Method and system for unmanned aerial vehicle autonomous flight patrol in public network communication area multi-machine nest cooperation

Abstract

The application provides a UAV autonomous flight patrol method and system for multiple machine nests in a public network communication area, and relates to the technical field of intelligent control. The method comprises the following steps: constructing a communication coverage model based on a patrol partition scheme and a machine nest resource state; generating a cooperative scheduling plan according to the communication coverage model; performing three-dimensional track optimization calculation by each machine nest according to the cooperative scheduling plan to obtain flight trajectory data; processing the flight trajectory data to obtain a three-dimensional environmental situation data set; generating a cooperative control instruction sequence based on the three-dimensional environmental situation data set, wherein the cooperative control instruction sequence comprises a track optimization strategy and a task reallocation strategy; and performing distributed control according to the cooperative control instruction sequence. The application can dynamically evaluate the communication coverage capability of each machine nest and generate an optimal cooperative scheduling plan.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent control, in particular to a method and system for autonomous flight patrol of unmanned aerial vehicles (UAVs) in a multi-UAV nest coordination area without public network communication. BACKGROUND

[0002] In the field of power inspection, autonomous flight patrol technology of UAVs has been widely applied. However, existing technologies still face severe challenges in the multi-UAV nest coordination in areas without public network communication. Traditional UAV inspection systems usually rely on public network communication for data transmission and coordination control. However, in remote mountainous areas, public network signal coverage is insufficient or even completely missing, which leads to the inability of UAV clusters to achieve effective coordination.

[0003] Currently, multi-UAV cooperative path planning technology mainly focuses on path optimization in single-UAV nest scenarios. The core goal is to plan flight trajectories that meet the multi-UAV coordination relationship and minimize the overall flight system cost under basic constraint conditions. However, when extended to multi-UAV nest distributed deployment scenarios, existing methods have difficulty in effectively solving the communication coverage blind area and task allocation conflict between nests. In complex terrain environments, the communication coverage range of each nest presents irregular distribution due to terrain obstruction and distance attenuation. Traditional uniform grid partitioning methods cannot adapt to terrain undulation characteristics, leading to insufficient or redundant coverage in some areas, which seriously affects the inspection efficiency and task completion rate.

[0004] In terms of path optimization, existing technologies usually use static environment models for offline planning, lacking real-time perception and response capability to dynamic environment situations. Especially in the absence of public network communication, nests may not be able to share environmental information and flight status in real time, and the paths calculated independently by each nest often have spatial conflicts or coverage gaps. SUMMARY

[0005] The present application provides a method and system for autonomous flight patrol of unmanned aerial vehicles (UAVs) in a multi-UAV nest coordination area without public network communication, which can dynamically evaluate the communication coverage capability of each nest and generate an optimal coordination scheduling plan.

[0006] To solve the above technical problems, the technical solutions of the present application are as follows:

[0007] In a first aspect, a method for autonomous flight patrol of unmanned aerial vehicles (UAVs) in a multi-UAV nest coordination area without public network communication is provided, the method comprising:

[0008] Step S1, by collecting the digital terrain data and nest distribution data of the target inspection area, discretizing the terrain data into continuous terrain grid cells with uniform spatial resolution, and calculating the coverage ability coefficient of each nest to the terrain grid based on the nest position and communication parameters, forming a nest-terrain grid coverage relationship matrix; from the nest-terrain grid coverage relationship matrix, the elevation change gradient of the terrain undulation feature and the nest coverage ability distribution data are extracted, the inspection area is dynamically divided into non-overlapping and complete coverage inspection sub-areas, and the final inspection zoning scheme is generated by dynamically adjusting the area size and shape complexity of each sub-area.

[0009] Step S2, based on the inspection zoning scheme and the nest resource state, a communication coverage model is constructed; according to the communication coverage model, a cooperative scheduling plan is generated;

[0010] Step S3, according to the cooperative scheduling plan, each nest respectively performs three-dimensional track optimization calculation to obtain flight trajectory data;

[0011] Step S4, processing the flight trajectory data to obtain a three-dimensional environmental situation data set;

[0012] Step S5, based on the three-dimensional environmental situation data set, a cooperative control instruction sequence is generated, which includes trajectory optimization strategy and task reallocation strategy;

[0013] Step S6, according to the cooperative control instruction sequence, distributed control is performed.

[0014] The second aspect is a multi-nest cooperative unmanned aerial vehicle autonomous flight inspection system in a non-public network communication area, comprising:

[0015] The processing module is configured to discretize the terrain data into continuous terrain grid cells with uniform spatial resolution, and calculate the coverage ability coefficient of each nest to the terrain grid based on the nest position and communication parameters, forming a nest-terrain grid coverage relationship matrix; from the nest-terrain grid coverage relationship matrix, the elevation change gradient of the terrain undulation feature and the nest coverage ability distribution data are extracted, the inspection area is dynamically divided into non-overlapping and complete coverage inspection sub-areas, and the final inspection zoning scheme is generated by dynamically adjusting the area size and shape complexity of each sub-area;

[0016] The construction module is configured to construct a communication coverage model based on the inspection zoning scheme and the nest resource state; according to the communication coverage model, a cooperative scheduling plan is generated; according to the cooperative scheduling plan, each nest respectively performs three-dimensional track optimization calculation to obtain flight trajectory data;

[0017] The control module processes flight trajectory data to obtain a three-dimensional environmental situation dataset; based on the three-dimensional environmental situation dataset, it generates a cooperative control command sequence, which includes a trajectory optimization strategy and a task reassignment strategy; and executes distributed control according to the cooperative control command sequence.

[0018] The above-described solution of the present invention has at least the following beneficial effects:

[0019] By constructing a precise communication coverage model, the problem of communication blind spots between racks in areas without public networks was solved. Based on the inspection zoning scheme and the real-time resource status of racks, this model can dynamically evaluate the communication coverage capability of each rack, generate the optimal collaborative scheduling plan, effectively avoid task failures caused by communication interruptions, and reduce equipment idle rate.

[0020] By employing distributed 3D trajectory optimization calculation, each pod independently generates its flight trajectory based on the collaborative scheduling plan. This ensures both computational efficiency and global optimization. The distributed computing architecture avoids the risk of single-point failures and enhances the robustness of the system.

[0021] By processing flight trajectory data to construct a three-dimensional environmental situation dataset, real-time dynamic perception of the inspection area was achieved. This dataset integrates observation information from multiple aircraft nests, forming a unified environmental digital twin model, which improves the accuracy and timeliness of anomaly detection. It also enables efficient collaboration among multiple aircraft nests in an environment without a public network, thereby improving the task completion rate. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating the autonomous flight patrol method for unmanned aerial vehicles (UAVs) with multi-nest collaboration in a public network communication zone provided by an embodiment of the present invention.

[0023] Figure 2 This is a schematic diagram of an autonomous drone patrol system with multi-nest collaboration in a public network communication zone provided by an embodiment of the present invention. Detailed Implementation

[0024] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0025] like Figure 1 As shown, embodiments of the present invention propose an autonomous flight patrol method for unmanned aerial vehicles (UAVs) with multi-nest collaboration in areas without public network communication. The method includes the following steps:

[0026] Step S1, by collecting the digital terrain data and nest distribution data of the target inspection area, discretizing the terrain data into continuous terrain grid cells with uniform spatial resolution, and calculating the coverage ability coefficient of each nest to the terrain grid based on the nest position and communication parameters, forming a nest-terrain grid coverage relationship matrix; extracting the elevation change gradient of the terrain relief feature and the nest coverage ability distribution data from the nest-terrain grid coverage relationship matrix, dynamically dividing the inspection area into non-overlapping and complete inspection sub-areas, and generating the final inspection zoning scheme by dynamically adjusting the area size and shape complexity of each sub-area;

[0027] Step S2, based on the inspection zoning scheme and the nest resource state, constructing a communication coverage model; according to the communication coverage model, generating a cooperative scheduling plan;

[0028] Step S3, according to the cooperative scheduling plan, each nest respectively performs three-dimensional trajectory optimization calculation to obtain flight trajectory data;

[0029] Step S4, processing the flight trajectory data to obtain a three-dimensional environmental situation data set;

[0030] Step S5, based on the three-dimensional environmental situation data set, generating a cooperative control instruction sequence, which includes trajectory optimization strategy and task reallocation strategy;

[0031] Step S6, according to the cooperative control instruction sequence, executing distributed control.

