Unmanned aerial vehicle communication protocol feature classification method and system based on artificial intelligence, electronic equipment and storage medium

Abstract

The invention provides an unmanned aerial vehicle communication protocol feature classification method and system based on artificial intelligence, electronic equipment and a storage medium, and relates to the technical field of unmanned aerial vehicle cluster communication and networking. Calculating signal propagation angles of adjacent unmanned aerial vehicles to generate coverage included angle distribution, and calculating electromagnetic reflection path loss to generate a multipath effect correction factor; constructing a polyhedral structure, determining an actual communication coverage center according to the geometric center and the offset of the polyhedral structure, and calculating a three-dimensional offset vector; inputting the vector, the coverage distribution data and the correction factor into a path planning neural network, comprehensively evaluating electromagnetic interference, node displacement and a reflection path, and then outputting a protocol feature classification result; the data is decomposed into feature subsets through a fixed standby node rule, and the feature subsets are transmitted in parallel through multiple relay nodes and recombined at a receiving end, so that the communication anti-interference and transmission reliability in a complex environment can be remarkably improved.

Description

Technical Field

[0001] The present application relates to the field of drone cluster communication and networking technology, and in particular to an artificial intelligence-based drone communication protocol feature classification method, system, electronic device and storage medium. Background Art

[0002] In emergency scenarios like disaster relief, drone swarms must collaborate to perform missions in environments with strong electromagnetic interference. These environments present multiple challenges: collapsed buildings or mountain obstructions complicate signal propagation paths and create significant multipath effects. Furthermore, densely distributed interference sources, such as damaged power facilities, frequently disrupt communication links.

[0003] Currently, the mainstream solution uses a static protocol switching mechanism based on signal strength threshold judgment. This solution uses onboard sensors to collect signal strength, signal-to-noise ratio, and link quality indicators, and combines them with a preset terrain database to match environmental parameters. It then selects a communication protocol based on fixed threshold rules. Furthermore, some drones are designated as static relay nodes, forwarding data along a preset path to maintain basic connectivity.

[0004] However, this solution has significant flaws: the static threshold cannot respond to sudden strong interference sources, resulting in frequent protocol mis-switching or link interruption, and the misjudgment rate remains high in actual measurements; static relay nodes cannot dynamically adjust according to topology changes, causing some nodes to be overloaded with energy and exhausted prematurely, and the parallel transmission capacity is limited. In actual applications, data loss is serious, making it difficult to meet the needs of efficient resource scheduling in disaster scenarios. Summary of the Invention

[0005] The purpose of this application is to provide an artificial intelligence-based drone communication protocol feature classification method, system, electronic device and storage medium to solve the problem of high misjudgment rate in the existing technology.

[0006] To solve the above technical problems, in the first aspect, the present application provides a method for classifying UAV communication protocol features based on artificial intelligence, comprising: Obtain the spatial position data of each drone in the drone group and the preset communication coverage center offset; Calculating signal propagation angles between adjacent drones based on the spatial position data, generating distribution data of signal coverage angles based on the signal propagation angles, calculating path loss due to electromagnetic reflection based on the spatial position data, and generating a correction factor for multipath effects based on the path loss; Constructing a polyhedron structure based on the spatial position data of the drone, determining the center position of actual communication coverage according to the geometric center of the polyhedron structure in combination with the preset communication coverage center offset, and calculating a three-dimensional offset vector pointing from the geometric center to the center position; Inputting the three-dimensional offset vector, the distribution data of the signal coverage angle, and the correction factor of the multipath effect into a path planning neural network, performing a comprehensive evaluation of the electromagnetic interference intensity, node displacement deviation, and signal reflection path through the path planning neural network to generate a protocol feature classification result; Based on the protocol feature classification results, a backup node rule is formulated, and the transmission protocol data of the target UAV is decomposed into multiple feature subsets according to the backup node rule. The multiple feature subsets are allocated to selected relay nodes, transmitted in parallel through multiple paths, and finally reassembled at the receiving end.

[0007] Optionally, the three-dimensional offset vector, the distribution data of the signal coverage angle, and the correction factor of the multipath effect are input into a path planning neural network, and the path planning neural network is used to comprehensively evaluate the electromagnetic interference intensity, node displacement deviation, and signal reflection path to generate a protocol feature classification result, including: Decomposing the three-dimensional offset vector into three axial components, and forming an input feature sequence together with the statistical density value in the distribution data of the signal coverage angle and the correction factor of the multipath effect; Performing a nonlinear transformation on the input feature sequence using a path planning neural network, generating an electromagnetic interference intensity evaluation value based on the concentration of high-density angle intervals in the distribution data, comparing the modulus of the three-dimensional offset vector with a preset threshold to generate a node displacement deviation evaluation value, and generating a signal reflection path evaluation value by taking the inverse of the multipath effect correction factor; The electromagnetic interference intensity evaluation value, the node displacement deviation evaluation value and the signal reflection path evaluation value are weighted and summed to obtain a comprehensive stability score; According to the preset interval range to which the comprehensive stability score belongs, the corresponding protocol feature classification results including high stability, medium stability and low stability are output.

[0008] Optionally, a backup node rule is formulated based on the protocol feature classification result, and the transmission protocol data of the target UAV is decomposed into multiple feature subsets according to the backup node rule. The multiple feature subsets are assigned to selected relay nodes, transmitted in parallel through multiple paths, and finally reassembled at the receiving end, including: Based on the stability level in the protocol feature classification result, select drones in the drone group with stability levels higher than a set threshold as backup relay nodes, and assign a priority value to each backup relay node; Formulate a backup node rule based on the field type determined in the protocol feature classification result, and decompose the transmission protocol data of the target UAV into three feature subsets: control instruction field, status feedback field, and mission data field through the backup node rule; According to the priority value, the control instruction field is allocated to the backup relay node with the highest priority value, the status feedback field is allocated to the backup relay node with the second highest priority value, and the task data field is allocated to the remaining backup relay nodes; Each backup relay node transmits the assigned feature subset in parallel through an independent communication path, and at the receiving end, reorganizes the control instruction field, status feedback field and task data field into complete transmission protocol data according to the field order preset by the protocol feature classification result.

[0009] Optionally, a backup node rule is formulated based on the field type determined in the protocol feature classification result, and the transmission protocol data of the target UAV is decomposed into three feature subsets: control instruction field, status feedback field, and mission data field through the backup node rule, including: Identifying a field type identifier in the protocol feature classification result, wherein the field type identifier includes a control instruction identifier, a status feedback identifier, and a task data identifier; generating extraction rules according to the numerical ranges of the control instruction identifier, the state feedback identifier, and the task data identifier, respectively, and combining the extraction rules to form a backup node rule; Traversing all field units of the transmission protocol data of the target UAV, if the identifier of the field unit falls within the numerical range of the control instruction identifier, assigning the field unit to the control instruction field set; if it falls within the numerical range of the status feedback identifier, assigning the field unit to the status feedback field set; if it falls within the numerical range of the mission data identifier, assigning the field unit to the mission data field set; The control instruction field set is encapsulated into an independent control instruction field, the status feedback field set is encapsulated into an independent status feedback field, and the task data field set is encapsulated into an independent task data field.

[0010] Optionally, calculating the signal propagation angle between adjacent drones based on the spatial position data, generating distribution data of the signal coverage angle based on the signal propagation angle, calculating the path loss of electromagnetic reflection based on the spatial position data, and generating a correction factor for the multipath effect based on the path loss include: Based on the spatial position data of each drone in the drone group, calculate the straight-line distance between any two adjacent drones, and determine the signal propagation angle between the adjacent drones based on the difference between the straight-line distance and the altitude coordinate, where the signal propagation angle includes an elevation component and an azimuth component in the signal propagation direction; Mapping the elevation component and the azimuth component to preset angle intervals respectively to generate distribution data of signal coverage angles, wherein the distribution data includes statistical density of signal propagation directions within each angle interval; Based on the straight-line distance and the preset electromagnetic wave attenuation coefficient, the direct path loss of the electromagnetic wave between adjacent drones is calculated. At the same time, the additional loss of the reflection path is calculated based on the relative position of the spatial position data and the environmental reflective surface, combined with the material property parameters of the reflective surface; The direct path loss and the additional loss are added to obtain a total path loss, and a correction factor for the multipath effect is generated according to a ratio of the total path loss to a preset reference loss.

