and the weighted track adjustment parameters are obtained; By adjusting the weighted track parameters and using the geometric transformation method to correct the coordinates of the original track sequence, a preliminary corrected track sequence is generated. For the preliminary corrected track sequence, a dynamic programming algorithm is used to optimize the path smoothness between track points to obtain the final corrected track sequence; If the deviation between the final corrected track sequence and the real-time meteorological data exceeds the preset threshold, it returns to the weighted fusion step, adjusts the weight parameters, and regenerates the corrected track sequence; If the deviation is within the threshold, the final corrected track sequence is output to obtain a corrected track sequence that adapts to meteorological dynamics.

[0010] The above-mentioned intelligent UAV flight control method based on 4D trajectory, wherein the trajectory adjustment speed of the track segment is optimized according to the modified track sequence using a dynamic programming algorithm to obtain a final track sequence including the adjusted speed, includes: Obtain the corrected track sequence from the satellite-based computing center, and use the segment analysis method to extract the coordinates and velocity information of the track segments to obtain the track segment data set; Based on the trajectory segment data set, a dynamic programming algorithm is used to calculate the adjusted speed of each trajectory segment, and the speed adjustment parameters are generated according to the dynamic constraints of complex airspace. By adjusting the speed parameters and combining the dynamic characteristics of the airspace, the track segment coordinates are fine-tuned using the geometric transformation method to obtain a preliminary adjusted track sequence. If the deviation between the initial adjustment trajectory sequence and the obstacle avoidance requirement exceeds the preset threshold, the system returns to the dynamic programming algorithm step, adjusts the speed parameters, and regenerates the initial adjustment trajectory sequence; If the deviation is within the threshold, the preliminary adjusted track sequence is output; According to the preliminary adjusted track sequence, the interpolation smoothing method is used to optimize the speed transition between track segments and generate a smoothed adjusted track sequence; Real-time airspace dynamic characteristics are obtained from the satellite-based computing center. If the matching degree between the smoothed adjusted track sequence and the airspace dynamic characteristics is lower than the preset threshold, the track segment coordinates are adjusted using a weighted correction method to obtain the final adjusted track sequence. If the matching degree is within the threshold, the final adjusted track sequence is directly output; Through the final adjustment of the track sequence, the instruction encoding method is used to generate a millisecond-level adjustment instruction set to obtain the final track sequence including speed adjustment.

[0011] The above-mentioned intelligent UAV flight control method based on 4D trajectory adopts a data distribution algorithm from the final trajectory sequence to transmit the adjustment instruction set to the UAV onboard controller via a satellite-based communication link to obtain a flight control sequence that can be directly executed by the UAV, including: From the final track sequence, the data distribution algorithm is used to decompose and generate the adjustment instruction set and determine the initial structure of the instruction set; Transmitting the adjusted instruction set to the UAV onboard controller via a satellite-based communication link to obtain the integrity status of the transmitted instruction set; If the integrity status of the transmitted instruction set meets a preset threshold, the instruction set is parsed in the onboard controller to obtain the parsed instruction data; If not, the transmission is re-transmitted via the satellite-based communication link; In view of the data processing capability of the airborne equipment, a lightweight algorithm is used to compress the parsed instruction data to generate lightweight execution instructions; Extract control parameters from lightweight execution instructions to generate flight control sequences that can be directly executed by the UAV; Verify the compatibility of the flight control sequence through the onboard controller to determine whether the sequence meets the UAV execution control requirements; If the flight control sequence meets the execution control requirements, the sequence is loaded into the UAV execution module to obtain a flight control sequence that can be directly executed by the UAV.

[0012] The above-mentioned intelligent UAV flight control method based on 4D trajectory, which obtains the actual trajectory data after execution from the UAV sensor according to the flight control sequence, and obtains the real-time adjusted control sequence by analyzing the deviation from the final trajectory sequence, includes: Acquire actual track data from the drone sensor, use data preprocessing technology to remove noise and outliers, and obtain the first track data set; According to the first track data set and the target track sequence, the Euclidean distance algorithm is used to calculate the deviation value between the two to obtain a deviation data set; If the deviation value in the deviation data set exceeds the preset deviation threshold, the deviation analysis module is used to determine the track segment with the deviation exceeding the threshold, and obtain the exceeding threshold track segment set; For the set of over-threshold track segments, a trajectory optimization algorithm is used to generate local correction instructions and obtain a correction instruction sequence; By modifying the instruction sequence and combining it with the onboard feedback algorithm, the target track sequence is updated to obtain the adjusted control sequence; The control parameters are extracted from the adjustment control sequence to generate a real-time adjusted control sequence.

[0013] The above-mentioned intelligent UAV flight control method based on 4D trajectory, for the control sequence adjusted in real time, uses a satellite-based data synchronization algorithm to upload the actual trajectory data to the satellite-based computing center, and updates the mission database through comparative analysis with the initial 4D trajectory set to obtain a trajectory data set for subsequent mission optimization, including: Through the satellite-based data synchronization algorithm, the real-time adjusted control sequence and actual track data are obtained from the navigation terminal to generate the initial data set; If the integrity of the initial data set meets the preset threshold, the data set is uploaded to the computing center through a secure transmission protocol to obtain an uploaded data record; Using comparative analysis methods, the uploaded data records and the initial 4D track set are obtained from the computing center, the track deviation value is calculated, and the deviation analysis results are generated; According to the deviation analysis results, historical track data matching the deviation value is obtained from the mission database, and the K-means clustering algorithm is used for classification to determine the optimized track candidate set; If the number of optimized track candidate sets is greater than the preset threshold, the candidate sets are fused by weighted averaging method to generate an optimized track data set; Through the mission database update interface, the optimized track data set is stored in the mission database to generate updated database records; According to the updated database records, the optimized track data set is obtained from the computing center and transmitted to the navigation terminal to complete the track optimization process.