[0032] In this embodiment, by constructing the nest-terrain grid coverage relationship matrix, the coverage ability of each nest to complex terrain is accurately calculated, effectively solving the communication blind area problem in the area without public network communication; based on the dynamic partition of the elevation change gradient and the coverage ability distribution, it is ensured that each inspection sub-area has reliable communication guarantee, and the communication interruption risk is reduced; through the coverage ability coefficient calculation and dynamic area segmentation, the scientific allocation of nest resources is realized, and the waste of resources and uneven load are avoided; dynamically adjusting the size and shape complexity of the sub-area, making the inspection task amount of each nest more balanced, and improving the overall system throughput; the terrain data is discretized into continuous grids with uniform spatial resolution, fully considering the terrain undulation characteristics, and ensuring full coverage inspection under complex terrain; each nest performs three-dimensional flight path optimization calculation, which can accurately adapt to terrain changes, improve inspection accuracy and data collection quality; a distributed control mode is executed, avoiding single point failure risk, even if part of the nest fails, the system can still maintain basic functions; the cooperative control instruction sequence contains task reassignment strategy, which can respond to sudden conditions (such as nest failure, weather change, etc.) in real time, and ensure the continuity and integrity of the inspection task; by processing flight trajectory data to generate three-dimensional environment situation data set, real-time and accurate environment information is provided for cooperative decision-making; the combination of trajectory optimization strategy and task reassignment strategy realizes the intelligent cooperation between multiple nests, and improves the adaptive ability of the overall system; the whole process automation from area segmentation, flight path planning to cooperative control greatly reduces the demand for manual intervention; through accurate coverage ability calculation and flight path optimization, unnecessary flight distance and hovering time of the unmanned aerial vehicle are reduced, and the battery endurance is prolonged; according to the real-time environmental change and task demand, the inspection strategy and area allocation can be quickly adjusted, and through the distributed control architecture, the communication delay of the central control node is reduced, and the overall response speed of the system is improved.

[0033] In a preferred embodiment of the present application, step S1 comprises:

[0034] In step S100, digital terrain data of a target inspection area is collected, and nest distribution data of each nest is obtained, the nest distribution data including geographical position coordinates of each nest, performance parameters of communication equipment, and number state information of schedulable unmanned aerial vehicles, specifically including: in a public network communication environment, each nest preferentially adopts an industrial-grade wireless communication protocol combining LoRa and ZigBee to construct a local communication network, and is assisted by industrial-grade WiFi to realize short-distance high-speed data interaction, forming a dual-link architecture of long-distance low-power consumption and short-distance high rate. The LoRa protocol works in the 433MHz industrial unlicensed frequency band, and the anti-terrain shielding capability is improved through spread spectrum technology, the communication distance covers 1 to 3 kilometers, and the long-distance data transmission between nests in remote mountainous areas is met; the ZigBee protocol adopts a star topology to access the slave devices of each nest, and realizes low-power consumption real-time reporting of the device state; the industrial-grade WiFi is used for high-definition data fast interaction between local devices of the nest. Relying on the local communication network, the laser radar equipment and high-definition camera carried by the unmanned aerial vehicle perform omnidirectional scanning and collection on the target inspection area, and obtain the digital terrain data of the area, which includes the three-dimensional space coordinates, elevation information, terrain slope value, ground attachment type and distribution of each point in the inspection area. At the same time, each nest interacts global data through the LoRa link, collects local device information through the ZigBee link, and extracts the nest distribution data by the deployed local data collection module, wherein the geographical position coordinates are obtained by real-time positioning of the global positioning system, and are accurate to latitude, longitude and altitude; the performance parameters of the communication equipment are specially matched with the above protocol characteristics, including the transmission power, receiving sensitivity and spread spectrum factor of the LoRa module, the network address and communication rate of the ZigBee module, and the frequency band bandwidth and access capacity of the industrial-grade WiFi; the number state information of the schedulable unmanned aerial vehicle covers the number of unmanned aerial vehicles in the standby state, the charging state and the maintenance state of each nest, the remaining battery capacity and the expected task execution time of each state unmanned aerial vehicle, and the related state information is reported to the nest master unit in real time through the ZigBee link, and then synchronized to other nests through the LoRa link, to ensure real-time sharing of data.

[0035] In step S101, the digital terrain data is discretized into a set of continuous terrain grid units with uniform spatial resolution; based on the geographical position coordinates and the performance parameters of the communication equipment in the nest distribution data, the communication coverage distance weight and the coverage ability coefficient of each nest to each terrain grid unit under the influence of terrain shielding are calculated, and a nest-terrain grid unit coverage relationship matrix is formed, specifically including:

[0036] Based on the digital terrain data and the nest distribution data obtained in step S100, first, the digital terrain data is discretized. In combination with the precision standard required by the inspection task and the detection capability of the unmanned aerial vehicle-mounted equipment, a uniform spatial resolution is set, and the length of a single terrain grid cell is usually set to a fixed value of 10-20 meters, so as to ensure that the grid cell can accurately reflect the terrain details and will not cause data redundancy. According to the spatial resolution, the entire target inspection area is divided into a continuous and non-overlapping set of terrain grid cells, each grid cell is assigned a unique identification number, and the three-dimensional spatial coordinate elevation information, terrain slope value and surface attached object type of the center point of each grid cell are extracted, so as to complete the discretization conversion of the terrain data.

[0037] Then, the coverage capability coefficient of each nest to each terrain grid cell is calculated. First, the calculation basis of the communication distance weight is determined, and the maximum communication distance of the LoRa module in the nest communication equipment performance parameter is taken as the reference value. The straight-line spatial distance between the nest geographic location coordinates provided by the global positioning system and the grid cell center point coordinates is calculated. When the straight-line distance is less than or equal to the maximum communication distance of the LoRa module, the distance weight value decreases linearly with the increase of the distance, and when the distance is 0, the weight value is 1, and when the distance reaches the maximum communication distance, the weight value is 0.1; when the straight-line distance exceeds the maximum communication distance, the distance weight value is directly set to 0, indicating that the grid cell is beyond the basic communication range of the nest.

[0038] Then, the terrain shielding influence coefficient is calculated, and the ray tracing method is used to simulate the communication signal propagation path between the nest and the grid cell. A virtual ray is emitted from the nest position to the center point of the grid cell, and the elevation information of all terrain grid cells through which the ray passes is checked one by one. If the ray is not blocked by the elevation of other terrain grid cells throughout the journey, i.e. the elevation of all points on the ray trajectory is lower than the elevation of the line connecting the nest and the target grid cell, the terrain shielding influence coefficient is 1; if there is shielding, the number of grid cells in the shielding area and the difference between the highest shielding elevation and the line elevation are counted, the more the number of shielding grid cells and the greater the difference, the smaller the shielding influence coefficient, which can be as low as 0.05.

[0039] The communication distance weight and the terrain shielding influence coefficient of each nest to the corresponding grid cell are multiplied to obtain the coverage capability coefficient of the nest to the grid cell. Taking the nest identification as the row index and the terrain grid cell identification as the column index, the coverage capability coefficients of all nests to all grid cells are sequentially filled in to form a complete nest-terrain grid cell coverage relationship matrix, and each value in the matrix represents the actual coverage capability of the corresponding nest to the corresponding grid cell.

[0040] Step S102, based on the nest-terrain grid cell coverage relationship matrix, extracting the elevation change gradient data reflecting the terrain undulation characteristics and the effective coverage capacity coefficient distribution data reflecting the nest coverage capacity; according to the elevation change gradient data and the effective coverage capacity coefficient distribution data, dynamically segmenting the target inspection area into multiple non-overlapping and complete coverage inspection sub-areas, specifically including: based on the nest-terrain grid cell coverage relationship matrix generated in step S101, first extracting the elevation change gradient data. For the elevation information of each terrain grid cell associated in the matrix, the elevation difference between the grid cell and the adjacent four direction (upper left lower right) grid cells is calculated, and the elevation difference of each direction is divided by the side length of the grid cell to obtain the elevation change rate of the direction. Take the maximum value of the four direction elevation change rate as the elevation change gradient value of the current grid cell, and record the change direction corresponding to the gradient value, so as to reflect the terrain undulation degree of the position of the grid cell. The elevation change gradient values and corresponding directions of all grid cells are sorted to form the elevation change gradient data set.

[0041] Then extract the effective coverage capacity coefficient distribution data. For each terrain grid cell in the matrix, compare all the coverage capacity coefficients of the nests to it, and select the largest coverage capacity coefficient as the effective coverage capacity coefficient of the grid cell, and mark the nest identifier corresponding to the coefficient to determine the main covering nest of each grid cell. The effective coverage capacity coefficients and corresponding main covering nest identifiers of all grid cells are summarized to form the effective coverage capacity coefficient distribution data set.

[0042] According to the above two types of data, the dynamic segmentation of the inspection area is carried out. First, according to the elevation change gradient data, the area with gradient value lower than the set threshold is divided into terrain flat area, and the area with gradient value higher than or equal to the threshold is divided into terrain complex area, realizing the preliminary classification of the area. In the terrain flat area, according to the effective coverage capacity coefficient distribution data, adjacent grid cells with the same main covering nest identifier are grouped to form the initial sub-area; in the terrain complex area, considering the influence of terrain on communication and inspection, the aggregation range of each group of grid cells is reduced, and grouping is also carried out according to the main covering nest identifier, ensuring that the grid cells in each group can be stably covered by the corresponding nest. In the grouping process, the boundaries of adjacent sub-areas are checked in real time to ensure that the sub-areas are non-overlapping and all grid cells are included in the sub-area range, realizing the complete segmentation of the inspection area and forming multiple initial inspection sub-areas.