[0011] Optionally, constructing a polyhedron structure based on the spatial position data of the drone, determining the center position of actual communication coverage according to the geometric center of the polyhedron structure in combination with a preset communication coverage center offset, and calculating a three-dimensional offset vector pointing from the geometric center to the center position includes: Generate a polyhedron structure using the spatial position coordinates in the spatial position data as vertex coordinates, and calculate the arithmetic mean of all the vertex coordinates of the polyhedron structure as the geometric center; Obtaining a preset communication coverage center offset, wherein the communication coverage center offset includes a horizontal offset parameter and a vertical offset parameter, and determining a maximum coverage plane based on the spatial vertex distribution of the polyhedron structure; Determine a horizontal projection point on the maximum coverage plane according to the horizontal offset parameter, and move the horizontal projection point along the normal direction of the maximum coverage plane by the vertical offset parameter to determine the center position of actual communication coverage; The direction vector and the Euclidean distance between the two coordinate points pointing from the geometric center to the center position are calculated, the direction vector is normalized and multiplied by the Euclidean distance to generate a three-dimensional offset vector.

[0012] Optionally, the electromagnetic interference intensity evaluation value, the node displacement deviation evaluation value, and the signal reflection path evaluation value are weighted and summed to obtain a comprehensive stability score, including: Obtaining preset weight ratios corresponding to the electromagnetic interference intensity evaluation value, the node displacement deviation evaluation value, and the signal reflection path evaluation value; Multiplying the electromagnetic interference intensity evaluation value, the node displacement deviation evaluation value, and the signal reflection path evaluation value by the corresponding weight ratio to generate a node displacement weighted result, and superimposing the electromagnetic interference weighted result, the node displacement weighted result, and the signal reflection weighted result to generate an initial comprehensive score; The initial comprehensive score is adjusted according to the preset proportional value storage data, and the electromagnetic interference weighted result, the node displacement weighted result and the signal reflection weighted result are re-superimposed to generate a comprehensive stability score.

[0013] In a second aspect, this application provides an artificial intelligence-based drone communication protocol feature classification system, including: The acquisition module is used to obtain the spatial position data of each drone in the drone group and the preset communication coverage center offset; a generation module, configured to calculate a signal propagation angle between adjacent UAVs based on the spatial position data, generate distribution data of a signal coverage angle based on the signal propagation angle, calculate a path loss of electromagnetic reflection based on the spatial position data, and generate a correction factor for a multipath effect based on the path loss; a calculation module, configured to construct a polyhedron structure based on the spatial position data of the drone, determine the center position of actual communication coverage according to the geometric center of the polyhedron structure in combination with a preset communication coverage center offset, and calculate a three-dimensional offset vector pointing from the geometric center to the center position; an evaluation module, configured to input the three-dimensional offset vector, the distribution data of the signal coverage angle, and the correction factor of the multipath effect into a path planning neural network, perform a comprehensive evaluation of the electromagnetic interference intensity, the node displacement deviation, and the signal reflection path through the path planning neural network, and generate a protocol feature classification result; A decomposition module is used to formulate a backup node rule based on the protocol feature classification result, decompose the transmission protocol data of the target UAV into multiple feature subsets according to the backup node rule, and distribute the multiple feature subsets to selected relay nodes, transmit them in parallel through multiple paths, and finally reassemble them at the receiving end.

[0014] In a third aspect, the present application provides an electronic device, comprising: memory for storing computer programs; A processor is configured to implement the steps of the method for classifying drone communication protocol features based on artificial intelligence as described in the first aspect above when executing the computer program.

[0015] In a fourth aspect, the present application provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it can implement the steps of an artificial intelligence-based drone communication protocol feature classification method as described in the first aspect above.

[0016] The present application provides an artificial intelligence-based drone communication protocol feature classification method. By acquiring the spatial position data of the drone group and the offset of the communication coverage center, a dynamic topology model can be constructed to provide basic data support for the calibration of the communication coverage center, thereby solving the signal blind spot problem caused by formation displacement. By calculating the signal propagation angle and generating coverage angle distribution data, and combining the path loss to generate the multipath effect correction factor, the signal attenuation and interference intensity in complex environments can be quantified to achieve accurate modeling of the physical layer channel. By constructing a polyhedron structure and calculating the three-dimensional offset vector to determine the actual communication coverage center, the signal coverage position can be dynamically calibrated to avoid coverage drift caused by formation changes and improve link stability. By inputting the three-dimensional offset vector, coverage distribution data and correction factor into the path planning neural network to generate the protocol feature classification result, electromagnetic interference, node displacement and reflection path can be comprehensively evaluated to achieve environmental adaptive dynamic protocol optimization decision-making. By decomposing the protocol data into feature subsets based on the classification results, and transmitting and recombining them in parallel through multiple relay nodes, the anti-interference capability and transmission robustness can be improved. Dynamic resource scheduling can be achieved by combining the backup node rules, significantly reducing the risk of packet loss. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present application or the technical solutions of the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0018] Figure 1 A flowchart of an artificial intelligence-based drone communication protocol feature classification method provided in an embodiment of the present application; Figure 2 A schematic diagram of a specific implementation of an artificial intelligence-based drone communication protocol feature classification method provided in an embodiment of the present application; Figure 3 A scene diagram of an artificial intelligence-based drone communication protocol feature classification method provided in an embodiment of the present application; Figure 4 A schematic diagram of the structure of an artificial intelligence-based drone communication protocol feature classification system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0019] Research has found that existing drone communication protocols exhibit significant flaws in complex electromagnetic environments and dynamic topology scenarios. Traditional solutions rely on static threshold mechanisms, which are unable to effectively respond to sudden strong interference sources, leading to frequent protocol switching misjudgments and link interruptions. Furthermore, static relay node allocation lacks dynamic adjustment capabilities, causing some nodes to fail prematurely due to excessive energy loads. Furthermore, insufficient parallel transmission capacity leads to severe data loss due to multipath effects and path loss. These issues are particularly prominent in scenarios requiring efficient resource scheduling, such as disaster relief. There is an urgent need for an intelligent protocol classification method that can adapt to environmental changes and achieve multi-dimensional collaboration.

[0020] To address these issues, this paper proposes an artificial intelligence-based UAV communication protocol feature classification method. Its core lies in integrating spatial topology modeling, multipath correction, and neural network dynamic evaluation to achieve intelligent decomposition and parallel scheduling of protocol features. Specifically, a polyhedron structure is first constructed using the spatial position data of the UAV group. A three-dimensional offset vector is generated by combining the geometric center with a preset offset to accurately locate the dynamic communication coverage center. Simultaneously, coverage angle distribution data is generated based on the signal propagation angle, and path loss is quantified to generate a multipath correction factor. The three-dimensional offset vector, coverage angle distribution, and correction factor are then input into a path planning neural network, which comprehensively evaluates electromagnetic interference intensity, node displacement deviation, and signal reflection path to generate a protocol feature classification result. Finally, based on the classification results, backup node rules are formulated, and the target protocol data is decomposed into feature subsets. These are then transmitted in parallel through multiple relay nodes and reassembled at the receiving end. This method significantly improves dynamic anti-interference response capabilities, optimizes thresholds and relay strategies, extends node lifespan through a load balancing mechanism, and enhances parallel reliability through multipath redundant transmission. This fundamentally addresses the three major shortcomings of static protocols: delayed response to sudden interference, poor topology adaptability, and low parallel efficiency.

[0021] In order to enable those skilled in the art to better understand the present application, the present application is further described in detail below in conjunction with the accompanying drawings and specific embodiments. Obviously, the embodiments described are only a part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without making any creative efforts are within the scope of protection of the present application.

[0022] The core of this application is to provide a method for classifying UAV communication protocol features based on artificial intelligence. The flowchart of a specific implementation method is as follows: Figure 1 As shown, the method includes: S101, obtaining the spatial position data of each drone in the drone group and a preset communication coverage center offset; In the above steps, spatial position data refers to the three-dimensional coordinate information collected by the positioning sensor carried by the drone, including longitude, latitude, altitude, and attitude angles such as pitch angle and yaw angle, which is used to describe the precise position and flight status of the drone in the air; the communication coverage center offset is a preset adjustment parameter used to compensate for the deviation between the actual communication center and the theoretical geometric center, and is usually pre-set based on environmental interference or mission requirements.

[0023] In the embodiment of the present application, first, in step S101, the positioning device carried by the drone collects original position information. The drone receives GPS satellite signals to obtain initial three-dimensional coordinates, longitude, latitude, and altitude. At the same time, the barometer is used to measure the current altitude. Then, combined with the gyroscope, accelerometer, and magnetometer data in the nine-axis attitude sensor, the current pitch angle, yaw angle, and other attitude parameters of the fuselage are calculated. Then, the host drone broadcasts its own position data to the entire drone group through high-precision wireless communication technology such as the RTK dynamic positioning system. After receiving the data, the other drones fuse the satellite positioning data with the local sensor data through the Kalman filter algorithm. At the same time, the system loads a preset communication coverage center offset from a pre-stored mission configuration file. The offset is a three-dimensional adjustment parameter generated by the ground control station based on electromagnetic interference data or terrain scanning results in historical missions. Finally, the fused and corrected spatial position data, including coordinates and attitude angles, are packaged together with the offset parameters into a standard data frame format, and a timestamp is added, and transmitted to the subsequent calculation module for path planning. For example, in an agricultural spraying mission with 10 drones, the original coordinates of the host drone No. 1 are 100 meters east longitude, 200 meters north latitude, and 50 meters above sea level. After broadcasting the position through RTK technology, the slave drone No. 2 receives the data and corrects the altitude value based on the actual altitude of 51 meters detected by its own barometer. Then, the preset communication coverage center offset parameters are loaded to offset 5 meters east and 1 meter upward. The final output corrected position data is 105 meters east longitude, 200 meters north latitude, and 51 meters above sea level, and the timestamp is marked at 14:30:00 and transmitted to the next link.