[0014] Compared with the existing technology, the present invention has the following advantages: the present invention addresses the four-dimensional trajectory planning problem of massive flight missions, obtains initial trajectory data from a satellite-based computing center, uses a distributed parallel computing framework to generate an initial four-dimensional trajectory set, obtains real-time airspace data through drone-mounted sensors, uses a deep reinforcement learning algorithm to optimize the trajectory, and generates a candidate trajectory set; combines satellite-based meteorological data, uses a prediction model and a weighted fusion algorithm to generate a corrected trajectory that adapts to meteorological dynamics; uses a dynamic programming algorithm to optimize the trajectory adjustment speed and generates millisecond-level adjustment instructions; finally, transmits the instructions to the drone for execution through a satellite-based communication link, and provides real-time feedback on the adjustments; realizes precise trajectory planning and dynamic optimization of large-scale drone swarms, and improves flight efficiency and safety in complex airspace.

[0015] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a flowchart of the method of the present invention. DETAILED DESCRIPTION

[0017] like Figure 1 As shown, the intelligent UAV flight control method based on 4D trajectory of the present invention includes the following steps: Obtain the initial 4D track data for task assignments and generate 4D track datasets for massive daily flight missions; including: The initial 4D track data of daily flight missions is obtained from the satellite-based computing center and stored as a structured data set containing time, position, and velocity attributes to obtain the initial data set; A distributed computing framework is used to assign tasks to the initial data set. If the data volume exceeds a preset threshold, the task is decomposed into multiple subtasks and assigned to computing nodes to obtain a subtask data set. The subtask data set is processed by the space-time grid partitioning algorithm to generate gridded track points of time, position and speed, and obtain a gridded track point set; For the gridded track point set, the k-means clustering algorithm is used to classify the track points. If the time interval of the track points is less than the preset threshold, they are classified into the same track sequence to obtain a preliminary track sequence set. Based on the preliminary track sequence set, the time, position and speed sequences are completed using the linear interpolation method. If there are missing points in the sequence, they are filled in by interpolation based on the previous and next point data to obtain a complete track sequence set. Through the distributed computing framework, the complete track sequence set is verified in parallel. If the deviation between the time, position and speed data of the sequence and the initial data is less than the preset threshold, it is determined to be a valid sequence and the final track point sequence set is obtained. For the final track point sequence set, a 4D track data structure containing time, position and velocity is generated and stored in the satellite-based computing center to obtain a standardized 4D track data set.

[0018] Specifically, when obtaining the initial 4D trajectory data for task assignments from the satellite-based computing center, the system receives a task list containing information such as flight number, departure time, and destination via a satellite communication link. For example, a task list received on a particular day might contain 1,000 flights, with each flight's task data size being 1KB. Using a distributed parallel computing framework, the system distributes the task data to 10 computing nodes, with each node processing data for 100 flights. Using a pre-defined spatiotemporal gridding algorithm, the airspace is divided into 1,000 grid cells, each measuring 10 km x 10 km x 1,000 meters, with a time interval of 1 minute.

[0019] To handle the massive daily flight missions, the system calculates the flight's initial four-dimensional trajectory within each grid cell. For example, if a flight's position at time t=0 is (x=100 km, y=200 km, z=5000 m) and its speed is 800 km / h, an interpolation algorithm is used to calculate its position at t=1 minute as (x=101.33 km, y=201.33 km, z=5000 m). This generates an initial four-dimensional trajectory set, ultimately resulting in a sequence of trackpoints containing time, position, and speed. For example, the trackpoint sequence for a flight might be [(t=0, x=100, y=200, z=5000, v=800), (t=1, x=101.33, y=201.33, z=5000, v=800), …]. Through parallel computing and a gridding algorithm, the system efficiently handles the massive daily flight missions, ensuring the accuracy and real-time nature of the track data.

[0020] For the 4D trajectory data set, real-time airspace data is obtained from the drone's onboard sensors, and the data is compressed to obtain a compressed real-time airspace data packet; including: For the 4D trajectory dataset, real-time airspace data is obtained from the drone's onboard sensors, and the environmental information is collected using a preset sampling frequency to obtain sensor data; For sensor data, a data compression algorithm is used to pre-process the environmental information. If the data redundancy exceeds the preset threshold, lossless compression is performed to obtain a compressed data packet. The compressed data packets are transmitted to the computing center via the satellite-based communication link. If the transmission bandwidth is lower than the preset threshold, the compression ratio is adjusted to obtain a stable data transmission flow. The data transmission stream is received at the computing center, and the data packet is restored using a decompression algorithm to obtain restored spatial domain data; For the restored spatial data, a clustering algorithm is used to extract features from the spatial data. If the dimension of the feature vector exceeds the preset threshold, dimensionality reduction is performed to obtain feature data. According to the feature data, the airspace data is classified using the preset matching rules to obtain the classified airspace data; For the classified airspace data, real-time airspace data packets are generated and stored in the preset database of the satellite-based computing center.

[0021] Specifically, within the initial four-dimensional track set, the drone's onboard sensors collect airspace data in real time using lidar and infrared cameras. The lidar scans the surrounding environment 1,000 times per second, generating point cloud data, while the infrared camera captures heat source information at 30 frames per second. To reduce the amount of data transmitted, a wavelet transform-based data compression algorithm is used to preprocess the sensor data. The wavelet transform decomposes the raw data into subbands of different frequencies. Thresholding processing removes high-frequency noise while retaining low-frequency information, achieving a compression ratio of 10:1. The compressed data is transmitted to the computing center via a satellite-based communication link, which uses LDPC coding technology to ensure reliable data transmission. Upon receiving the compressed data packets, the computing center uses the JPEG2000 decoding algorithm to decompress them and restore the original data.