[0043] Step S103, dynamically adjust the area size and shape complexity of the inspection sub-regions, and generate an inspection zoning scheme, specifically including: based on the initial inspection sub-regions formed in step S102, first, statistics the core parameters of each sub-region. Calculate the total number of terrain grid cells contained in each sub-region, and derive the actual area of the sub-region in combination with the area of each grid cell; calculate the average value according to the elevation change gradient value of each grid cell in the sub-region, reflecting the overall terrain complexity of the sub-region; determine the assignable unmanned aerial vehicle resources of the sub-region in combination with the number of schedulable unmanned aerial vehicles corresponding to the main coverage nest in the nest distribution data and the task execution capability of each unmanned aerial vehicle.

[0044] Based on the terrain complexity of the sub-region, in combination with the assignable unmanned aerial vehicle resources, a task load evaluation standard is established. The sub-region with high terrain complexity requires longer inspection time per unit area, and the corresponding task load is higher; the sub-region with flat terrain has high inspection efficiency per unit area, and the task load is relatively low. According to the standard, the task load value of each sub-region is calculated, and the load difference of all sub-regions is compared. If the ratio of the maximum load value to the minimum load value exceeds the set balance threshold, the sub-region adjustment process is started.

[0045] For the sub-region with too high task load, find its boundary with the adjacent sub-region, preferentially select the grid cells with small elevation change gradient value and effective coverage ability coefficient close to the main coverage nest coefficient of the adjacent sub-region, and divide these grid cells to the adjacent sub-region with lower load. During the splitting process, ensure that the shape of the split sub-region remains continuous to avoid scattered grids. For the sub-region with too low task load, try to merge the adjacent small sub-regions or absorb the grid cells split from the surrounding sub-region with too high load, and expand its own area.

[0046] During the adjustment process, the coverage integrity of each sub-region is monitored in real time to ensure that all grid cells in the adjusted sub-region can be effectively covered by the corresponding main coverage nest, and at the same time, the sub-regions do not overlap each other. After completing the preliminary adjustment, the task load value of each sub-region is recalculated, and the load balance is checked to see if it meets the requirements. If it does not meet the requirements, repeat the above adjustment steps until the task load difference of all sub-regions is controlled within the balance threshold. The area size of each sub-region finally formed matches the terrain complexity and unmanned aerial vehicle resources, the shape complexity is moderate and convenient for unmanned aerial vehicles to perform inspection tasks, thereby generating the final inspection zoning scheme.

[0047] In a preferred embodiment of the present application, step S2, based on the inspection zoning scheme and the nest resource state, a communication coverage model is constructed, including:

[0048] Step S200, based on the inspection zoning scheme and the nest resource state, calculate the theoretical communication coverage radius of each nest in the three-dimensional space, determine the actual communication coverage range of each nest under the influence of considering the terrain shelter, specifically including: based on the inspection zoning scheme generated in step S1 and the communication equipment performance parameters in the nest resource state, first calculate the theoretical communication coverage radius of each nest. With the LoRa module carried by the nest as the core communication equipment, combined with its transmission power, receiving sensitivity and spread factor parameters, reference the wireless communication propagation model, first determine the basic communication distance in the ideal open environment without terrain shelter. Then combine the maximum flight height and the minimum flight height of the unmanned aerial vehicle in the nest resource state, decompose the basic communication distance into three-dimensional space, calculate the theoretical communication coverage radius of the nest at different vertical height levels, form a three-dimensional theoretical coverage sphere data centered on the nest.

[0049] Then determine the actual communication coverage range of each nest considering the terrain shelter. Call the nest-terrain grid unit coverage relationship matrix generated in step S101, extract the coverage ability coefficient of all terrain grid units corresponding to the current nest. Mark the grid units with a coverage ability coefficient greater than 0 as potential coverage areas, and combine the three-dimensional space coordinates of these grid units to construct the initial space area of the nest communication coverage. Then use the three-dimensional ray tracing method, from the nest position, emit virtual signal rays in all directions within the initial space area, and the propagation path of the rays strictly follows the elevation data of the terrain grid unit. For the directions blocked by the terrain grid unit, truncate the coverage range beyond the shelter point in that direction; for the unblocked directions, determine the effective coverage range with the theoretical communication coverage radius as the boundary. Integrate the effective coverage boundaries of all directions to finally form the actual communication coverage range of each nest in the three-dimensional space, which must completely contain all terrain grid units within the corresponding inspection sub-area.

[0050] Step S201, based on the actual communication coverage range of each nest, model the communication coverage area of each nest as a conical model with the nest geographic location coordinates as the vertex, the actual communication coverage radius as the generatrix length, and the terrain relief angle as the opening angle, specifically including: based on the actual communication coverage range of each nest determined in step S200, first determine the core parameters of the conical model. Take the geographic location coordinates of the nest as the vertex of the conical model, which contains accurate latitude and longitude and altitude information, ensuring the accuracy of the model positioning. Take the actual communication coverage radius calculated in step S200 as the generatrix length of the conical model, and the extension direction of the generatrix covers all spatial directions within the actual communication coverage range of the nest.

[0051] Then the opening angle of the cone is determined, which needs to be set in combination with the terrain fluctuation angle of the corresponding patrol sub-region. The elevation change gradient data of all terrain grid cells in the corresponding patrol sub-region obtained in step S102 is extracted, and the maximum angle of terrain fluctuation in the sub-region, i.e. the maximum value of terrain slope, is calculated. Based on the maximum terrain fluctuation angle, the opening angle of the cone is set to be 10 degrees larger than the maximum terrain fluctuation angle, ensuring that the cone model can completely contain the communication coverage requirement changes caused by terrain fluctuations, and avoiding the fact that part of the area is out of the model coverage range due to the terrain protrusion.

[0052] According to the above vertex generatrix length and opening angle parameters, the communication coverage cone model of each nest is constructed in three-dimensional space. The axis direction of the model can be fine-tuned according to the overall terrain trend of the sub-region, to ensure that the cone can best fit the actual communication coverage range, and each cone model needs to be associated with the identification information of the corresponding nest, so as to facilitate subsequent analysis of the coverage area interaction between nests.

[0053] In step S202, the intersection region volume between the corresponding cone models of the adjacent nests in three-dimensional space position is calculated, and when the intersection region volume is greater than a preset minimum communication relay volume threshold, the corresponding intersection region is determined as an effective communication relay region, which specifically includes: based on the communication coverage cone models of each nest constructed in step S201, first, the adjacent nest combinations in three-dimensional space position are screened out. By comparing the geographical position coordinates of each nest, the straight-line distance between the nests is calculated, and when the distance is less than or equal to the sum of the generatrix lengths of the two corresponding cone models, the two nests are determined to be adjacent in three-dimensional space position, and they are combined into an adjacent nest combination, to avoid invalid calculation of nests that are too far apart and have no intersection of coverage ranges.

[0054] For each adjacent nest group, the cone model parameters of each nest, including the vertex coordinate generatrix length and opening angle, are extracted, and the models are superimposed in a unified three-dimensional coordinate system; a layered calculation method is used to solve the intersection region volume between the two cone models. The space covered by the two cones is divided into multiple equidistant thin layers along the vertical height direction, and the height interval of each thin layer is set to 1 meter, to ensure the calculation accuracy.

[0055] After determining the cone models of the adjacent nests, the three-dimensional space covered by the two cones is first divided into several parallel horizontal thin layers along the vertical height direction at a fixed interval of 1 meter, each thin layer is a flat cuboid with uniform thickness, and the total height of all thin layers covers the complete height range where the two cones may intersect. For each thin layer, the volume is calculated according to the following specific steps:

[0056] Determine the three-dimensional spatial position of the current processing thin layer, record the elevation of the upper surface and the lower surface of the thin layer, take the average of the two heights as the intermediate plane height of the thin layer, and all subsequent cross section calculations are based on the intermediate plane. At the same time, the basic parameters of the two cones are called out, including the respective vertex coordinates (machine nest longitude and latitude and elevation), generatrix length, opening angle and axis direction in three-dimensional space, ensuring that all parameters correspond to the same three-dimensional coordinate system, avoiding calculation errors caused by coordinate deviation.

[0057] For the first cone, first calculate the vertical distance from its vertex to the intermediate plane of the thin layer. Take the vertex as the starting point and draw a perpendicular line to the intermediate plane of the thin layer. The length of the perpendicular line is the vertical distance. Combined with the opening angle of the cone, the cross section boundary of the cone on the thin layer plane is determined through geometric relationship: if the vertical distance is less than the generatrix length of the cone, and the thin layer plane is completely within the coverage range of the cone, the cross section is a complete circle, and the radius of the circle is determined by the vertical distance and the opening angle. The smaller the vertical distance and the larger the opening angle, the larger the cross section radius. If the thin layer plane partially exceeds the coverage range of the cone, the cross section is a sector. At this time, the radius of the complete circle needs to be determined first, and then the central angle of the sector is calculated according to the actual coverage angle of the cone on the plane. The central angle ranges from 0 degrees to 360 degrees.