[0024] In a real-world disaster relief scenario, five drones deployed in Area A form an emergency communications group, requiring a coordinated communication network to be established in a complex electromagnetic environment. First, the spatial position data of each drone is acquired using a radar detection system and an optoelectronic tracking system: UAV U1 is located at coordinates (120m, 80m, 50m), U2 at (90m, 110m, 55m), U3 at (150m, 70m, 48m), U4 at (105m, 95m, 52m), and U5 at (130m, 100m, 53m). Furthermore, a communication coverage center offset of (5m, 3m, 2m) is preset to compensate for deviations from the ideal communication center point due to terrain obstruction or signal interference in the real environment. This position data is then processed through spatiotemporal alignment, fusing the range information provided by the radar with the high-precision angle information from the optoelectronic system to generate a set of three-dimensional coordinates for each drone, which serves as the basic input for subsequent communication protocol optimization.

[0025] In the overall solution of the above-mentioned step S101, by obtaining the spatial position data of each drone in the drone group and the preset communication coverage center offset, a cluster spatial topology relationship network is dynamically constructed, and the communication coverage center direction of each drone is corrected based on the vector superposition of the position data and the offset, so as to realize the adaptive alignment of the multi-machine collaborative communication beam, effectively suppress the signal attenuation caused by the fluctuation of the flight attitude, and significantly improve the stability and anti-interference capability of the drone group communication link in a complex airspace environment.

[0026] S102: Calculate signal propagation angles between adjacent UAVs based on the spatial position data, generate distribution data of signal coverage angles based on the signal propagation angles, calculate electromagnetic reflection path loss based on the spatial position data, and generate a multipath correction factor based on the path loss; Optionally, step S102 may specifically include the following steps: S1021. Calculate the straight-line distance between any two adjacent drones based on the spatial position data of each drone in the drone group, and determine the signal propagation angle between the adjacent drones based on the difference between the straight-line distance and the altitude coordinate, where the signal propagation angle includes an elevation component and an azimuth component in the signal propagation direction. S1022. Map the elevation component and the azimuth component to preset angle intervals, respectively, to generate distribution data of signal coverage angles, the distribution data including statistical density of signal propagation directions within each angle interval; S1023. Calculate the direct path loss of electromagnetic waves between adjacent UAVs based on the straight-line distance and a preset electromagnetic wave attenuation coefficient. Simultaneously, calculate the additional loss of the reflection path based on the spatial position data and the relative position of the environmental reflective surface, combined with the material property parameters of the reflective surface. S1024: Add the direct path loss and the additional loss to obtain a total path loss, and generate a correction factor for the multipath effect based on a ratio of the total path loss to a preset reference loss.

[0027] In the above steps, the signal propagation angle refers to the electromagnetic wave transmission direction parameter calculated based on the longitude, latitude and altitude of the three-dimensional coordinate data of the adjacent drones, including the elevation component of the electromagnetic wave propagation direction relative to the horizontal plane and the azimuth component relative to the true north direction; the distribution data of the signal coverage angle is a data set generated by mapping the calculated elevation and azimuth angles to preset angle intervals, and counting the frequency density of the signal direction in each interval, reflecting the concentration trend of the signal propagation direction within the group; the path loss is the signal strength attenuation value caused by distance diffusion and environmental reflection during the transmission of electromagnetic waves, including the inherent attenuation of the direct path which is proportional to the square of the distance and the additional attenuation of the reflected path caused by the surrounding The correction factor for multipath effect is an adjustment coefficient calculated by the ratio of total path loss to the reference loss in an ideal non-reflective environment. It is used to quantify the impact of multipath interference and the distortion caused by the superposition of signal reflections along different paths on communication quality. A larger ratio indicates more severe multipath interference. The straight-line distance is the Euclidean distance between the three-dimensional coordinates of adjacent drones and is used to calculate the signal propagation angle and path loss. The electromagnetic wave attenuation coefficient is a physical parameter that describes the inherent attenuation of electromagnetic waves when propagating in free space and is positively correlated with frequency. The material property parameter of the environmental reflective surface is a coefficient that quantifies the reflection loss of electromagnetic waves due to different materials, such as concrete and glass, and is used to calculate the additional loss of the reflection path.

[0028] In the embodiment of the present application, first, the signal propagation angle is calculated through step S1021: according to the three-dimensional coordinates of two adjacent drones, such as drone A is located at 100 meters east longitude, 200 meters north latitude, and 50 meters altitude, and drone B is located at 130 meters east longitude, 180 meters north latitude, and 55 meters altitude, first calculate the horizontal distance difference rice, Meters, get the horizontal distance Meters, then calculate the height difference Meters; then, use the geometric relationship to calculate the elevation component of the signal propagation, that is, the angle between the signal and the horizontal plane , and the azimuth component, which is the horizontal deviation angle of the signal relative to the north direction , and corrected according to the coordinate system quadrant as , because drone B is located in the southwest of A, it is necessary to add For example, in an agricultural spraying mission, the coordinate difference between machine 1 and machine 2 is rice, rice, meters, the final output elevation angle , azimuth .

[0029] Secondly, step S1022 generates distribution data for signal coverage angles: the elevation and azimuth angles calculated above are categorized into preset intervals, such as 0°-15° and 15°-30° for elevation angles, and 315°-360° and 270°-315° for azimuth angles. The frequency of angle values ​​within each interval is counted. For example, if 65 of the 100 measurements show that the elevation angles are within the 0°-15° interval, then the density value is divided by the total number of measurements to obtain a density value of 0.65. This ultimately forms an angle density mapping table, which is used to describe the concentration trend of signal propagation directions within the drone group. For example, in a certain mission, 70% of the elevation angles are within the 0°-15° interval and 60% of the azimuth angles are within the 270°-315° interval, indicating that the signal mainly propagates westward at low elevation angles.

[0030] Next, the path loss is calculated in step S1023: The path loss is divided into two parts: First, the direct path loss is calculated based on the natural attenuation model of electromagnetic waves in the air. The formula is: direct path loss ,in, is the signal frequency in megahertz MHz, is the distance in kilometers. For example, when the frequency distance When , the calculation process is: ,but decibel Then calculate the additional loss caused by environmental reflection, such as the reflection coefficient of the concrete wall between the drone and the receiving point. , the reflection attenuation formula is The total path loss is the sum of the two For example, when transmitter 1 transmits to transmitter 2, the direct distance of 36.1 meters results in a loss of 71.26 dB, and the reflection from the concrete wall adds an additional 3.1 dB, for a total loss of 74.36 dB.

[0031] Finally, in step S1024, a correction factor for the multipath effect is generated: the total path loss 74.36 dB is compared with the benchmark loss in an ideal non-reflective environment, i.e., the direct loss 71.26 dB. This value is used to quantify the severity of the distortion caused by multipath interference such as signal reflection superposition. For example, if the correction factor of a transmission is 1.2, it means that the signal attenuation is increased by an additional , the anti-interference coding strategy needs to be enabled.