[0022] Based on the compressed real-time airspace data packet, the track point sequence in the 4D track data set is optimized to generate an adjusted candidate track set, and a candidate track point sequence containing the optimized path is obtained; including: By decompressing the real-time airspace data packet, the initial data set containing the track point sequence and environmental information is obtained to obtain structured airspace data; Using a deep reinforcement learning algorithm, we perform preliminary optimization of the track point sequence based on structured airspace data at the satellite-based computing center to obtain a preliminarily optimized track point sequence. Based on the dynamic obstacle data acquired in real time, if a conflict between an obstacle and the initially optimized track point sequence is detected, the track point coordinates are adjusted to obtain a conflict-avoiding track point sequence. By integrating weather change data, if the weather conditions exceed the preset threshold, the conflict avoidance trackpoint sequence is adjusted twice to obtain a trackpoint sequence that adapts to weather changes; A genetic algorithm is used to perform global optimization on the trackpoint sequence that adapts to weather changes, generating a set of candidate trackpoint sequences containing multiple optimized paths. By comparing the cost function values of each path in the candidate track point sequence set, the track point sequence with the lowest cost function value is determined to obtain the final optimized track point sequence; For the final optimized track point sequence, a track point sequence data packet that meets the airspace constraints is generated, and a candidate track point sequence containing the optimized path is obtained and output to the satellite-based computing center.

[0023] Specifically, after receiving compressed real-time airspace data packets at the satellite-based computing center, they are first decoded using a decompression module based on the LZMA algorithm. The decompressed data contains aircraft position coordinates (e.g., longitude 112.5°, latitude 28.2°, altitude 9,000 meters), updated at 10Hz per second, dynamic obstacle information (a swarm of drones moving at 12 meters per second), and meteorological data (a turbulent area with a wind speed of 15 knots in the southwest). A trajectory optimization model is constructed using the Proximal Policy Optimization (PPO) algorithm, a deep reinforcement learning framework. The model's input layer is designed as an LSTM network containing a 128-dimensional feature vector. The sequence of track points within each 5-second time window (e.g., 50 consecutive sampling points) and dynamic environmental parameters together constitute the state space. During the strategy network output phase, the model generates three candidate track adjustment plans for sudden meteorological changes (such as a newly emerged thunderstorm zone with a radius of 5 kilometers): Plan A uses Bezier curve smoothing to insert seven control points between the original track points (112.5°, 28.2°) and (112.8°, 28.5°), with a horizontal offset of no more than 300 meters; Plan B replans based on the A-star algorithm, taking into account the predicted position of dynamic obstacles (moving to 112.6°, 28.3° within the next 30 seconds), and generates an avoidance path containing 12 track points; Plan C combines the results of meteorological data fusion and uses the gradient descent method to calculate the optimal energy consumption path, adjusting the flight altitude from 9,000 meters to 8,700 meters to avoid strong winds. The final set of candidate tracks was ranked using a Q-value evaluation module. Option B, with the highest overall score (98.7% obstacle avoidance success rate and only a 2.1% increase in energy consumption), was designated as the preferred option. The latitude and longitude coordinates of all candidate tracks were stored in the onboard database with an accuracy of 0.01°.

[0024] From the candidate track point sequence, a spatiotemporal synchronization constraint algorithm is used to verify the timestamp and position deviation of each track point to obtain an optimized track sequence that satisfies spatiotemporal synchronization; this includes: Obtain initial track data from the candidate track point sequence, use a preset millisecond-level time window to determine the timestamp and position coordinates of each track point, and obtain the initial track point set; For the initial track point set, a spatiotemporal synchronization constraint algorithm is used to check the timestamp and position deviation of each track point. If the deviation exceeds the preset threshold, it is marked as an abnormal track point to obtain an abnormal point set; Obtain the marked track points from the outlier point set, use the local interpolation algorithm to recalculate the coordinates and timestamps of the outlier track points, and obtain the corrected local track segment; The corrected local track segment is used to replace the corresponding abnormal track points in the initial track point set to generate an updated track point set; For the updated track point set, a smoothing filtering algorithm is used to eliminate the sudden changes between track points and obtain a smooth track sequence; From the smoothed track sequence, the timestamps and position deviations of adjacent track points are checked. If the deviation exceeds the preset threshold, the local interpolation calculation is repeated to obtain the final optimized track sequence. Through the final optimized track sequence, track data that meets the time and space synchronization constraints are generated, and the optimized track sequence that meets the time and space synchronization is obtained.

[0025] Specifically, during the processing of the candidate track point sequence, the timestamp of each track point is first verified using a preset millisecond time window (e.g., 100 milliseconds). Assume that the timestamp of the current track point is T1 and the position coordinates are (X1, Y1), and the timestamp of the next track point is T2 and the position coordinates are (X2, Y2). The time difference ΔT = T2 - T1 is calculated. If ΔT exceeds 100 milliseconds, it is determined to be a time deviation. Next, the position deviation ΔD = √((X2 - X1)² + (Y2 - Y1)²) is calculated. If ΔD exceeds a preset threshold (e.g., 10 meters), it is determined to be a position deviation. When the time deviation or position deviation exceeds the threshold, the spatiotemporal synchronization constraint algorithm is used to recalculate the local track segment.