[0058] The same method is used to process the second cone. The vertical distance from its vertex to the intermediate plane of the thin layer is calculated, and it is determined whether the cross section is a circle or a sector, and the radius of the circle or the sector and the central angle are determined. At the same time, the center coordinates of the two cross sections on the thin layer plane (i.e. the intersection of the cone axis and the thin layer plane) are recorded to ensure accurate position correspondence for subsequent intersection calculation.

[0059] First, determine the positional relationship of the two cross sections. By calculating the distance between the centers of the two cross sections, comparing it with the sum and difference of the two radii, it is preliminarily determined whether there is an intersection area. If the distance between the centers is greater than the sum of the two radii, or one cross section is completely outside the other, the intersection area is 0. If there is a possibility of intersection, the two cross sections are processed according to their types (circle and circle, circle and sector, sector and sector):

[0060] Circle and circle intersection: determine the position of the common chord formed by the intersection of the two circles based on the line connecting the two centers, and then divide the intersection area into a lens shape (surrounded by two circular arcs) by the chord length and the radii of the two circles. The area of the intersection region is obtained by calculating the area of each part and summing them up.

[0061] Circle and sector intersection: first determine the straight line where the two radii of the sector are located, judge the position relationship between the two straight lines and the circle, find out the intersection points of the boundary of the circle and the sector, combine the central angle range of the sector, determine the boundary of the intersection area composed of the arc of the circle and the radius or arc of the sector, and then disassemble the intersection area into regular geometric figures (such as triangle, sector, arc), calculate the area respectively and add up.

[0062] Sector and sector intersection: first determine the central angle range and radius length of the two sectors respectively, find out the intersection points of the boundary radii of the two sectors, determine the outline of the intersection area composed of the arcs or radii of the two sectors, and then according to the shape characteristics of the intersection area, divide it into triangle and small sector and other basic figures, and get the total area of the intersection area through the area calculation of each basic figure.

[0063] Multiply the intersection area obtained by the thickness of the thin layer (fixed as 1 meter), and the product is the volume of the intersection area corresponding to the current thin layer. In the calculation process, a separate record item is established for each thin layer, and the height range, intersection area and volume value are marked to avoid missing calculation or repeated calculation; according to the order from bottom to top or from top to bottom, the above operation is performed on all thin layers in turn, and after the intersection volume of each thin layer is obtained, the volume values of all thin layers are continuously added, and the cumulative result is the total intersection volume between the two adjacent nest cone models.

[0064] According to the minimum bandwidth requirement and communication stability requirement of unmanned aerial vehicle data transmission, set the minimum communication relay volume threshold value. Compare the calculated intersection volume with the threshold value, when the intersection volume is greater than the threshold value, it means that the intersection area can provide stable signal relay space for two unmanned aerial vehicles, and it is determined as an effective communication relay area, and the two nest identifiers corresponding to the area and the three-dimensional space coordinate range of the area are recorded; when the intersection volume is less than or equal to the threshold value, it is determined as an invalid relay area, which is not included in the subsequent communication model construction.

[0065] Step S203, divide each conical model communication coverage space into multiple horizontal communication layers along the vertical height direction; according to the signal strength attenuation degree and data transmission bandwidth demand in the height interval of each horizontal communication layer, assign a corresponding communication priority level and dynamic bandwidth allocation strategy to each horizontal communication layer, which specifically includes: based on the conical model constructed in step S201, first divide the communication coverage space of each conical along the vertical height direction; refer to the commonly used flight height range of the unmanned aerial vehicle in the inspection task, combine the average elevation of the terrain, divide the conical from the bottom to the top into multiple horizontal communication layers; usually take 50 meters as a height interval, for the area with large terrain undulations, the height interval can be reduced to 20 meters, ensure that the terrain environment and signal propagation conditions in each horizontal communication layer are relatively stable, assign a unique number to each horizontal communication layer, and record the corresponding vertical height interval of each layer.

[0066] Then test the signal strength attenuation of each horizontal communication layer, through the signal receiving equipment carried by the unmanned aerial vehicle, collect the communication signal strength data of the corresponding nest at multiple evenly distributed points in the height interval of the layer, calculate the average value and attenuation amplitude of the signal strength in the layer; at the same time, combine the corresponding inspection task demand of the layer to determine the data transmission bandwidth requirement, for example, the layer used for high-definition image transmission needs higher bandwidth, and the layer used for state information reporting needs lower bandwidth.

[0067] According to the signal strength attenuation degree and bandwidth demand, assign a communication priority level to each horizontal communication layer, adopt a grading standard of 1 to 5 levels, and level 1 is the highest priority; for the horizontal communication layer with small signal strength attenuation and high bandwidth demand, assign a high priority of level 1 to 2; for the layer with medium signal strength attenuation and medium bandwidth demand, assign a medium priority of level 3; for the layer with large signal strength attenuation or low bandwidth demand, assign a low priority of level 4 to 5.

[0068] Formulate a dynamic bandwidth allocation strategy for different priority levels, the horizontal communication layer with high priority adopts the strategy of priority guaranteeing bandwidth, reserves a fixed proportion of bandwidth resources to ensure the real-time and stability of data transmission, and the remaining bandwidth is dynamically adjusted according to the real-time data volume; the medium priority layer adopts the strategy of bandwidth allocation on demand, allocates bandwidth resources according to the real-time data transmission demand of the current layer, and can appropriately compress when resources are tight but needs to ensure basic transmission demand; the low priority layer adopts the strategy of elastic bandwidth allocation, uses the remaining bandwidth for data transmission on the basis of meeting the demand of high priority and medium priority layers, and can suspend non-critical data transmission when the bandwidth is insufficient; associate and store the height interval priority level and dynamic bandwidth allocation strategy of each horizontal communication layer, provide basic data for subsequent communication coverage model construction.

[0069] Step S204, based on the effective communication relay area, communication priority level and dynamic bandwidth allocation strategy, a communication coverage model is constructed, specifically including: based on the effective communication relay area determined in step S202 and the communication priority level dynamic bandwidth allocation strategy obtained in step S203, first, the basic framework of the communication coverage model is constructed; taking the three-dimensional space coordinate system as the carrier, the communication coverage cone body model of each nest is superimposed in the coordinate system according to the actual position, and the identification information and the corresponding inspection sub-area range of each nest are marked on the model, and the spatial distribution of the overall communication coverage is clearly presented.

[0070] All effective communication relay areas are marked in the three-dimensional coordinate system, and the relay areas corresponding to different adjacent nest combinations are distinguished by visual identification, and the volume size and three-dimensional coordinate boundary of each relay area are marked. At the same time, the association between the relay area and the corresponding nest is established, and it is clear that each relay area can serve the nest pair, which provides a basis for the subsequent communication relay switching between unmanned aerial vehicles; the priority level and dynamic bandwidth allocation strategy of each horizontal communication layer are integrated into the corresponding cone body model, and the priority level of each cone body at different height intervals is marked, and the dynamic bandwidth allocation strategy is converted into executable rules and embedded into the model. For example, in the high-priority communication layer, the rule of ensuring 50Mbps bandwidth and dynamically allocating the remaining bandwidth is marked, so that the model can directly guide the actual communication resource scheduling.

[0071] Finally, when a certain unmanned aerial vehicle flies in a specific horizontal communication layer, the communication coverage model can automatically match the priority level and bandwidth allocation strategy of the layer, and judge whether the unmanned aerial vehicle enters the effective communication relay area according to the real-time position of the unmanned aerial vehicle. If it enters, the corresponding relay nest information is automatically associated; through this association mechanism, the communication coverage model can respond to the position change and communication demand of the unmanned aerial vehicle in real time, realize the dynamic optimization configuration of the communication resource, and finally form a complete communication coverage model.

[0072] In this embodiment, the actual coverage range is calculated by combining the nest resources and the terrain shielding, the signal transmission limitation in the public network environment is broken through, the communication blind area caused by terrain shielding is avoided, and stable communication foundation is provided for the unmanned aerial vehicle; by identifying the effective communication relay area, a redundant communication link between nests is constructed, even if the communication range of a single nest is limited, data interaction can be realized through the relay area, and the risk of communication interruption is greatly reduced; the communication priority and bandwidth strategy are allocated according to the height layering, the high-priority bandwidth is guaranteed for the key demand such as high-definition data transmission, and the elastic bandwidth allocation is adopted for the non-key data such as device state reporting, so as to avoid the waste of communication resources and improve the overall bandwidth utilization rate; the coverage area is modeled as a cone body, the nest communication capability is materialized, accurate basis is provided for resource matching in subsequent cooperative scheduling, and resource mismatching problem is reduced.