[0032] In actual applications, a drone group consists of 5 drones, performing low-altitude collaborative monitoring tasks. The spatial position data of each drone obtained by the positioning system, that is, the three-dimensional coordinates, in meters, are drone A: (100, 200, 300), drone B: (150, 220, 320), drone C: (90, 180, 290), drone D: (120, 250, 310), drone E: (80, 210, 305). Based on these data, the signal propagation angle is first calculated using drones A and B as an example. The straight-line distance between the two is calculated to be about 57.45 meters using the three-dimensional distance formula, and the height difference is 20 meters, the horizontal distance is about 53.85 meters, and then the elevation component of B relative to A is calculated to be about 20.56° by arctan (height difference / horizontal distance), and the azimuth component of B relative to A is calculated to be about 21.80° by arctan ((220-200) / (150-100)). Similarly, the elevation and azimuth components of other adjacent drone pairs can be calculated; then these angle components are mapped to the preset angle intervals, where the elevation intervals are set to [0°, 30°), [30°, 60°), [60°, 90°), and the azimuth intervals are set to [0°, 90°), [90°, 180°), [ 180°, 270°), [270°, 360°), and statistics show that among the total 8 pairs of adjacent relationships, there are 6 pairs with elevation angles in the interval [0°, 30°), accounting for 75%, 2 pairs in the interval [30°, 60°), accounting for 25%, 5 pairs in the interval [0°, 90°), accounting for 62.5%, and 3 pairs in other intervals, accounting for 37.5%, forming a distribution data containing the statistical density of signal propagation direction in each angle interval; then, taking drones A and B as an example to calculate the path loss, the free space loss formula is used to calculate the direct path loss, where the signal wavelength is 0.1 meter, and the calculated value is about 85.2dB. At the same time, considering the environment The reflecting surface is a concrete floor, and the reflection loss coefficient is 8dB. Combined with the relative positions of A and B to the ground, and the heights of 300 meters and 320 meters respectively, it is calculated that the reflection path is 12 meters longer than the direct path, and the additional loss is about 1.3dB; finally, the direct path loss and the additional loss are added together to obtain a total path loss of 86.5dB. The free space loss of 92dB at a preset distance of 100 meters is used as the benchmark loss, and the multipath effect correction factor is calculated to be approximately 0.94. Through this process, the signal propagation angle analysis, coverage distribution statistics and multipath effect correction factor calculation of the drone group are completed, providing data support for subsequent communication parameter optimization.

[0033] In the overall solution of step S102, the aforementioned steps accurately capture the elevation and azimuth components of signal propagation from adjacent drones, generating signal coverage angle distribution data that provides a basis for optimizing the coverage of drone group communications. Furthermore, the path loss and multipath correction factors calculated based on spatial location and environmental parameters accurately reflect electromagnetic propagation characteristics and improve the accuracy of the signal transmission model. This lays the foundation for dynamic adjustment of inter-UAV communication parameters, enhanced anti-interference capabilities, and guaranteed communication link stability, effectively improving the reliability and efficiency of drone group collaborative operations.

[0034] S103: constructing a polyhedron structure based on the spatial position data of the drone, determining the center position of actual communication coverage according to the geometric center of the polyhedron structure and the preset communication coverage center offset, and calculating a three-dimensional offset vector pointing from the geometric center to the center position; Optionally, step S103 may specifically include the following steps: S1031, generating a polyhedron structure using the spatial position coordinates in the spatial position data as vertex coordinates, and calculating the arithmetic mean of all vertex coordinates of the polyhedron structure as the geometric center; S1032. Obtain a preset communication coverage center offset, where the communication coverage center offset includes a horizontal offset parameter and a vertical offset parameter, and determine a maximum coverage plane based on the spatial vertex distribution of the polyhedron structure; S1033: Determine a horizontal projection point on the maximum coverage plane according to the horizontal offset parameter, and move the horizontal projection point along the normal direction of the maximum coverage plane by a distance equal to the vertical offset parameter to determine the center position of actual communication coverage; S1034. Calculate the direction vector and the Euclidean distance between the two coordinate points pointing from the geometric center to the center position, normalize the direction vector and multiply it by the Euclidean distance to generate a three-dimensional offset vector.

[0035] In the above steps, spatial position coordinates are the specific location data of each drone in three-dimensional space, including values ​​in the x, y, and z directions. Vertex coordinates are the coordinates of the corner points of the polyhedron, and here, the spatial position coordinates of the drone are directly used. The geometric center is the arithmetic mean of the coordinates of all the vertices of the polyhedron, representing the center position of the polyhedron in space. The communication coverage center offset is a preset parameter used to adjust the coverage center position. It includes horizontal and vertical offset parameters. The horizontal offset parameter adjusts the position horizontally, and the vertical offset parameter adjusts the position vertically. The maximum coverage plane is the plane with the largest number of vertices in the polyhedron and serves as the reference plane for horizontal offset. The horizontal projection point is the point on the maximum coverage plane determined by the horizontal offset parameter and is the base position of the coverage center on the plane. The normal direction of the maximum coverage plane is the direction perpendicular to the plane and is used to determine the direction of the vertical offset. The actual communication coverage center position is the final coverage center determined after horizontal and vertical offsets, and is used to locate the core area of ​​communication coverage. The direction vector is a vector pointing from the geometric center to the actual communication coverage center. It is composed of the difference between the coordinates of two points and is used to indicate the direction of the offset. The Euclidean distance is the straight-line distance from the geometric center to the actual coverage center and is used to indicate the length of the offset. The three-dimensional offset vector is the vector obtained by multiplying the normalized direction vector by the Euclidean distance. It fully indicates the direction and magnitude of the offset and is used for subsequent adjustment of communication parameters.

[0036] In the embodiment of the present application, first, a polyhedron structure is generated and its geometric center is calculated in step S1031. Specifically, the spatial position coordinates of each drone in the drone group are used as vertices, these vertices are connected to form a polyhedron, and the arithmetic mean of the x, y, and z coordinates of all vertices is calculated to obtain the geometric center. For example, there are 5 drones with coordinates A(100,200,300), B(150,220,320), C(90,180,290), D(120,250,310), and E(80,210,305). After generating a pentahedron with these coordinates as vertices, the sum of the x-coordinates is 100+150+90+120+80=540, and the average is 540÷5=108; the sum of the y-coordinates is 200+220+180+250+210=1060, and the average is 1060÷5=212; the sum of the z-coordinates is 300+320+290+310+305=1525, and the average is 1525÷5=305, so the geometric center is (108,212,305).

[0037] Next, the preset offset is obtained in step S1032 and the maximum coverage plane is determined. First, the communication coverage center offset, which includes the horizontal and vertical offset parameters, is obtained. All vertices of the polyhedron are analyzed, and the number of vertices contained in each plane is counted. The plane with the most vertices is found as the maximum coverage plane. For example, if the preset horizontal offset parameters are 5 and -3, and the vertical offset parameter is 2, and the pentahedron composed of A, B, C, D, and E is analyzed, it is found that the four points A, B, D, and E are located in the same plane, and this plane contains the most vertices, so this plane is determined as the maximum coverage plane.

[0038] Next, the center position of the actual communication coverage is determined through step S1033. First, find the projection point of the geometric center on the maximum coverage plane. Using this as the starting point, move the horizontal projection point on the plane according to the horizontal offset parameter to obtain the horizontal projection point. Then, move the horizontal projection point along the normal direction of the maximum coverage plane according to the vertical offset parameter to obtain the actual coverage center. For example, the projection of the geometric center (108, 212, 305) on the maximum coverage plane is (108, 212, 300). After moving the horizontal offset parameters 5 and -3 on the plane, the horizontal projection point is (108+5, 212-3, 300), that is, (113, 209, 300). The normal direction of the plane is vertically upward. After moving according to the vertical offset parameter 2, the actual coverage center is (113, 209, 300+2), that is, (113, 209, 302).

[0039] Finally, the three-dimensional offset vector is generated through step S1034. First, the direction vector from the geometric center to the actual coverage center is calculated, that is, the actual coverage center coordinates minus the geometric center coordinates to obtain the direction components; then the Euclidean distance between the two points is calculated, that is, the square root of the sum of the squares of the components, and the formula is ; Then normalize the direction vector, that is, divide each component by the Euclidean distance; finally, multiply the normalized direction vector by the Euclidean distance to obtain the three-dimensional offset vector. For example, the geometric center To the actual coverage center The direction vector is Right now ; Euclidean distance is ; The normalized direction vector is the sum of each component divided by 6.56. ; After multiplying it by the Euclidean distance, the three-dimensional offset vector is .

[0040] In practical applications, a group of 5 drones has spatial coordinates of A(100,200,300), B(150,220,320), C(90,180,290), D(120,250,310), and E(80,210,305). First, a polyhedron structure is generated with these coordinates as vertices. When calculating the geometric center, the sum of the x coordinates is calculated as 100+150+ 90+120+80=540, the average value is 540÷5=108, the sum of the y coordinates is 200+220+180+250+210=1060, the average value is 1060÷5=212, the sum of the z coordinates is 300+320+290+310+305=1525, the average value is 1525÷5=305, and the geometric center is (108,212,305); then, get the preset pass The offset of the coverage center is calculated, where the horizontal offset parameter is (5, -3) and the vertical offset parameter is 2. After analyzing the distribution of the polyhedron vertices, it is found that the plane where the four points A, B, D, and E are located contains the largest number of vertices, and it is determined to be the maximum coverage plane; then the projection point of the geometric center (108, 212, 305) on the maximum coverage plane is found to be (108, 212, 300), and the horizontal projection point is moved on the plane according to the horizontal offset parameter to obtain (108+5, 212-3, 300) = (113, 209, 300), and then moved along the normal direction of the plane (vertically upward) according to the vertical offset parameter 2, and the actual communication coverage center position is (113, 209, 302); finally, the direction vector pointing from the geometric center to the center position is calculated to be (113-108, 209-212, 302-305) = (5, -3, -3), and the Euclidean distance is , normalize the direction vector and multiply it by the Euclidean distance to generate a three-dimensional offset vector of (5,-3,-3).