[0026] For example, an interpolation algorithm is used to insert new track points between T1 and T2, ensuring uniform time intervals and smooth position changes. Assume that N points are inserted between T1 and T2, with a time interval of ΔT / N between each point. Position coordinates are calculated using linear interpolation. Ultimately, an optimized track sequence is obtained that meets spatiotemporal synchronization, ensuring that the timestamp and position deviation of each track point are within a preset range, thereby improving the accuracy and reliability of the track data.

[0027] To optimize the track sequence, real-time weather update data is obtained from the satellite-based meteorological database. The weather change trend is combined with the track sequence to obtain a corrected track sequence that adapts to the weather dynamics. This includes: Obtain real-time meteorological data from satellite-based meteorological databases, parse the data format using standard protocols, and obtain structured meteorological data sets; The structured meteorological dataset is analyzed through a prediction model, and the long short-term memory network (LSTM) is used to calculate the time series features to obtain the meteorological change trend; If the fluctuation amplitude of the meteorological change trend exceeds the preset threshold, the sliding window method is used to smooth the change trend to obtain a smooth change trend; If the fluctuation range of the meteorological change trend is lower than the threshold, the change trend is directly output; According to the smooth change trend, a weighted fusion algorithm is used to calculate the matching weight of the track sequence and the change trend, and the weighted track adjustment parameters are obtained; By adjusting the weighted track parameters and using the geometric transformation method to correct the coordinates of the original track sequence, a preliminary corrected track sequence is generated. For the preliminary corrected track sequence, a dynamic programming algorithm is used to optimize the path smoothness between track points to obtain the final corrected track sequence; If the deviation between the final corrected track sequence and the real-time meteorological data exceeds the preset threshold, it returns to the weighted fusion step, adjusts the weight parameters, and regenerates the corrected track sequence; If the deviation is within the threshold, the final corrected track sequence is output to obtain a corrected track sequence that adapts to meteorological dynamics.

[0028] Specifically, during the track sequence optimization process, real-time meteorological data is first obtained from a satellite-based meteorological database. For example, NOAA's GOES-16 satellite collects parameters such as atmospheric pressure (1013 hPa ± 5 hPa), wind speed (12 m / s ± 3 m / s), and turbulence index (0.15 ± 0.05) within the current latitude and longitude range of 30°N to 40°N. The data is updated every five minutes. An LSTM prediction model is used to analyze meteorological trends. The model has 128 neurons in the input layer, a tanh activation function in the hidden layer, and a mean square error (MSE) loss function of 0.02. The model predicts that the wind speed will increase to 15 m / s and the turbulence index will rise to 0.22 within the next 30 minutes. The prediction results are integrated with the original track using a weighted fusion algorithm. The weather impact weight α is set to 0.7 (based on a weather sensitivity analysis), and the track stability weight β is set to 0.3. The Euclidean distance formula is used to calculate the corrected offset Δd = α × √(Δx² + Δy²) + β × |Δθ|, where Δx = 1.2 km and Δy = 0.8 km represent the horizontal displacement caused by weather, and Δθ = 5° represents the heading deviation. The final corrected track is generated using cubic spline interpolation to ensure track smoothness (radius of curvature ≥ 8 km). The corrected waypoint sequence (for example, the original longitude of waypoint WPT01 is adjusted from 118.75°E to 118.73°E) is then written to the flight management system. Throughout this process, a Kalman filter (process noise Q = 0.1, observation noise R = 0.05) continuously estimates the state of the fusion results to ensure that the output track meets ICAO RNP-1 navigation performance requirements.

[0029] Based on the corrected track sequence, the trajectory adjustment speed of the track segment is optimized using a dynamic programming algorithm to obtain the final track sequence including the adjustment speed; including: Obtain the corrected track sequence from the satellite-based computing center, and use the segment analysis method to extract the coordinates and velocity information of the track segments to obtain the track segment data set; Based on the trajectory segment data set, a dynamic programming algorithm is used to calculate the adjusted speed of each trajectory segment, and the speed adjustment parameters are generated according to the dynamic constraints of complex airspace. By adjusting the speed parameters and combining the dynamic characteristics of the airspace, the track segment coordinates are fine-tuned using the geometric transformation method to obtain a preliminary adjusted track sequence. If the deviation between the initial adjustment trajectory sequence and the obstacle avoidance requirement exceeds the preset threshold, the system returns to the dynamic programming algorithm step, adjusts the speed parameters, and regenerates the initial adjustment trajectory sequence; If the deviation is within the threshold, the preliminary adjusted track sequence is output; According to the preliminary adjusted track sequence, the interpolation smoothing method is used to optimize the speed transition between track segments and generate a smoothed adjusted track sequence; Real-time airspace dynamic characteristics are obtained from the satellite-based computing center. If the matching degree between the smoothed adjusted track sequence and the airspace dynamic characteristics is lower than the preset threshold, the track segment coordinates are adjusted using a weighted correction method to obtain the final adjusted track sequence. If the matching degree is within the threshold, the final adjusted track sequence is directly output; Through the final adjustment of the track sequence, the instruction encoding method is used to generate a millisecond-level adjustment instruction set to obtain the final track sequence including speed adjustment.

[0030] Specifically, a dynamic programming algorithm is used at the satellite-based computing center to optimize the trajectory adjustment speed of each track segment based on the corrected track sequence. First, the satellite-based ADS-B system acquires real-time airspace obstacle information. For example, a cumulonimbus cloud cluster with an altitude of 10,000 meters, a horizontal range of 5 kilometers, and a moving speed of 10 meters per second is detected within the latitude and longitude range of 25°N to 35°N. Using a dynamic programming algorithm, with the track segment length as a constraint, the objective function is to minimize the weighted sum of adjustment time and energy consumption, with a time weight of γ = 0.6 and an energy weight of δ = 0.4. During the algorithm iteration, the state transition step size is set to 0.1 seconds, and the optimal adjustment speed is calculated for each state point. For example, at waypoint WPT02, the initial speed is 250 meters per second, which is adjusted to 240 meters per second after optimization to maintain a safe distance from the obstacle (≥ 3 kilometers). To address the multi-obstacle avoidance requirements in complex airspace, a multi-objective optimization algorithm is used to generate a set of millisecond-level adjustment instructions.