[0073] In a preferred embodiment of the present application, a cooperative scheduling plan is generated according to the communication coverage model, comprising:

[0074] In step S205, based on the distribution of effective communication relay areas in the communication coverage model, the communication connection relationship between each inspection sub-area is identified, and a communication topology network between the nests is constructed, specifically including: based on all the effective communication relay area distribution data recorded in the communication coverage model, first determine the two adjacent nest identifiers corresponding to each effective communication relay area, and the two nests responsible for the inspection sub-area. The main coverage nest corresponding to each inspection sub-area is taken as a network node, and the node information includes nest identifier, belonging to the inspection sub-area number, the number of schedulable unmanned aerial vehicles and the maximum bandwidth of communication equipment.

[0075] If the main coverage nests of two inspection sub-areas have a common effective communication relay area, and the volume of the relay area meets the unmanned aerial vehicle data interaction demand, it is determined that the inspection sub-areas corresponding to the two nests have a communication connection relationship; the node represents the nest, and the connection line segment represents the communication connection relationship, and an initial communication topology network is constructed. Add attribute labels to each connection line segment in the network, and the attribute content includes the three-dimensional coordinate range of the corresponding effective communication relay area, the signal transmission strength and the available bandwidth capacity. The signal transmission strength is obtained by inversely converting the average signal attenuation value in the relay area, and the available bandwidth capacity refers to the dynamic bandwidth allocation result of the corresponding horizontal communication layer in the communication coverage model. Finally, a communication topology network between nests is formed, which includes node attributes and connection attributes, ensuring that the network can intuitively reflect the communication accessibility and connection quality between each inspection sub-area.

[0076] In step S206, according to the connection strength and bandwidth capacity of each node in the communication topology network, the main control nest and the standby relay nest are allocated for each inspection sub-area, and a hierarchical cooperative control architecture is formed, specifically including: for each inspection sub-area, first extract the three core index basic data of all candidate main coverage nests. The coverage capacity coefficient directly uses the average value of the coverage capacity coefficient of the nest to all grid units in the sub-area in the nest-terrain grid coverage relationship matrix, and the numerical range is stable between 0 and 1. The unmanned aerial vehicle resource sufficiency is determined by the ratio of the number of on-call unmanned aerial vehicles of the nest to the estimated number of unmanned aerial vehicles required by the inspection sub-area. If the ratio is greater than 1, it is calculated as 1, and if it is less than 0.5, it is calculated as 0.5, to ensure that the value is in the reasonable interval of 0.5 to 1. The connection strength takes the average value of the communication connection strength of the nest and other potential cooperative nests in the sub-area. First, take the average of the signal transmission strength (0 to 1) and the effective communication relay area volume (normalized value scaled to 0 to 1), and then summarize the average values of all connections as the final connection strength, which is also controlled in the range of 0 to 1.

[0077] The three indicators are assigned equal weights, and each indicator accounts for one third. When calculating, the coverage capacity coefficient, the unmanned aerial vehicle resource sufficiency, and the connection strength are multiplied by one third respectively, and then the three products are added to obtain the comprehensive capability value of the nest. After completing the calculation of the comprehensive capability value of all candidate main coverage nests, the main coverage nest with the largest comprehensive capability value is determined as the main control nest of the inspection sub-region. The main control nest is responsible for the overall scheduling of the inspection task of the sub-region, including task decomposition, unmanned aerial vehicle assignment, data aggregation, and coordination with other nests.

[0078] For each inspection sub-region, all adjacent nodes (i.e., other nests with communication connections) of the corresponding main coverage nest in the communication topology network are selected. The main coverage nest is scored comprehensively based on the coverage capacity coefficient, the unmanned aerial vehicle resource sufficiency, and the connection strength. The higher the coverage capacity coefficient, the more sufficient the unmanned aerial vehicle resources, and the greater the connection strength, the higher the score. The main coverage nest with the highest score is determined as the main control nest of the inspection sub-region, which is responsible for the overall scheduling of the inspection task of the sub-region. From the remaining adjacent nodes, the nest with the largest connection bandwidth capacity and stable resource state is selected as the standby relay nest. Each inspection sub-region is configured with at least one standby relay nest, and if the number of adjacent nodes in the communication topology network is sufficient, two standby relay nests can be configured to improve redundancy.

[0079] The responsibilities of the main control nest and the standby relay nest are determined. The main control nest is responsible for generating inspection task instructions, receiving core data, and scheduling unmanned aerial vehicles in real time. The standby relay nest is responsible for relay transmission tasks when the main control nest communication is interrupted, and assists in sharing light inspection tasks if necessary. In this way, a hierarchical collaborative control architecture of main control and standby is formed. Each architecture unit is bound to the corresponding inspection sub-region and recorded in the system scheduling database.

[0080] In step S207, according to the collaborative control architecture, the number of unmanned aerial vehicles, the battery endurance time, and the task urgency in the resource state of the nest are combined to assign specific inspection task sequences and take-off time windows to each nest. Specifically, based on the constructed hierarchical collaborative control architecture, the task demands of all inspection sub-regions are first aggregated, and the task urgency is classified. The urgency is determined according to the importance of the inspection region (such as whether it is the region of a key facility) and the potential risk level (such as the stability of the terrain and the historical failure frequency), which is classified into three levels: urgent, regular, and low. The urgent level task is preferentially allocated resources.

[0081] The resource state data of each nest is called, the number of unmanned aerial vehicles in standby state in each nest, the remaining battery endurance time of each unmanned aerial vehicle and the executable task type are counted, the unmanned aerial vehicles are sorted according to the endurance time from long to short, and the unmanned aerial vehicles with long endurance time are preferentially allocated to the long-time inspection task. According to the area, the terrain complexity and the task urgency level of the inspection sub-region, the required number of unmanned aerial vehicles and the task execution time of a single unmanned aerial vehicle are estimated. The higher the terrain complexity, the larger the area, the more the required number of unmanned aerial vehicles, and the longer the task execution time. The emergency level task is compressed by 20% of the redundant time based on the estimated time, to ensure fast completion. The estimated task demand is allocated to the main control nest of the corresponding inspection sub-region, and if the number of standby unmanned aerial vehicles in the main control nest is insufficient, the standby relay nest will share the task gap in turn according to the connection strength. The specific inspection task sequence allocated to each nest is sorted according to the task urgency level, and the emergency task is arranged at the front, and the same level task is sorted according to the geographical direction of the inspection area from near to far, to avoid the time-consuming of unmanned aerial vehicles back and forth.

[0082] According to the execution time of each task in the task sequence and the preparation time of the unmanned aerial vehicle, a take-off time window is allocated to each task, and a 5-minute buffer period is reserved between the take-off time windows of adjacent tasks to avoid unmanned aerial vehicle scheduling conflicts. At the same time, the take-off time windows of the unmanned aerial vehicles in the same nest are ensured not to overlap, and the task take-off time of the standby relay nest is kept synchronous and associated with the main control nest.

[0083] Step S208, according to the inspection task sequence and the take-off time window, the time synchronization point of the cooperative work of the unmanned aerial vehicles between adjacent nests is calculated; according to the time synchronization point of the cooperative work of the unmanned aerial vehicles between adjacent nests, the terrain complexity and the communication signal stability, the inspection speed and the data transmission frequency of each nest are dynamically adjusted, and the cooperative scheduling plan is generated; the cooperative scheduling plan includes the task allocation scheme, the time scheduling table, the communication relay path and the abnormal handling plan, which specifically includes:

[0084] The inspection task sequence and the take-off time window data of each nest are extracted, and the inspection tasks involving the same effective communication relay area between adjacent nests are focused on. The execution time overlap interval of these tasks is calculated, and the midpoint of the overlap interval is taken as the time synchronization point of the cooperative work of the unmanned aerial vehicles between adjacent nests. If there is no direct overlap in the execution time, the time when the unmanned aerial vehicle executing the task first arrives at the relay area is taken as the synchronization reference point, and the take-off time window of the unmanned aerial vehicle executing the task later is adjusted to ensure that it arrives at the relay area within 10 minutes before and after the reference point to realize cooperation; the terrain complexity data of the area corresponding to each inspection task is collected, and the terrain complexity is determined by the average value of the elevation change gradient of the area. The larger the average value, the more complex the terrain; at the same time, the communication signal stability data of the task execution height interval is obtained through the communication coverage model, and the signal stability is measured by the signal strength fluctuation amplitude. The smaller the fluctuation amplitude, the higher the stability.

[0085] For the area with complex terrain and poor signal stability, the inspection speed of the unmanned aerial vehicle is reduced by 20% to 30%, and the data transmission frequency is increased from the default transmission once every 30 seconds to transmission once every 10 seconds, so as to ensure real-time data transmission and no omission of terrain details; for the area with flat terrain and stable signal, the inspection speed is appropriately increased, and the data transmission frequency is reduced to once every 60 seconds, prolonging the battery life of the unmanned aerial vehicle.