[0041] In the overall solution of step S103 above, a precise polyhedron structure is constructed based on the spatial positions of the drones. Its geometric center intuitively reflects the spatial distribution core of the drone swarm. The maximum coverage plane is determined by combining it with a preset offset, providing a benchmark for adjusting the coverage center to align with the actual distribution. By adjusting the horizontal and vertical offset parameters in steps, the actual communication coverage center can be precisely located, ensuring that the coverage range matches the operational requirements of the drone swarm. The generated three-dimensional offset vector quantifies the center offset, providing a spatial reference for dynamic optimization of communication parameters, effectively improving the targeted and adaptable communication coverage and ensuring the stability and effectiveness of the communication link during collaborative operations of the drone swarm.

[0042] The embodiment of the present application provides a specific implementation diagram of a method for classifying features of drone communication protocols based on artificial intelligence, as shown in FIG. Figure 2 As shown, including the following: S104: Input the three-dimensional offset vector, the distribution data of the signal coverage angle, and the correction factor of the multipath effect into a path planning neural network, and use the path planning neural network to comprehensively evaluate the electromagnetic interference intensity, node displacement deviation, and signal reflection path to generate a protocol feature classification result; Optionally, step S104 may specifically include the following steps: S1041. Decompose the three-dimensional offset vector into three axial components, and use the three axial components together with the statistical density value in the distribution data of the signal coverage angle and the correction factor of the multipath effect to form an input feature sequence. S1042. Perform a nonlinear transformation on the input feature sequence using a path planning neural network, generate an electromagnetic interference intensity evaluation value based on the concentration of high-density angle intervals in the distribution data, compare the modulus of the three-dimensional offset vector with a preset threshold to generate a node displacement deviation evaluation value, and generate a signal reflection path evaluation value by taking the inverse of the multipath effect correction factor. S1043. Weighted sum of the electromagnetic interference intensity evaluation value, the node displacement deviation evaluation value, and the signal reflection path evaluation value to obtain a comprehensive stability score; Among them, step S1043 may specifically include the following processes: obtaining a preset weight ratio corresponding to the electromagnetic interference intensity evaluation value, the node displacement deviation evaluation value and the signal reflection path evaluation value; multiplying the electromagnetic interference intensity evaluation value, the node displacement deviation evaluation value and the signal reflection path evaluation value by the corresponding weight ratio to generate a node displacement weighted result, superimposing the electromagnetic interference weighted result, the node displacement weighted result and the signal reflection weighted result to generate an initial comprehensive score; adjusting the initial comprehensive score according to the preset ratio value storage data, re-superimposing the electromagnetic interference weighted result, the node displacement weighted result and the signal reflection weighted result to generate a comprehensive stability score.

[0043] S1044. Output corresponding protocol feature classification results including high stability, medium stability, and low stability according to the preset interval range to which the comprehensive stability score belongs.

[0044] In the above steps, the three-dimensional offset vector is a vector containing three axial components, which is used to represent the offset direction and size from the geometric center to the actual communication coverage center; the distribution data of the signal coverage angle is a data set containing the statistical density of the signal propagation direction in each angle interval, reflecting the distribution of signal coverage; the correction factor for the multipath effect is a parameter generated based on the ratio of the total path loss to the reference loss, and is used to correct the impact of multipath propagation. The input feature sequence is a data sequence consisting of the three axial components of the three-dimensional offset vector, the statistical density value in the distribution data, and the correction factor, which serves as the input of the neural network; the path planning neural network is a network model used to comprehensively evaluate electromagnetic interference, node displacement deviation, and signal reflection path; the electromagnetic interference intensity evaluation value is an evaluation value generated by calculating the concentration of high-density angle intervals in the distribution data, reflecting the degree of electromagnetic interference; the node displacement deviation evaluation value is an evaluation value generated by calculating the proportional relationship between the modulus of the three-dimensional offset vector and a preset threshold, reflecting the degree of node displacement deviation; the signal reflection path evaluation value is an evaluation value generated by calculating the inverse normalization of the correction factor, reflecting the quality of the signal reflection path; the comprehensive stability score is a score obtained by weighted summing and adjusting the proportion of the three evaluation values, reflecting the overall stability of the communication; the weight ratio is a preset weight value corresponding to the three evaluation values, used to adjust the influence of each evaluation value; the initial comprehensive score is a preliminary score obtained by multiplying the three evaluation values ​​by the corresponding weights; the protocol feature classification result is a classification result including high stability, medium stability, and low stability output according to the interval to which the comprehensive stability score belongs, which is used to identify the stability of the communication protocol feature.

[0045] In the embodiment of the present application, first, an input feature sequence is constructed through step S1041. Specifically, the three-dimensional offset vector is decomposed into three axial components in the x, y, and z axis directions, and then the statistical density values ​​of each high-density interval in the signal coverage angle distribution data are extracted. Finally, these axial components, statistical density values ​​and the correction factor of the multipath effect are arranged in sequence and combined into an input feature sequence. For example, the three-dimensional offset vector is (5, -3, -3), and the decomposed axial components are 5, -3, and -3; the statistical density of the elevation angle [0°, 30°) in the signal coverage angle distribution data is 75%, and the statistical density of the azimuth angle [0°, 90°) is 62.5%; the correction factor of the multipath effect is 0.94, then the input feature sequence is constructed in the order of 5, -3, -3, 75%, 62.5%, and 0.94.

[0046] Secondly, three evaluation values ​​are generated through step S1042. Specifically, the path planning neural network performs nonlinear transformation processing on the input feature sequence, first calculating the electromagnetic interference intensity evaluation value: the proportion of all high-density angle intervals in the statistical distribution data, with the elevation angle and azimuth angle each accounting for 50% weight, and summing them to obtain the concentration, that is, 75%×0.5+62.5%×0.5=68.75%, and then mapping the concentration to the 0-100 scoring range, 68.75% corresponds to the generated evaluation value of 85; then calculate the node displacement deviation evaluation value: first calculate the modulus of the three-dimensional offset vector, the formula is ,Right now , the preset threshold is 10, and the ratio of the modulus to the threshold is calculated as 6.56 / 10=0.656. This ratio is subtracted from 1 and multiplied by 100, that is, (1-0.656)×100≈34.4, and the integer is taken to generate the evaluation value 34; finally, the signal reflection path evaluation value is calculated: the reciprocal of the multipath effect correction factor is taken, that is, 1 / 0.94≈1.06. It is known that the normal range of the correction factor is 0.5-1.5, corresponding to the reciprocal range of 1.0-2.0, and normalized to a 0-100 score, that is, (1.06-1.0) / (2.0-1.0)×100=6, and the generated evaluation value is 100-6=94.

[0047] Next, the comprehensive stability score is calculated in step S1043. Specifically, the preset weight ratios are first obtained: the electromagnetic interference intensity evaluation value has a weight of 0.4, the node displacement deviation evaluation value has a weight of 0.3, and the signal reflection path evaluation value has a weight of 0.3. The weighted results of the three evaluation values ​​are calculated, namely 85×0.4=34, 34×0.3=10.2, and 94×0.3=28.2, and the initial comprehensive score is calculated by adding them together to obtain 34+10.2+28.2=72.4. The preset ratio adjustment value is 1.05, and the initial comprehensive score is multiplied by this ratio, namely 72.4×1.05≈76.02. The weighted results are then added together and rounded to obtain a final comprehensive stability score of 76.

[0048] Finally, step S1044 outputs the protocol feature classification results. Specifically, the preset comprehensive stability score ranges are: high stability (80-100), medium stability (60-79), and low stability (0-59). The calculated comprehensive stability score of 76 is compared with the range, and 76 falls within the 60-79 range. Therefore, the corresponding medium stability protocol feature classification result is output.