[0031] For example, at waypoint WPT03, two obstacles were detected, located 2 kilometers to the left and 3 kilometers to the right of the track. Calculation generated adjustment instructions: within 0.5 seconds, the heading angle was deflected 3° to the left and the speed was reduced to 230 meters per second, ensuring a safe distance of 2.5 kilometers from the left obstacle and 3.2 kilometers from the right obstacle. Finally, the track sequence, including the adjusted speed, was written into the flight management system. Bezier curve interpolation was used to ensure track smoothness (radius of curvature ≥ 10 kilometers). This was then transmitted in real time to the airborne system via a satellite-based communication link, ensuring the immediacy and accuracy of the track adjustments.

[0032] From the final trajectory sequence, a data distribution algorithm is used to transmit the adjustment instruction set to the UAV onboard controller via a satellite-based communication link, resulting in a flight control sequence that can be directly executed by the UAV; including: From the final track sequence, the data distribution algorithm is used to decompose and generate the adjustment instruction set and determine the initial structure of the instruction set; Transmitting the adjusted instruction set to the UAV onboard controller via a satellite-based communication link to obtain the integrity status of the transmitted instruction set; If the integrity status of the transmitted instruction set meets a preset threshold, the instruction set is parsed in the onboard controller to obtain the parsed instruction data; If not, the transmission is re-transmitted via the satellite-based communication link; In view of the data processing capability of the airborne equipment, a lightweight algorithm is used to compress the parsed instruction data to generate lightweight execution instructions; Extract control parameters from lightweight execution instructions to generate flight control sequences that can be directly executed by the UAV; Verify the compatibility of the flight control sequence through the onboard controller to determine whether the sequence meets the UAV execution control requirements; If the flight control sequence meets the execution control requirements, the sequence is loaded into the UAV execution module to obtain a flight control sequence that can be directly executed by the UAV.

[0033] Specifically, in the final trajectory sequence, the adjustment instruction set is first transmitted to the drone's onboard controller via a data distribution algorithm. The data distribution algorithm uses a priority-based multi-path transmission strategy to ensure that critical instructions are transmitted first.

[0034] For example, when the drone is flying at an altitude of 500 meters, the algorithm selects the optimal path based on current network conditions and transmits adjustment instructions to the onboard controller at a rate of 1000 bits per millisecond. After receiving the instructions, the onboard controller generates lightweight execution instructions based on the data processing capabilities of the onboard equipment.

[0035] Specifically, the controller adopts a finite state machine-based instruction generation algorithm to simplify the complex adjustment instruction set into a series of state transitions.

[0036] For example, when a drone needs to adjust its heading, the controller converts complex heading adjustment instructions into simple state sequences, such as "turn left 30 degrees, maintain altitude 500 meters," and transmits these state sequences to the flight control system at a rate of 500 bits per millisecond. After receiving the lightweight execution instructions, the flight control system generates a flight control sequence that the drone can directly execute.

[0037] For example, when a drone is instructed to "turn left 30 degrees and maintain an altitude of 500 meters," the flight control system calculates specific control parameters, such as servo angle and throttle level, based on the current flight state. These parameters are then transmitted to the actuator at a rate of 200 bits per millisecond. The actuator then adjusts the drone's flight state in real time based on the received control parameters, ensuring the drone follows the predetermined flight path.

[0038] Through the above steps, the UAV can perform flight missions efficiently and accurately, ensuring flight safety and mission completion.

[0039] According to the flight control sequence, the actual trajectory data after execution is obtained from the drone sensor. By analyzing the deviation from the final trajectory sequence, the real-time adjusted control sequence is obtained; including: Acquire actual track data from the drone sensor, use data preprocessing technology to remove noise and outliers, and obtain the first track data set; According to the first track data set and the target track sequence, the Euclidean distance algorithm is used to calculate the deviation value between the two to obtain a deviation data set; If the deviation value in the deviation data set exceeds the preset deviation threshold, the deviation analysis module is used to determine the track segment with the deviation exceeding the threshold, and obtain the exceeding threshold track segment set; For the set of over-threshold track segments, a trajectory optimization algorithm is used to generate local correction instructions and obtain a correction instruction sequence; By modifying the instruction sequence and combining it with the onboard feedback algorithm, the target track sequence is updated to obtain the adjusted control sequence; The control parameters are extracted from the adjustment control sequence to generate a real-time adjusted control sequence.

[0040] Specifically, during drone flight control, the onboard real-time feedback algorithm uses sensors to acquire actual track data, such as the current latitude and longitude coordinates (116.3975, 39.9087) using a GPS module. This data is then compared with the preset final track sequence. Assuming the preset track is (116.3970, 39.9085), the Euclidean distance deviation is calculated. If the deviation exceeds a preset threshold of 10 meters, a local correction command is triggered. The correction algorithm uses a PID controller with a proportional coefficient Kp of 0.8, an integral coefficient Ki of 0.2, and a differential coefficient Kd of 0.1. The correction is calculated based on the deviation.

[0041] For example, if the current deviation is 15 meters, the PID controller outputs a correction of 12 meters, generating a new control sequence to adjust the drone's flight direction. This adjusted control sequence is executed by the flight control system, which updates the track data in real time to ensure the drone follows the intended path. The entire process is processed in real time by the onboard computer, ensuring flight accuracy and stability.