[0086] Integrate all the adjusted task information to generate a complete collaborative scheduling plan. The task allocation scheme specifies the specific task content, responsible area and unmanned aerial vehicle number of each nest; the time scheduling table marks the take-off time window, execution time and time synchronization point of each task; the communication relay path corresponds to a specific effective communication relay area, and the relay node and path order of data transmission are clear; the abnormal handling plan includes the response measures for the scenarios such as nest failure, unmanned aerial vehicle disconnection and signal interruption, such as taking over the task by the standby relay nest when the nest fails, starting the homing instruction when the unmanned aerial vehicle is disconnected, etc.

[0087] In this embodiment, the nest communication topology network constructed by the effective communication relay area forms an interconnected communication link for each inspection sub-area, and breaks away from the dependence on the public network; the clear standby relay nest mechanism enables quick switching of the relay path when the communication of a single nest is interrupted, minimizes the impact of communication failure on the task, and ensures the continuity of data transmission. The number of unmanned aerial vehicles, endurance and task urgency are combined to allocate tasks and take-off windows, avoiding idle or overloaded nest resources, allowing long-endurance unmanned aerial vehicles to undertake complex tasks first, and giving emergency tasks resource inclination, improving resource allocation accuracy; the time synchronization point coordinates the operation rhythm of adjacent nests, reduces the conflict of unmanned aerial vehicles in the air, and reduces the loss of invalid flight. The hierarchical collaborative control architecture clearly defines the responsibilities of the master and standby nests, avoids confusion of scheduling instructions, and improves task coordination efficiency; the dynamically adjusted inspection speed and data transmission frequency make the inspection of complex terrain areas more detailed and the operation of signal stable areas more efficient.

[0088] In a preferred embodiment of the present application, step S3, according to the collaborative scheduling plan, each nest performs three-dimensional trajectory optimization calculation to obtain flight trajectory data, including:

[0089] Step S300, based on the task allocation scheme and the time schedule table contained in the cooperative scheduling plan, extract the boundary coordinate set of each nest responsible for the inspection sub-region and the corresponding priority sequence, combine the performance parameters of the unmanned aerial vehicle in the nest resource state, determine the initial search space range of three-dimensional flight path optimization calculation, specifically including: first, from the task allocation scheme of the cooperative scheduling plan, locate all the inspection sub-regions corresponding to the current nest, extract the complete boundary coordinate set of each sub-region, these coordinates contain the latitude, longitude and altitude information of each point on the boundary, arrange them in clockwise order into a closed coordinate sequence, and clearly define the horizontal range and vertical boundary of the inspection. At the same time, extract the priority sequence of each sub-region inspection task from the time schedule table, mark the sub-region corresponding to the emergency task and the inspection route that needs to be completed first.

[0090] Then call the performance parameters of the unmanned aerial vehicle to be assigned in the nest resource state, focus on sorting the maximum flight altitude, minimum flight altitude, maximum range, climb rate and turning radius and other key data. Combined with the horizontal range determined by the sub-region boundary coordinates, take the intersection of the area covered by the maximum range of the unmanned aerial vehicle and the sub-region boundary range as the horizontal search space of the flight path optimization, taking the geographical position of the nest as the starting point.

[0091] According to the maximum and minimum flight altitude of the unmanned aerial vehicle, combined with the highest and lowest altitudes of the terrain in the sub-region, determine the vertical search space of the flight path optimization, ensure that the vertical range is within the performance range of the unmanned aerial vehicle, and is at least 20 meters higher than the highest terrain to avoid collision, while being lower than the maximum flight altitude of the unmanned aerial vehicle to ensure flight safety. For the emergency task area in the priority sequence, increase the resolution of its corresponding search space, reduce the search interval in the horizontal direction and the vertical direction to half of the regular area, forming the final three-dimensional flight path optimization initial search space range, which needs to completely contain all the inspection sub-regions responsible for the current nest.

[0092] Step S301, based on the initial search space range, the inspection sub-region is divided into a plurality of continuous triangular element grids in three-dimensional space, the vertex coordinates of each triangular element grid are determined by the sub-region boundary points and the terrain elevation data, a three-dimensional triangular mesh model covering the entire inspection sub-region is formed, specifically including: based on the initial search space range determined in step S300, first extract all boundary feature points of the inspection sub-region in the range, these feature points include the vertices in the sub-region boundary coordinate set, the boundary inflection points of the terrain mutation and the corner points of the initial search space range, the boundary feature point set is formed and the latitude and longitude information of each point is recorded. The boundary feature point set is processed by the convex hull algorithm, the feature points located at the outermost periphery of the sub-region boundary are selected by judging the positional relationship of each feature point relative to the connecting line of any two points, and the feature points are connected in a clockwise order to form a closed convex polygon, that is, the convex hull boundary of the inspection sub-region, so as to accurately define the core range of triangular partitioning and exclude the redundant area in the initial search space which does not belong to the inspection sub-region.

[0093] With the convex hull boundary as the reference, uniform grids are divided on the horizontal plane inside the convex hull at an interval of 50 meters, ensuring that all grid points fall within the convex hull boundary range, obtaining a series of horizontal grid points, and recording the latitude and longitude coordinates of each grid point; if the latitude and longitude of a grid point coincides with the convex hull boundary point, it is marked as a boundary grid point and subsequently participates in triangular partitioning preferentially; the digital terrain data obtained in step S1 is called, the elevation information corresponding to each horizontal grid point is extracted, and the latitude, longitude and elevation are combined to form three-dimensional space point coordinates, wherein the three-dimensional space point corresponding to the convex hull boundary point is used as a key boundary node for subsequent partitioning.

[0094] The triangular partitioning method is used to connect these three-dimensional space points into triangular element grids. In the specific operation, the three-dimensional space point corresponding to the convex hull boundary is taken as the basis, and the adjacent convex hull boundary points are connected in a clockwise order to form a boundary triangle chain around the sub-region, thereby constructing the peripheral framework of the triangular mesh. Subsequently, the boundary triangle chain is taken as the starting point, and triangular elements are gradually filled into the sub-region, and in the filling process, the principle of adjacent triangular elements sharing a side is followed to ensure that all triangular elements have no overlap and no gap, and the three vertices of each triangular element are three-dimensional space point coordinates that have been obtained; in the filling process, if it is detected that the terrain elevation difference between a certain horizontal grid point and the adjacent grid point exceeds the preset threshold, it is determined that the terrain elevation is mutated. At this time, the horizontal grid points within a 20-meter range around the mutation point are densified, smaller three-dimensional space points are regenerated, and triangular partitioning is supplemented based on these densified points to form smaller triangular elements, so as to accurately fit the terrain undulation details and avoid distortion of terrain features caused by excessively large grid elements.

[0095] After all the triangulation is completed, a range check is performed to check whether the three vertices of each triangular element all fall within the convex hull boundary and are within the initial search space range, and triangular elements whose vertices are outside the range are removed; for the local area in the overlapping area of the convex hull boundary and the initial search space which is not covered, small triangular elements are supplemented to fill in, and finally a three-dimensional triangular mesh model covering the entire inspection sub-region is formed, which not only guarantees the range accuracy through the convex hull algorithm, but also restores the terrain features through the encryption processing.

[0096] In step S302, based on the three-dimensional triangular mesh model, the geometric center point coordinates of each triangular element grid are taken as the ideal hovering observation position of the unmanned aerial vehicle; according to the communication quality level of the corresponding ideal hovering observation position in the communication coverage model, the flight height weight parameter of the geometric center point is dynamically adjusted to generate a preliminary set of track points, which specifically includes: for each triangular element grid in the three-dimensional triangular mesh model generated in step S301, the geometric center point coordinates are calculated, and the calculation method is to take the average of the three vertex coordinates to obtain the latitude and longitude and initial elevation of the center point. This center point is directly taken as the ideal hovering observation position of the unmanned aerial vehicle, and the unmanned aerial vehicle can completely observe the entire triangular element area at this position.

[0097] The communication coverage model is called, and according to the three-dimensional coordinates of the ideal hovering observation position, the horizontal communication layer where it is located is located, and the corresponding communication quality level of the layer is queried. The communication quality level is divided into four levels of excellent, good, general and poor, and the level division is based on signal strength attenuation degree and data transmission stability. For different communication quality levels, the flight height weight parameter of the ideal hovering observation position is dynamically adjusted: for the position with excellent communication quality, the height weight is set to 1, and the initial elevation is kept unchanged; for the position with good communication quality, the height weight is set to 1.1, and the initial elevation is increased by 10% to further optimize the signal; for the position with general communication quality, the height weight is set to 0.9, and the elevation is reduced by 10% to approach the terrain to ensure communication; for the position with poor communication quality, the height weight is set to 0.8, and the elevation is greatly reduced while recording the position, which needs to be optimized later.