[0049] In practical applications, for a certain drone group, the three-dimensional offset vector obtained through preliminary calculation is (5, -3, -3). The statistical density of the elevation angle [0°, 30°) in the signal coverage angle distribution data is 75%, the statistical density of the azimuth angle [0°, 90°) is 62.5%, and the correction factor of the multipath effect is 0.94. First, the three-dimensional offset vector is decomposed into three axial components of x=5, y=-3, and z=-3, which together with the statistical density values ​​of 75%, 62.5% and the correction factor of 0.94 constitute the input feature sequence. Then, through the path planning neural network processing, the high-density interval concentration is calculated to be 75%×0.5+62.5%×0.5=68.75%, and the mapping generates an electromagnetic interference intensity evaluation value of 85; the modulus of the three-dimensional offset vector is , compared to the preset threshold of 10, the ratio is 0.656, generating a node displacement deviation evaluation value of 34. The reciprocal of the correction factor is approximately 1.06, and after normalization, the signal reflection path evaluation value is 94. Then, the weight ratios 0.4, 0.3, and 0.3 are obtained, and the weighted results are calculated as 85 × 0.4 = 34, 34 × 0.3 = 10.2, and 94 × 0.3 = 28.2. The initial comprehensive score is 72.4, and after adjusting the preset ratio by 1.05, the comprehensive stability score is 76. Finally, since 76 falls between 60 and 79, the protocol feature classification result is output as moderately stable.

[0050] In the overall solution of step S104 above, the three-dimensional offset vector, signal coverage angle distribution data, and multipath effect correction factor are effectively integrated as input features to provide a comprehensive evaluation basis for the path planning neural network. The neural network accurately generates evaluation values ​​for electromagnetic interference, node displacement deviation, and signal reflection path through nonlinear transformations. It combines preset weight ratios to achieve scientific weighting of multi-dimensional indicators. The comprehensive stability score obtained after proportional adjustment can objectively reflect the overall status of the communication system. The protocol feature classification results output based on the scoring interval can intuitively identify the communication stability level, providing clear guidance for the optimization of drone group communication parameters and the formulation of anti-interference strategies, effectively improving the adaptability and reliability of communication links, and ensuring the smooth implementation of collaborative operations.

[0051] S105. Formulate a backup node rule based on the protocol feature classification result, decompose the transmission protocol data of the target UAV into multiple feature subsets according to the backup node rule, and distribute the multiple feature subsets to the selected relay nodes, transmit them in parallel through multiple paths, and finally reassemble them at the receiving end.

[0052] Optionally, step S105 may specifically include the following steps: S1051. Based on the stability level in the protocol feature classification result, select drones in the drone group whose stability level is higher than a set threshold as backup relay nodes, and assign a priority value to each backup relay node; S1052. Formulate a backup node rule based on the field type determined in the protocol feature classification result, and decompose the transmission protocol data of the target UAV into three feature subsets: a control instruction field, a status feedback field, and a mission data field using the backup node rule; Among them, step S1052 may specifically include the following process: identifying the field type identifier in the protocol feature classification result, the field type identifier includes a control instruction identifier, a state feedback identifier and a task data identifier; generating extraction rules according to the numerical ranges of the control instruction identifier, the state feedback identifier and the task data identifier, respectively, and combining the extraction rules to form a backup node rule; traversing all field units of the transmission protocol data of the target UAV, if the identifier of the field unit falls within the numerical range of the control instruction identifier, then assigning the field unit to the control instruction field set; if it falls within the numerical range of the state feedback identifier, then assigning it to the state feedback field set; if it falls within the numerical range of the task data identifier, then assigning it to the task data field set; encapsulating the control instruction field set into an independent control instruction field, encapsulating the state feedback field set into an independent state feedback field, and encapsulating the task data field set into an independent task data field.

[0053] S1053. Based on the priority values, allocate the control instruction field to the backup relay node with the highest priority value, allocate the status feedback field to the backup relay node with the second highest priority value, and allocate the task data field to the remaining backup relay nodes. S1054. Each backup relay node transmits the assigned feature subset in parallel through an independent communication path, and at the receiving end, reorganizes the control instruction field, status feedback field and task data field into complete transmission protocol data according to the field order preset by the protocol feature classification result.

[0054] In the above steps, the protocol feature classification result is a classification identifier including the stability level. The stability level is a numerically quantified communication stability evaluation, which is used to guide the formulation of backup node rules. The backup node rules are rules for data decomposition and allocation formulated based on the field type and priority in the protocol feature classification result. The target UAV is the UAV that needs to transmit data. The transmission protocol data is the structured data that the target UAV needs to transmit, which contains multiple field units. The feature subset is an independent data fragment formed by the decomposition of the transmission protocol data according to the field type, including the control instruction field, the status feedback field and the task data field. The relay node is the UAV used to forward data, and the backup relay node is a relay node selected from the UAV group with a stability score higher than the set threshold. The set threshold is the judgment The minimum stability score for determining whether a drone can be used as a backup relay node is usually set to 60 points; the priority value is a serial number assigned from high to low according to the stability score of the backup relay node, with 1 being the highest priority; the field type is the content category of different functions in the transmission protocol data; the field type identifier is the numerical identifier of the field unit header, with the control instruction identifier ranging from 1 to 100, the status feedback identifier ranging from 101 to 200, and the mission data identifier ranging from 201 to 300; the extraction rule is the rule for filtering the corresponding field according to the identifier value range; the field unit is the smallest data unit with an identifier in the transmission protocol data; the independent communication path is the communication link exclusive to each relay node; the receiving end is the terminal that receives and reassembles the data; the complete transmission protocol data is the original data restored after reassembly by the receiving end.

[0055] In this embodiment of the present application, a backup relay node is first selected and assigned a priority in step S1051. Specifically, the stability score of each drone in the protocol feature classification results is first obtained. For example, drone A scores 95 points, B scores 90 points, C scores 85 points, D scores 70 points, and E scores 65 points, with a threshold of 60 points. The scores of each drone are compared with the threshold, and drones A, B, C, D, and E with scores greater than or equal to 60 points are selected as backup relay nodes. The nodes are then sorted from high to low by stability score, with A scoring 95 points higher than B, 90 points higher than C, 85 points higher than D, 70 points higher than E, and 65 points higher than E. Finally, priority values ​​are assigned according to the sorting order, with A being 1, B being 2, C being 3, D being 4, and E being 5.

[0056] Next, in step S1052, a backup node rule is formulated and the transmission protocol data is decomposed. Specifically, the field type definitions in the protocol feature classification results are first read, identifying that the control instruction identifier value range is 1 to 100, the status feedback identifier value range is 101 to 200, and the mission data identifier value range is 201 to 300. Based on these ranges, an extraction rule is generated. The rule content is that field units with identifiers between 1 and 100 are classified into the control instruction set, 101 to 200 into the status feedback set, and 201 to 300 into the mission data set. These rules are combined to form a backup node rule. Then, all field units of the target drone's transmission protocol data are traversed, and the identifier value of each field unit is checked. For example, the field unit with identifier 50 is classified into the control instruction set, 150 into the status feedback set, and 250 into the mission data set. Finally, the three sets are encapsulated into independent control instruction fields, status feedback fields, and mission data fields.

[0057] Next, step S1053 allocates feature subsets to backup relay nodes. Specifically, based on the backup relay node's priority value, the most important control instruction field is allocated to the node with the highest priority, i.e., A with priority 1. The less important status feedback field is allocated to the node with the next highest priority, i.e., B with priority 2. The data volume of the task data field is calculated. If the data volume is large, it is split according to the number of nodes. For example, the data is split into three parts and allocated to C with priority 3, D with priority 4, and E with priority 5, respectively.

[0058] Finally, the data is transmitted and reassembled in step S1054. Specifically, each standby relay node activates its own independent communication path. A transmits the control instruction field, B transmits the status feedback field, and C, D, and E transmit the split task data fields. All data is sent in parallel. The receiving end pre-stores the field reassembly order: the control instruction field first, the status feedback field in the middle, and the task data field last. The receiving end receives each field in sequence, first receiving and storing the control instruction field, then receiving and splicing the status feedback field, and finally receiving the task data fields transmitted by C, D, and E and splicing them in order, ultimately reassembling the complete transmission protocol data.

[0059] In a real-world application, for a group of drones, protocol feature classification results show that drone A has a high stability rating of 95, B has a high stability rating of 90, C has a medium stability rating of 85, D has a medium stability rating of 70, and E has a medium stability rating of 65. A threshold of 60 points is set. First, A, B, C, D, and E are selected as backup relay nodes, and their priorities are assigned from high to low stability levels: A is 1, B is 2, C is 3, D is 4, and E is 5. Next, the field type identifiers in the protocol feature classification results are identified: control command identifiers range from 1-100, status feedback identifiers range from 101-200, and mission data identifiers range from 201-300. Backup node rules are generated based on these identifiers. The field units of the target drone's transmitted protocol data are traversed, and identifiers 50 and 80 are classified into the control command field set, 120 and 150 into the status feedback field set, and 230, 260, and 290 into the mission data field set. These are then encapsulated into three feature subsets, respectively. Then, the control instruction field is assigned to A at priority 1, the status feedback field is assigned to B at priority 2, and the task data field is split into three and assigned to C at priority 3, D at priority 4, and E at priority 5. Finally, each node transmits in parallel through an independent path. The receiving end splices the control instruction, status feedback, and task data in the preset order and reassembles them into complete transmission protocol data.