[0042] For the control sequence adjusted in real time, the satellite-based data synchronization algorithm is used to upload the actual track data to the satellite-based computing center. By comparing and analyzing it with the initial 4D track set, the mission database is updated to obtain the track data set for subsequent mission optimization. This includes: Through the satellite-based data synchronization algorithm, the real-time adjusted control sequence and actual track data are obtained from the navigation terminal to generate the initial data set; If the integrity of the initial data set meets the preset threshold, the data set is uploaded to the computing center through a secure transmission protocol to obtain an uploaded data record; Using comparative analysis methods, the uploaded data records and the initial 4D track set are obtained from the computing center, the track deviation value is calculated, and the deviation analysis results are generated; According to the deviation analysis results, historical track data matching the deviation value is obtained from the mission database, and the K-means clustering algorithm is used for classification to determine the optimized track candidate set; If the number of optimized track candidate sets is greater than the preset threshold, the candidate sets are fused by weighted averaging method to generate an optimized track data set; Through the mission database update interface, the optimized track data set is stored in the mission database to generate updated database records; According to the updated database records, the optimized track data set is obtained from the computing center and transmitted to the navigation terminal to complete the track optimization process.

[0043] Specifically, during the real-time control sequence adjustment phase, a Kalman filter-based prediction algorithm is used to dynamically correct the aircraft's trajectory. For example, if the aircraft deviates from the preset track by more than 300 meters, the system automatically generates control commands including the new altitude (e.g., from 9,000 meters to 9,500 meters) and speed (from 850 km / h to 880 km / h). Satellite-based data is synchronized via the BeiDou-3 satellite's L-band channel, with ADS-B packets uploaded twice per second. These packets contain latitude and longitude (e.g., 116.4 degrees east longitude, 39.9 degrees north latitude), altitude (accuracy ±0.5 meters), and UTC timestamps (synchronized to the nanosecond level). CRC-32 checksums and TCP retransmission mechanisms ensure transmission integrity. Upon receiving the data, the computing center compares the actual trajectory with the initial four-dimensional trajectory (consisting of longitude, latitude, altitude, and time) using the Dynamic Time Warping (DTW) algorithm. Track updates are triggered when the time deviation of a track point exceeds 15 seconds or the spatial distance difference exceeds 500 meters. The mission database is updated using a Redis-based stream processing architecture. The ZADD command stores new track data (such as waypoint serial number WPT_20231125_0042) with timestamps as scores in an ordered set, while a Bloom filter is used to quickly remove duplicates. The resulting optimized track dataset uses a cubic spline interpolation algorithm to complete missing points, forming a continuous curve containing 200 track points for use by the trajectory prediction module. Each track point includes environmental parameters such as the QNH pressure correction value (e.g., 1013.25 hPa) and wind field data (wind direction 270 degrees, wind speed 12 m / s).

[0044] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any way. Any simple modification, change and equivalent structural change made to the above embodiment based on the technical essence of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. An intelligent UAV flight control method based on 4D trajectory, characterized in that: The method comprises the following steps: Obtain the initial 4D track data of the task assignment and generate 4D track data sets for the massive daily flight tasks; For the 4D trajectory data set, real-time airspace data is obtained from the drone's onboard sensors, and the data is compressed to obtain a compressed real-time airspace data packet; Based on the compressed real-time airspace data packet, the track point sequence in the 4D track data set is optimized to generate an adjusted candidate track set, thereby obtaining a candidate track point sequence containing the optimized path. From the candidate track point sequence, the time-space synchronization constraint algorithm is used to verify the timestamp and position deviation of each track point to obtain the optimized track sequence that meets the time-space synchronization. To optimize the track sequence, real-time weather update data is obtained from the satellite-based meteorological database, and the weather change trend is combined with the track sequence to obtain a corrected track sequence that adapts to the weather dynamics; According to the corrected track sequence, the trajectory adjustment speed of the track segment is optimized using a dynamic programming algorithm to obtain the final track sequence including the adjustment speed; From the final trajectory sequence, a data distribution algorithm is used to transmit the adjustment instruction set to the UAV onboard controller via a satellite-based communication link, obtaining a flight control sequence that can be directly executed by the UAV; According to the flight control sequence, the actual trajectory data after execution is obtained from the drone sensor, and the real-time adjusted control sequence is obtained by analyzing the deviation from the final trajectory sequence; For the control sequence adjusted in real time, the satellite-based data synchronization algorithm is used to upload the actual track data to the satellite-based computing center. By comparing and analyzing it with the initial 4D track set, the mission database is updated to obtain the track data set for subsequent mission optimization.

2. The intelligent UAV flight control method based on 4D trajectory according to claim 1, characterized in that: The acquisition of the 4D track initial data of the task assignment generates a 4D track data set for the massive daily flight tasks, including: The initial 4D track data of daily flight missions is obtained from the satellite-based computing center and stored as a structured data set containing time, position, and velocity attributes to obtain the initial data set; A distributed computing framework is used to assign tasks to the initial data set. If the data volume exceeds a preset threshold, the task is decomposed into multiple subtasks and assigned to computing nodes to obtain a subtask data set. The subtask data set is processed by the space-time grid partitioning algorithm to generate gridded track points of time, position and speed, and obtain a gridded track point set; For the gridded track point set, the k-means clustering algorithm is used to classify the track points. If the time interval of the track points is less than the preset threshold, they are classified into the same track sequence to obtain a preliminary track sequence set. Based on the preliminary track sequence set, the time, position and speed sequences are completed using the linear interpolation method. If there are missing points in the sequence, they are filled in by interpolation based on the previous and next point data to obtain a complete track sequence set. Through the distributed computing framework, the complete track sequence set is verified in parallel. If the deviation between the time, position and speed data of the sequence and the initial data is less than the preset threshold, it is determined to be a valid sequence and the final track point sequence set is obtained. For the final track point sequence set, a 4D track data structure containing time, position and velocity is generated and stored in the satellite-based computing center to obtain a standardized 4D track data set.