[0098] The center point coordinates corresponding to the adjusted elevation are arranged in the priority order of the inspection sub-region, the points outside the initial search space range after height adjustment are removed, the replacement points with qualified communication quality in the periphery are supplemented, and finally a preliminary set of track points containing all the effective hovering positions is generated, and each track point is marked with the corresponding triangular element number and communication quality level.

[0099] Step S303, based on the preliminary set of track points, the connection order between adjacent track points is optimized and sorted, and the flight path segment is generated; spare observation positions are added near the key track points of the flight path segment, and finally the flight trajectory data containing the complete flight trajectory coordinate sequence, flight speed parameters and observation time length configuration are obtained, which specifically includes: based on the preliminary set of track points generated in step S302, first sort the patrol sub-regions according to the priority from high to low, and sort the track points in the same priority according to the geographical direction from the nest to the direction away from the nest, to ensure that the flight path reduces the round trip distance from near to far. Then calculate the straight line distance and elevation difference between adjacent track points, and preferentially select track points with moderate distance and gentle elevation change for connection to generate continuous flight path segments, avoiding paths with sharp turns or steep ascents.

[0100] Spare observation positions are set near the key track points of the flight path segment, and the key track points include track points at the boundary points of the sub-regions and terrain changes, as well as track points with poor communication quality. Two spare positions are set around each key track point, located 30 meters to the left and right of the main track point, respectively. The elevation of the spare position is adjusted based on the elevation of the main track point to ensure that it is at the same level of communication and the communication quality level is not lower than that of the main track point. Flight speed parameters are configured for each flight path segment according to the time schedule of cooperative scheduling and the complexity of the sub-regional terrain: the speed of the path segment in the flat terrain area is set to 15 meters per second, and the speed in the complex terrain area is reduced to 8 meters per second, and the speed near the key track points.

[0101] In this embodiment, the initial search space is determined based on the task allocation scheme and the performance parameters of the unmanned aerial vehicle, so that the track calculation is always anchored to the actual task demand and equipment capability, avoiding the generation of invalid tracks that exceed the performance of the unmanned aerial vehicle or deviate from the inspection range; the track points are planned in combination with the task priority sequence to ensure that the sub-regions corresponding to emergency tasks are preferentially covered, guarantee the consistency of task execution and scheduling planning, and improve the task response efficiency; the three-dimensional triangular mesh model is constructed to completely fit the terrain features of the inspection sub-region, and the track points are generated based on the terrain elevation data to avoid collision risks between the unmanned aerial vehicle and the terrain and surface attachments from the source; the flight speed parameter is matched according to the terrain complexity, and the flight speed is reduced in complex terrain areas to further improve the flight safety and adapt to the scene demand in the environment without public network. According to the communication coverage model, the height of the track point is dynamically adjusted, so that the unmanned aerial vehicle is always in an area with better communication quality in the ideal hovering position, solving the problem that the signal is easily blocked by the terrain in the environment without public network, and ensuring the real-time and stable return of the inspection data; the height optimization of the track point is bound with the communication quality, reducing the data loss caused by signal interruption and improving the data collection integrity. The optimization sorting of the connection sequence of the track points shortens the invalid flight distance of the unmanned aerial vehicle, and in combination with the flight speed and observation time length configuration, the inspection rhythm is accurately matched with the task time window, improving the overall inspection efficiency; the design of the standby observation position near the key track point provides an alternative solution for sudden failures and temporary signal fluctuations of the unmanned aerial vehicle, which can quickly adjust without re-planning the track, enhancing the continuity of the inspection task and the system fault tolerance capability.

[0102] In a preferred embodiment of the present application, step S4, processing flight trajectory data to obtain a three-dimensional environmental situation data set, includes:

[0103] Step S400, based on the complete flight trajectory coordinate sequence contained in the flight trajectory data, extracts the environmental perception data continuously collected by the unmanned aerial vehicle during the inspection, and processes the environmental perception data according to the flight time stamp to form a three-dimensional data matrix, specifically including: extracting the complete flight trajectory coordinate sequence from the flight trajectory data, which contains the three-dimensional space coordinates of the unmanned aerial vehicle at each time during the inspection and the corresponding flight time stamp, and arranging these time stamps into a continuous time sequence in chronological order. At the same time, the various types of environmental perception data collected by the laser radar device, high-definition camera and environmental sensor carried by the unmanned aerial vehicle during the inspection are called, including the terrain point cloud data generated by the laser radar, the surface image pixel data photographed by the high-definition camera, the temperature, humidity and air visibility data recorded by the environmental sensor, and the attitude data of the unmanned aerial vehicle itself such as pitch angle, roll angle and heading angle.

[0104] The collection time stamp of each piece of environment perception data is checked one by one and matched with the flight time stamp in the flight trajectory coordinate sequence. For environment perception data and flight coordinate data that are completely consistent in time stamp, direct association and binding are performed; for the case that there is a slight deviation between the environment perception data collection time stamp and the flight time stamp, the environment perception data is time calibrated in a linear interpolation manner based on the flight time stamp, so as to ensure that there is corresponding environment perception data for each flight time point. The flight time sequence after arrangement is taken as the first dimension, the longitude, latitude and altitude of the three-dimensional space coordinates of the unmanned aerial vehicle is taken as the second dimension, and the core parameters of various environment perception data are taken as the third dimension, so as to construct a three-dimensional data matrix. Each element in the matrix corresponds to an environment perception parameter value at a specific time point and a specific spatial position. For missing data caused by temporary device failure, the average value of adjacent data in the same time period is used for supplement, so as to ensure the integrity and continuity of the three-dimensional data matrix.

[0105] In step S401, based on the three-dimensional data matrix, the flight trajectory is divided into a plurality of continuous time interval segments on the horizontal projection plane, each time interval segment corresponds to the flight path between two adjacent flight path points on the flight trajectory, and the environment perception data in each time interval segment is projected onto the vertical plane of the flight path to form a plurality of trapezoidal data regions. Specifically, based on the three-dimensional data matrix generated in step S400, the flight trajectory coordinate sequence is extracted, the altitude value of each coordinate point is removed, and only the longitude and latitude information is retained, the projection operation of the flight trajectory on the horizontal projection plane is completed, and the flight path curve on the two-dimensional plane is obtained.

[0106] On the horizontal projection flight path curve, adjacent two flight path points are taken as nodes to divide time interval segments, the starting time of each interval segment is the flight time stamp of the previous flight path point, and the ending time is the flight time stamp of the next flight path point. The time length of each interval segment is calculated and recorded. At the same time, the flight path corresponding to each time interval segment is determined, that is, the straight line segment between the adjacent two flight path points on the horizontal projection plane, which is the reference path for subsequent data projection. For each time interval segment, a corresponding vertical plane is constructed, which needs to meet two conditions, one is to contain the horizontal projection flight path corresponding to the interval segment, and the other is to be perpendicular to the horizontal projection plane. All environment perception data in the current time interval segment are extracted from the three-dimensional data matrix, and these data all have corresponding three-dimensional space coordinate information.

[0107] The extracted environmental perception data is projected onto the constructed vertical plane according to its three-dimensional spatial coordinates, and the elevation information and perception parameter values of the data are kept unchanged during the projection process, and only the relative position of the data in the plane is converted. Due to the intersection of the vertical planes corresponding to adjacent time interval segments at a certain angle, the projection graph formed by each interval segment naturally forms a plurality of continuous trapezoidal data regions after the projection graphs of adjacent interval segments are connected. The upper and lower bases of each trapezoid correspond to the projection lengths of the flight paths of the adjacent two interval segments, and the waist corresponds to the intersection line of the adjacent vertical planes. Each trapezoidal data region contains all the projected environmental perception data in the time interval segment.

[0108] In step S402, the coverage area value and the data density distribution characteristics of the trapezoidal data region in the three-dimensional space are calculated; the coverage area value is converted into the confidence weight of the environmental situation, and the environmental situation feature vector corresponding to each trapezoidal data region is generated, which specifically includes: for each trapezoidal data region, first, according to the three-dimensional spatial coordinates of its four vertices, the actual coverage area of the trapezoid in the three-dimensional space is calculated. When calculating, first convert the four vertex coordinates of the trapezoid into vectors in a unified coordinate system, and calculate the area of the plane where the trapezoid is located by vector cross multiplication. The area is the coverage area value of the trapezoidal data region.

[0109] The total number of environmental perception data points contained in each trapezoidal data region is counted, and the total number of data points is divided by the coverage area value of the region to obtain the number of data points per unit area, which is used as the core indicator of data density. At the same time, the distribution of data points in the region is analyzed. If the data points are concentrated in the central region of the trapezoid, it is determined to be densely distributed; if the data points are evenly distributed in the entire trapezoidal region, it is determined to be evenly distributed; if the data points are concentrated in the edge region of the trapezoid, it is determined to be edge distribution. These characteristics together constitute the data density distribution characteristics.