[0060] In the overall solution of step S105 above, the protocol feature classification results are used to accurately screen out highly stable backup relay nodes and reasonably assign priorities to ensure that data transmission relies on reliable nodes. The backup node rules formulated based on field types achieve scientific splitting of transmission protocol data, allowing different types of data to be processed in a targeted manner. Allocating feature subsets by priority and transmitting them in parallel through multiple paths greatly improves data transmission efficiency while reducing the impact of single node or path failures on overall transmission. The orderly reorganization mechanism at the receiving end ensures data integrity. The overall process significantly enhances the reliability, flexibility, and anti-interference capability of drone group communications, providing stable data transmission support for efficient collaborative operations.

[0061] The following is a complete embodiment of steps S101 to S105: like Figure 3 As shown in the figure, a drone group performs a low-altitude collaborative monitoring mission. The group contains 5 drones. The spatial position data (three-dimensional coordinates, unit: meter) of each drone obtained by the positioning system are: drone A (100, 200, 300), B (150, 220, 320), C (90, 180, 290), D (120, 250, 310), E (80, 210, 305), and the preset communication coverage center offset is the horizontal offset parameter (5, -3) and the vertical offset parameter 2.

[0062] First, the signal propagation angle and multipath correction factor between adjacent drones are calculated based on the spatial position data. Taking A and B as an example, the straight-line distance is calculated using the three-dimensional distance formula. The height difference is 20 meters, the horizontal distance is approximately 53.85 meters, the elevation component is approximately 20.56°, and the azimuth component is approximately 21.80°. Similarly, the angles of other adjacent drone pairs are calculated to generate the signal coverage angle distribution data. The elevation angle range of [0°, 30°] accounts for 75%, and the azimuth angle range of [0°, 90°] accounts for 62.5%. When calculating path loss, the direct path loss between A and B is approximately 85.2 dB, with an additional loss of approximately 1.3 dB due to concrete floor reflection, for a total path loss of 86.5 dB. The ratio of this to the preset baseline loss of 92 dB generates a multipath correction factor of 0.94.

[0063] Next, a pentahedron structure was generated with the spatial position coordinates of the five drones as vertices. When calculating the geometric center, the average x-coordinate was (100+150+90+120+80)÷5=108, the average y-coordinate was (200+220+180+250+210)÷5=212, and the average z-coordinate was (300+320+290+310+305)÷5=305, resulting in the geometric center (108, 212, 305). Combined with the communication coverage center offset, the plane where A, B, D, and E are located is determined to be the maximum coverage plane. The projection of the geometric center on this plane is (108, 212, 300). The horizontal projection point is (113, 209, 300) after moving according to the horizontal offset parameter. After moving 2 meters along the normal direction, the actual communication coverage center position is (113, 209, 302). The three-dimensional offset vector pointing from the geometric center to this center is calculated to be (5, -3, -3).

[0064] Subsequently, the input feature sequence was formed from the axial component (5, -3, -3) of the three-dimensional offset vector, the statistical densities of the signal coverage angle distribution data (75% and 62.5%), and a correction factor of 0.94, and then fed into the path planning neural network. The neural network generated an electromagnetic interference intensity evaluation value of 85 based on the high-density interval concentration, a node displacement deviation evaluation value of 34 (modulus length ≈ 6.56) calculated by the ratio of the threshold of 10, and a signal reflection path evaluation value of 94 (normalized by the inverse of the correction factor). The initial score was 72.4, weighted by 0.4, 0.3, and 0.3, and the overall stability score after proportional adjustment was 76, generating a moderately stable protocol feature classification result.

[0065] Finally, a backup node rule is formulated based on the classification results: A, B, C, D, and E with a stability level higher than 60 points are selected as backup relay nodes, and are assigned priorities 1 to 5 according to the stability ranking; the field type identifiers in the protocol characteristics are identified, control instruction identifiers 1-100, status feedback 101-200, and mission data 201-300, and the target UAV transmission protocol data is decomposed into three types of fields accordingly; the control instruction field is assigned to A with priority 1, the status feedback field is assigned to B with priority 2, and the mission data field is split and assigned to C, D, and E; each node transmits in parallel through an independent path, and the receiving end reassembles the data in the preset order of control instruction → status feedback → mission data to obtain the complete transmission protocol data.

[0066] The artificial intelligence-based drone communication protocol feature classification method provided in this application realizes the accurate classification and optimized transmission of communication protocol features by integrating drone spatial position data and preset parameters. First, by calculating the signal propagation angle, path loss and correction factor, the spatial electromagnetic characteristics of drone group communication are fully captured, providing a solid data foundation for subsequent evaluation. A polyhedron structure is constructed and a three-dimensional offset vector is generated to accurately locate the core area of ​​communication coverage and ensure that the evaluation dimension fits the actual distribution. The comprehensive evaluation of multi-dimensional parameters by the path planning neural network realizes the quantitative classification of protocol feature stability and provides a scientific basis for the formulation of data transmission strategies. The backup node rules and multi-path parallel transmission mechanism based on the classification results greatly improve the reliability and efficiency of data transmission and reduce the impact of single node or path failures. The overall process deeply integrates spatial perception and communication optimization through artificial intelligence technology, significantly enhancing the communication adaptability, anti-interference ability and data integrity of drone groups during collaborative operations, and providing an efficient solution for drone cluster communications in complex environments.

[0067] Figure 4 This is a structural diagram of a specific implementation of an artificial intelligence-based drone communication protocol feature classification system provided in an embodiment of the present application, with reference to Figure 4 , the system may include: An acquisition module 41 is used to obtain the spatial position data of each drone in the drone group and a preset communication coverage center offset; a generating module 42 configured to calculate a signal propagation angle between adjacent UAVs based on the spatial position data, generate distribution data of a signal coverage angle based on the signal propagation angle, calculate a path loss due to electromagnetic reflection based on the spatial position data, and generate a correction factor for a multipath effect based on the path loss; A calculation module 43 is configured to construct a polyhedron structure based on the spatial position data of the drone, determine the center position of actual communication coverage based on the geometric center of the polyhedron structure and the preset communication coverage center offset, and calculate a three-dimensional offset vector pointing from the geometric center to the center position; An evaluation module 44 is configured to input the three-dimensional offset vector, the distribution data of the signal coverage angle, and the correction factor of the multipath effect into a path planning neural network, and perform a comprehensive evaluation of the electromagnetic interference intensity, node displacement deviation, and signal reflection path through the path planning neural network to generate a protocol feature classification result; The decomposition module 45 is used to formulate a backup node rule based on the protocol feature classification result, decompose the transmission protocol data of the target UAV into multiple feature subsets according to the backup node rule, and distribute the multiple feature subsets to the selected relay nodes, transmit them in parallel through multiple paths and finally reassemble them at the receiving end.

[0068] The artificial intelligence-based drone communication protocol feature classification system of the embodiment of the present application is used to implement the aforementioned artificial intelligence-based drone communication protocol feature classification method. Therefore, the specific implementation method of the artificial intelligence-based drone communication protocol feature classification system can be seen in the embodiment part of the artificial intelligence-based drone communication protocol feature classification method in the previous text. Its specific implementation method can refer to the description of the corresponding various parts of the embodiment, which will not be repeated here.

[0069] The present application also provides an electronic device, comprising: a memory for storing a computer program; and a processor for implementing the steps of any of the above-mentioned artificial intelligence-based drone communication protocol feature classification methods when executing the computer program.

[0070] The present application also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of any of the above-mentioned artificial intelligence-based drone communication protocol feature classification methods are implemented.

[0071] In an exemplary embodiment, the computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as a USB flash drive, a read-only memory, a random access memory, a mobile hard disk, a magnetic disk, or an optical disk.

[0072] An embodiment of the present invention also provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the steps of any of the above-mentioned artificial intelligence-based drone communication protocol feature classification method embodiments.

[0073] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the components and steps of each example according to their functions. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0074] The above is a detailed introduction to the artificial intelligence-based drone communication protocol feature classification method, system, electronic device, and storage medium provided by this application. This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of this application, several improvements and modifications can be made to this application, and these improvements and modifications also fall within the scope of protection of this application.