3. The intelligent UAV flight control method based on 4D trajectory according to claim 1, characterized in that: The method of acquiring real-time airspace data from the UAV's onboard sensor for the 4D track data set and compressing the data to obtain a compressed real-time airspace data packet includes: For the 4D trajectory dataset, real-time airspace data is obtained from the drone's onboard sensors, and the environmental information is collected using a preset sampling frequency to obtain sensor data; For sensor data, a data compression algorithm is used to pre-process the environmental information. If the data redundancy exceeds the preset threshold, lossless compression is performed to obtain a compressed data packet. The compressed data packets are transmitted to the computing center via the satellite-based communication link. If the transmission bandwidth is lower than the preset threshold, the compression ratio is adjusted to obtain a stable data transmission flow. The data transmission stream is received at the computing center, and the data packet is restored using a decompression algorithm to obtain restored spatial domain data; For the restored spatial data, a clustering algorithm is used to extract features from the spatial data. If the dimension of the feature vector exceeds the preset threshold, dimensionality reduction is performed to obtain feature data. According to the feature data, the airspace data is classified using the preset matching rules to obtain the classified airspace data; For the classified airspace data, real-time airspace data packets are generated and stored in the preset database of the satellite-based computing center.

4. The intelligent UAV flight control method based on 4D trajectory according to claim 1, characterized in that: The method optimizes the track point sequence in the 4D track data set according to the compressed real-time airspace data packet to generate an adjusted candidate track set, thereby obtaining a candidate track point sequence containing the optimized path, including: By decompressing the real-time airspace data packet, the initial data set containing the track point sequence and environmental information is obtained to obtain structured airspace data; Using a deep reinforcement learning algorithm, we perform preliminary optimization of the track point sequence based on structured airspace data at the satellite-based computing center to obtain a preliminarily optimized track point sequence. Based on the dynamic obstacle data acquired in real time, if a conflict between an obstacle and the initially optimized track point sequence is detected, the track point coordinates are adjusted to obtain a conflict-avoiding track point sequence. By integrating weather change data, if the weather conditions exceed the preset threshold, the conflict avoidance trackpoint sequence is adjusted twice to obtain a trackpoint sequence that adapts to weather changes; A genetic algorithm is used to perform global optimization on the trackpoint sequence that adapts to weather changes, generating a set of candidate trackpoint sequences containing multiple optimized paths. By comparing the cost function values of each path in the candidate track point sequence set, the track point sequence with the lowest cost function value is determined to obtain the final optimized track point sequence; For the final optimized track point sequence, a track point sequence data packet that meets the airspace constraints is generated, and a candidate track point sequence containing the optimized path is obtained and output to the satellite-based computing center.

5. The intelligent UAV flight control method based on 4D trajectory according to claim 1, characterized in that: The method uses a spatiotemporal synchronization constraint algorithm to verify the timestamp and position deviation of each track point from the candidate track point sequence to obtain an optimized track sequence that satisfies spatiotemporal synchronization, including: Obtain initial track data from the candidate track point sequence, use a preset millisecond-level time window to determine the timestamp and position coordinates of each track point, and obtain the initial track point set; For the initial track point set, a spatiotemporal synchronization constraint algorithm is used to check the timestamp and position deviation of each track point. If the deviation exceeds the preset threshold, it is marked as an abnormal track point to obtain an abnormal point set; Obtain the marked track points from the outlier point set, use the local interpolation algorithm to recalculate the coordinates and timestamps of the outlier track points, and obtain the corrected local track segment; The corrected local track segment is used to replace the corresponding abnormal track points in the initial track point set to generate an updated track point set; For the updated track point set, a smoothing filtering algorithm is used to eliminate the sudden changes between track points and obtain a smooth track sequence; From the smoothed track sequence, the timestamps and position deviations of adjacent track points are checked. If the deviation exceeds the preset threshold, the local interpolation calculation is repeated to obtain the final optimized track sequence. Through the final optimized track sequence, track data that meets the time and space synchronization constraints are generated, and the optimized track sequence that meets the time and space synchronization is obtained.

6. The intelligent UAV flight control method based on 4D trajectory according to claim 1, characterized in that: The method for optimizing the track sequence is to obtain real-time weather update data from a satellite-based weather database, combine the weather change trend with the track sequence, and obtain a corrected track sequence that adapts to the weather dynamics, including: Obtain real-time meteorological data from satellite-based meteorological databases, parse the data format using standard protocols, and obtain structured meteorological data sets; The structured meteorological data set is analyzed through the prediction model, and the time series characteristics are calculated using the long short-term memory network to obtain the meteorological change trend; If the fluctuation amplitude of the meteorological change trend exceeds the preset threshold, the sliding window method is used to smooth the change trend to obtain a smooth change trend; If the fluctuation range of the meteorological change trend is lower than the threshold, the change trend is directly output; According to the smooth change trend, a weighted fusion algorithm is used to calculate the matching weight of the track sequence and the change trend, and the weighted track adjustment parameters are obtained; By adjusting the weighted track parameters and using the geometric transformation method to correct the coordinates of the original track sequence, a preliminary corrected track sequence is generated. For the preliminary corrected track sequence, a dynamic programming algorithm is used to optimize the path smoothness between track points to obtain the final corrected track sequence; If the deviation between the final corrected track sequence and the real-time meteorological data exceeds the preset threshold, it returns to the weighted fusion step, adjusts the weight parameters, and regenerates the corrected track sequence; If the deviation is within the threshold, the final corrected track sequence is output to obtain a corrected track sequence that adapts to meteorological dynamics.

7. The intelligent UAV flight control method based on 4D trajectory according to claim 1, characterized in that: The method of optimizing the trajectory adjustment speed of the track segment using a dynamic programming algorithm based on the modified track sequence to obtain a final track sequence including the adjustment speed includes: Obtain the corrected track sequence from the satellite-based computing center, and use the segment analysis method to extract the coordinates and velocity information of the track segments to obtain the track segment data set; Based on the trajectory segment data set, a dynamic programming algorithm is used to calculate the adjusted speed of each trajectory segment, and the speed adjustment parameters are generated according to the dynamic constraints of complex airspace. By adjusting the speed parameters and combining the dynamic characteristics of the airspace, the track segment coordinates are fine-tuned using the geometric transformation method to obtain a preliminary adjusted track sequence. If the deviation between the initial adjustment trajectory sequence and the obstacle avoidance requirement exceeds the preset threshold, the system returns to the dynamic programming algorithm step, adjusts the speed parameters, and regenerates the initial adjustment trajectory sequence; If the deviation is within the threshold, the preliminary adjusted track sequence is output; According to the preliminary adjusted track sequence, the interpolation smoothing method is used to optimize the speed transition between track segments and generate a smoothed adjusted track sequence; Real-time airspace dynamic characteristics are obtained from the satellite-based computing center. If the matching degree between the smoothed adjusted track sequence and the airspace dynamic characteristics is lower than the preset threshold, the track segment coordinates are adjusted using a weighted correction method to obtain the final adjusted track sequence. If the matching degree is within the threshold, the final adjusted track sequence is directly output; Through the final adjustment of the track sequence, the instruction encoding method is used to generate a millisecond-level adjustment instruction set to obtain the final track sequence including speed adjustment.

8. The intelligent UAV flight control method based on 4D trajectory according to claim 1, characterized in that: The final track sequence is obtained by using a data distribution algorithm to transmit the adjustment instruction set to the UAV onboard controller via a satellite-based communication link, thereby obtaining a flight control sequence that can be directly executed by the UAV, including: From the final track sequence, the data distribution algorithm is used to decompose and generate the adjustment instruction set and determine the initial structure of the instruction set; Transmitting the adjusted instruction set to the UAV onboard controller via a satellite-based communication link to obtain the integrity status of the transmitted instruction set; If the integrity status of the transmitted instruction set meets a preset threshold, the instruction set is parsed in the onboard controller to obtain the parsed instruction data; If not, the transmission is re-transmitted via the satellite-based communication link; In view of the data processing capability of the airborne equipment, a lightweight algorithm is used to compress the parsed instruction data to generate lightweight execution instructions; Extract control parameters from lightweight execution instructions to generate flight control sequences that can be directly executed by the UAV; Verify the compatibility of the flight control sequence through the onboard controller to determine whether the sequence meets the UAV execution control requirements; If the flight control sequence meets the execution control requirements, the sequence is loaded into the UAV execution module to obtain a flight control sequence that can be directly executed by the UAV.

9. The intelligent UAV flight control method based on 4D trajectory according to claim 1, characterized in that: The actual flight path data after execution is obtained from the drone sensor according to the flight control sequence, and the real-time adjusted control sequence is obtained by analyzing the deviation from the final flight path sequence, including: Acquire actual track data from the drone sensor, use data preprocessing technology to remove noise and outliers, and obtain the first track data set; According to the first track data set and the target track sequence, the Euclidean distance algorithm is used to calculate the deviation value between the two to obtain a deviation data set; If the deviation value in the deviation data set exceeds the preset deviation threshold, the deviation analysis module is used to determine the track segment with the deviation exceeding the threshold, and obtain the exceeding threshold track segment set; For the set of over-threshold track segments, a trajectory optimization algorithm is used to generate local correction instructions and obtain a correction instruction sequence; By modifying the instruction sequence and combining it with the onboard feedback algorithm, the target track sequence is updated to obtain the adjusted control sequence; The control parameters are extracted from the adjustment control sequence to generate a real-time adjusted control sequence.

10. The intelligent UAV flight control method based on 4D trajectory according to claim 1, characterized in that: For the control sequence adjusted in real time, a satellite-based data synchronization algorithm is used to upload the actual track data to the satellite-based computing center. By comparing and analyzing the actual track data with the initial 4D track data set, the mission database is updated to obtain a track data set for subsequent mission optimization, including: Through the satellite-based data synchronization algorithm, the real-time adjusted control sequence and actual track data are obtained from the navigation terminal to generate the initial data set; If the integrity of the initial data set meets the preset threshold, the data set is uploaded to the computing center through a secure transmission protocol to obtain an uploaded data record; Using comparative analysis methods, the uploaded data records and the initial 4D track set are obtained from the computing center, the track deviation value is calculated, and the deviation analysis results are generated; According to the deviation analysis results, historical track data matching the deviation value is obtained from the mission database, and the K-means clustering algorithm is used for classification to determine the optimized track candidate set; If the number of optimized track candidate sets is greater than the preset threshold, the candidate sets are fused by weighted averaging method to generate an optimized track data set; Through the mission database update interface, the optimized track data set is stored in the mission database to generate updated database records; According to the updated database records, the optimized track data set is obtained from the computing center and transmitted to the navigation terminal to complete the track optimization process.

Images