Claims

1. A UAV communication protocol feature classification method based on artificial intelligence, characterized by: include: Obtain the spatial position data of each drone in the drone group and the preset communication coverage center offset; Calculating signal propagation angles between adjacent drones based on the spatial position data, generating distribution data of signal coverage angles based on the signal propagation angles, calculating path loss due to electromagnetic reflection based on the spatial position data, and generating a correction factor for multipath effects based on the path loss; Constructing a polyhedron structure based on the spatial position data of the drone, determining the center position of actual communication coverage according to the geometric center of the polyhedron structure in combination with the preset communication coverage center offset, and calculating a three-dimensional offset vector pointing from the geometric center to the center position; Inputting the three-dimensional offset vector, the distribution data of the signal coverage angle, and the correction factor of the multipath effect into a path planning neural network, performing a comprehensive evaluation of the electromagnetic interference intensity, node displacement deviation, and signal reflection path through the path planning neural network to generate a protocol feature classification result; Based on the protocol feature classification results, a backup node rule is formulated, and the transmission protocol data of the target UAV is decomposed into multiple feature subsets according to the backup node rule. The multiple feature subsets are allocated to selected relay nodes, transmitted in parallel through multiple paths, and finally reassembled at the receiving end.

2. The method according to claim 1, characterized in that The three-dimensional offset vector, the distribution data of the signal coverage angle, and the correction factor of the multipath effect are input into a path planning neural network. The path planning neural network comprehensively evaluates the electromagnetic interference intensity, node displacement deviation, and signal reflection path to generate a protocol feature classification result, including: Decomposing the three-dimensional offset vector into three axial components, and forming an input feature sequence together with the statistical density value in the distribution data of the signal coverage angle and the correction factor of the multipath effect; Performing a nonlinear transformation on the input feature sequence using a path planning neural network, generating an electromagnetic interference intensity evaluation value based on the concentration of high-density angle intervals in the distribution data, comparing the modulus of the three-dimensional offset vector with a preset threshold to generate a node displacement deviation evaluation value, and generating a signal reflection path evaluation value by taking the inverse of the multipath effect correction factor; The electromagnetic interference intensity evaluation value, the node displacement deviation evaluation value and the signal reflection path evaluation value are weighted and summed to obtain a comprehensive stability score; According to the preset interval range to which the comprehensive stability score belongs, the corresponding protocol feature classification results including high stability, medium stability and low stability are output.

3. The method according to claim 1, characterized in that Formulate a backup node rule based on the protocol feature classification result, decompose the transmission protocol data of the target UAV into multiple feature subsets according to the backup node rule, and distribute the multiple feature subsets to selected relay nodes, transmit them in parallel through multiple paths, and finally reassemble them at the receiving end, including: Based on the stability level in the protocol feature classification result, select drones in the drone group with stability levels higher than a set threshold as backup relay nodes, and assign a priority value to each backup relay node; Formulate a backup node rule based on the field type determined in the protocol feature classification result, and decompose the transmission protocol data of the target UAV into three feature subsets: control instruction field, status feedback field, and mission data field through the backup node rule; According to the priority value, the control instruction field is allocated to the backup relay node with the highest priority value, the status feedback field is allocated to the backup relay node with the second highest priority value, and the task data field is allocated to the remaining backup relay nodes; Each backup relay node transmits the assigned feature subset in parallel through an independent communication path, and at the receiving end, reorganizes the control instruction field, status feedback field and task data field into complete transmission protocol data according to the field order preset by the protocol feature classification result.

4. The method according to claim 3, characterized in that A backup node rule is formulated based on the field type determined in the protocol feature classification result. The backup node rule is used to decompose the transmission protocol data of the target UAV into three feature subsets: control instruction field, status feedback field, and mission data field, including: Identifying a field type identifier in the protocol feature classification result, wherein the field type identifier includes a control instruction identifier, a status feedback identifier, and a task data identifier; generating extraction rules according to the numerical ranges of the control instruction identifier, the state feedback identifier, and the task data identifier, respectively, and combining the extraction rules to form a backup node rule; Traversing all field units of the transmission protocol data of the target UAV, if the identifier of the field unit falls within the numerical range of the control instruction identifier, assigning the field unit to the control instruction field set; if it falls within the numerical range of the status feedback identifier, assigning the field unit to the status feedback field set; if it falls within the numerical range of the mission data identifier, assigning the field unit to the mission data field set; The control instruction field set is encapsulated into an independent control instruction field, the status feedback field set is encapsulated into an independent status feedback field, and the task data field set is encapsulated into an independent task data field.

5. The method according to claim 1, wherein Calculating signal propagation angles between adjacent UAVs based on the spatial position data, generating distribution data of signal coverage angles based on the signal propagation angles, calculating path loss of electromagnetic reflection based on the spatial position data, and generating a correction factor for multipath effects based on the path loss, including: Based on the spatial position data of each drone in the drone group, calculate the straight-line distance between any two adjacent drones, and determine the signal propagation angle between the adjacent drones based on the difference between the straight-line distance and the altitude coordinate, where the signal propagation angle includes an elevation component and an azimuth component in the signal propagation direction; Mapping the elevation component and the azimuth component to preset angle intervals respectively to generate distribution data of signal coverage angles, wherein the distribution data includes statistical density of signal propagation directions within each angle interval; Based on the straight-line distance and the preset electromagnetic wave attenuation coefficient, the direct path loss of the electromagnetic wave between adjacent drones is calculated. At the same time, the additional loss of the reflection path is calculated based on the relative position of the spatial position data and the environmental reflective surface, combined with the material property parameters of the reflective surface; The direct path loss and the additional loss are added to obtain a total path loss, and a correction factor for the multipath effect is generated according to a ratio of the total path loss to a preset reference loss.

6. The method according to claim 1, characterized in that Constructing a polyhedron structure based on the spatial position data of the drone, determining the center position of actual communication coverage according to the geometric center of the polyhedron structure in combination with the preset communication coverage center offset, and calculating a three-dimensional offset vector pointing from the geometric center to the center position, including: Generate a polyhedron structure using the spatial position coordinates in the spatial position data as vertex coordinates, and calculate the arithmetic mean of all the vertex coordinates of the polyhedron structure as the geometric center; Obtaining a preset communication coverage center offset, wherein the communication coverage center offset includes a horizontal offset parameter and a vertical offset parameter, and determining a maximum coverage plane based on the spatial vertex distribution of the polyhedron structure; Determine a horizontal projection point on the maximum coverage plane according to the horizontal offset parameter, and move the horizontal projection point along the normal direction of the maximum coverage plane by the vertical offset parameter to determine the center position of actual communication coverage; The direction vector and the Euclidean distance between the two coordinate points pointing from the geometric center to the center position are calculated, the direction vector is normalized and multiplied by the Euclidean distance to generate a three-dimensional offset vector.

7. The method according to claim 2, characterized in that The electromagnetic interference intensity evaluation value, the node displacement deviation evaluation value, and the signal reflection path evaluation value are weighted and summed to obtain a comprehensive stability score, including: Obtaining preset weight ratios corresponding to the electromagnetic interference intensity evaluation value, the node displacement deviation evaluation value, and the signal reflection path evaluation value; Multiplying the electromagnetic interference intensity evaluation value, the node displacement deviation evaluation value, and the signal reflection path evaluation value by the corresponding weight ratio to generate a node displacement weighted result, and superimposing the electromagnetic interference weighted result, the node displacement weighted result, and the signal reflection weighted result to generate an initial comprehensive score; The initial comprehensive score is adjusted according to the preset proportional value storage data, and the electromagnetic interference weighted result, the node displacement weighted result and the signal reflection weighted result are re-superimposed to generate a comprehensive stability score.

8. An artificial intelligence-based UAV communication protocol feature classification system, characterized by: include: The acquisition module is used to obtain the spatial position data of each drone in the drone group and the preset communication coverage center offset; a generation module, configured to calculate a signal propagation angle between adjacent UAVs based on the spatial position data, generate distribution data of a signal coverage angle based on the signal propagation angle, calculate a path loss of electromagnetic reflection based on the spatial position data, and generate a correction factor for a multipath effect based on the path loss; a calculation module, configured to construct a polyhedron structure based on the spatial position data of the drone, determine the center position of actual communication coverage according to the geometric center of the polyhedron structure in combination with a preset communication coverage center offset, and calculate a three-dimensional offset vector pointing from the geometric center to the center position; an evaluation module, configured to input the three-dimensional offset vector, the distribution data of the signal coverage angle, and the correction factor of the multipath effect into a path planning neural network, perform a comprehensive evaluation of the electromagnetic interference intensity, the node displacement deviation, and the signal reflection path through the path planning neural network, and generate a protocol feature classification result; A decomposition module is used to formulate a backup node rule based on the protocol feature classification result, decompose the transmission protocol data of the target UAV into multiple feature subsets according to the backup node rule, and distribute the multiple feature subsets to selected relay nodes, transmit them in parallel through multiple paths, and finally reassemble them at the receiving end.

9. An electronic device, characterized in that: include: memory for storing computer programs; A processor, configured to implement the steps of the artificial intelligence-based drone communication protocol feature classification method as described in any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, which, when executed by a processor, can implement the artificial intelligence-based drone communication protocol feature classification method according to any one of claims 1 to 7.

Images