[0110] The coverage area values of all trapezoidal data regions are collected, and the maximum and minimum values are found. The coverage area value of each region is normalized according to the difference between the maximum value and the minimum value, and the normalized value range is controlled between 0.1 and 1. The normalized value is the confidence weight of the environmental situation of the region. The larger the coverage area, the more comprehensive the environmental data of the region, and the higher the confidence weight.

[0111] The coverage area value data density distribution characteristics confidence weight of each trapezoidal data region and the corresponding time interval segment information starting track point coordinates and ending track point coordinates are integrated to form a unique environmental situation feature vector for the region. The parameters in each feature vector are arranged in a fixed order to ensure consistency in subsequent processing.

[0112] Step S403, the environmental situation feature vector is weighted and fused, the spatial correlation and the time continuity between different trapezoidal data regions are identified, and a three-dimensional environmental situation dataset is constructed, specifically including: taking the confidence weight in each environmental situation feature vector as a weighting coefficient, the feature vectors of all trapezoidal data regions are weighted and fused. When fusing, the corresponding parameters in each feature vector are multiplied by the confidence weight of itself respectively to obtain the weighted parameter value, and then all the weighted feature vectors are preliminarily arranged in chronological order.

[0113] The spatial correlation between adjacent trapezoidal data regions is analyzed, the straight line distance between the center coordinates of adjacent regions is calculated, and the similarity of the environmental perception data core parameters such as terrain elevation and obstacle type of the two regions is compared. If the distance is less than the preset spatial correlation threshold and the parameter similarity is higher than the preset feature similarity threshold, it is determined that the two regions have strong spatial correlation, and the correlation relationship is marked in the dataset; if only one condition is met, it is determined as weak spatial correlation; if neither condition is met, it is determined as no spatial correlation; based on the chronological order of the time interval segment, the environmental situation change of adjacent trapezoidal data regions is checked, and the change amplitude of the key parameters in the feature vectors of the two regions is calculated. If the change amplitude is less than the preset time continuity threshold, it means that the environmental situation remains stable in this time period, and is marked as time continuity; if the change amplitude exceeds the threshold, it is recorded as a situation mutation point, and the parameter type and change value of the mutation are marked in detail, so as to reflect the time continuity feature.

[0114] All the environmental situation feature vectors after weighted fusion, together with the marked spatial correlation information, time continuity information and situation mutation point record, are collected and arranged; these data are reorganized according to the chronological order and spatial position distribution, forming a three-dimensional environmental situation dataset containing data layer, correlation layer and time layer; the data layer stores the original data of the feature vectors of each region, the correlation layer stores the spatial correlation relationship between the regions, and the time layer stores the time change sequence and mutation record of the situation, finally completing the construction of the dataset.

[0115] In this embodiment, the spatio-temporal alignment processing accurately binds the multi-source environmental perception data with the flight trajectory, eliminates the time deviation and spatial misplacement of data collection, solves the confusion problem caused by asynchronous data transmission in the environment without public network, makes each item of environmental data correspond to a clear inspection spatio-temporal position; through the trapezoidal data area division, the continuous inspection data is segmented and focused according to the flight path, the abstract environmental data is converted into concrete regionalized data units, and the environmental characteristics of different inspection sections are intuitively presented; the generated situation characteristic vector, combined with the coverage area and data density, quantifies the integrity and reliability of the environmental information of each region, makes the situation measurable, and avoids the fuzzy environmental description; weighted fusion and correlation identification integrate the dispersed regional data into an overall situation with spatial correlation and temporal continuity, which can not only locate the environmental mutation points (such as terrain mutation and obstacle appearance), but also present the dynamic change trend of the situation, so that the trajectory optimization and task reassignment are more in line with the actual environment.

[0116] In a preferred embodiment of the present application, step S5, based on the three-dimensional environmental situation data set, a cooperative control instruction sequence is generated, which includes trajectory optimization strategy and task reassignment strategy, including:

[0117] Step S500, based on the three-dimensional environmental situation data set, extract spatial distribution data; according to the spatial distribution data, combined with the task priority sequence and time schedule table in the cooperative scheduling plan, generate environmental situation evaluation results, specifically including: extracting spatial distribution data from the three-dimensional environmental situation data set, focusing on organizing the center coordinate coverage range of each trapezoidal data area and the distribution of core environmental parameters in the area, these parameters include the maximum and minimum values of terrain elevation, the type and spatial position of obstacles, the distribution interval of communication signal strength, and the coverage density of ground attachments; classify and summarize these data according to the inspection sub-regions to form a spatial distribution data set exclusive to each sub-region, clearly presenting the environmental differences of each region; retrieve the task priority sequence and time schedule table in the cooperative scheduling plan, and match the spatial distribution data of each inspection sub-region with the corresponding task requirements in order from high to low according to the task priority. The task requirements include the expected completion time of the sub-region inspection task, the environmental accuracy requirement of the key inspection points, and the types of dangerous environments that must be avoided, etc.

[0118] For each inspection sub-region, extract the indicators related to the task requirements from the spatial distribution data for evaluation; if there is no unmarked sudden obstacle in the sub-region, the communication signal strength meets the data transmission requirements, the terrain complexity is within the scope of the unmanned aerial vehicle operation capability, and the deviation between the current time and the planned execution time of the time schedule table is within 10 minutes, then it is determined that the regional environmental situation is good and suitable for the current task arrangement.

[0119] If a terrain mutation or a new obstacle appears in a sub-region, which leads to a collision risk of the original flight trajectory, or the communication signal strength is continuously lower than the transmission threshold, or the current time deviates from the planned time by more than 20 minutes, it is determined that the environmental situation is abnormal, and the task execution scheme needs to be adjusted; the evaluation results of all sub-regions are sorted by priority, and the situation level of each region is marked as good, abnormal or pending, as well as the type and specific location of the abnormality, to form a complete environmental situation evaluation result.

[0120] In step S501, based on the environmental situation evaluation result, the flight trajectory data is corrected in real time to generate a trajectory optimization strategy; based on the environmental situation evaluation result, the real-time number of available unmanned aerial vehicles, the remaining battery capacity and the task urgency in the nest resource state are combined to calculate the task load redistribution ratio of each inspection sub-region, and a task redistribution strategy is generated, which specifically includes: when generating the trajectory optimization strategy based on the environmental situation evaluation result, first select the inspection sub-regions with abnormal evaluation results, and extract the three-dimensional spatial coordinates and environmental parameters of the abnormal points in these regions. Check whether each track point in the original flight trajectory data is within the abnormal environment range; for the track points within the abnormal range, select the nearest point with good environmental situation from the standby observation positions in the three-dimensional environmental situation data set as the replacement track point.

[0121] The flight path between the replacement track point and the normal track points before and after it is recalculated to ensure that the turning angle of the new path does not exceed the maximum turning ability of the unmanned aerial vehicle, and the climb or descent rate is controlled within the performance range allowed by the unmanned aerial vehicle; at the same time, the flight height of the track points on the new path is adjusted according to the communication coverage model to ensure stable communication signals throughout the journey; the flight speed parameters and observation time of the adjusted track point sequence are integrated to form a trajectory optimization strategy for abnormal regions, and the flight trajectory of normal regions remains unchanged, only the communication weak sections that need to be monitored are marked; when generating the task redistribution strategy, first collect the real-time resource state of each nest through the nest local communication network, count the number of unmanned aerial vehicles in standby state in each nest, record the remaining battery capacity and expected endurance time of each unmanned aerial vehicle, and check the completion progress of the tasks undertaken by each nest. Combined with the environmental situation evaluation result, the task urgency of the abnormal sub-region is increased by one level, and if the original task is marked as urgent, an additional resource tilt weight is added.

[0122] The current task load of each inspection sub-region is calculated, and the load value is obtained by multiplying the number of required drones by the expected execution time; for sub-regions whose original nest resources cannot meet the load demand, such as insufficient remaining power of the drones to cover the flight distance of the adjusted trajectory, or the number of standby drones is less than the required number, the gap proportion of the task load is calculated; the standby relay nest closest to the sub-region and with high communication connection strength is selected from the resource sufficient nest, and the task gap is distributed according to the resource redundancy of the standby nest, which is obtained by subtracting the number of standby drones from the number required by the task load, and the higher the redundancy, the higher the redistribution proportion of the nest, and the task content added or reduced by each nest and the time node of task handover are determined to form a task redistribution strategy.

[0123] In step S502, based on the trajectory optimization strategy and the task redistribution strategy, the instruction sequence is processed in time sequence and spatial position relationship to generate a cooperative control instruction sequence containing a complete trajectory optimization instruction set and a task redistribution instruction set, which specifically includes: collecting all trajectory adjustment instructions in the trajectory optimization strategy and task handover instructions in the task redistribution strategy, and preliminarily sorting them in time sequence; taking the time schedule table of cooperative scheduling as the benchmark, the instructions to be executed within 30 minutes after the current time point are arranged in priority, such as the take-off adjustment instruction of the drone of the emergency task, or the flight path correction instruction about to enter the abnormal area.

[0124] For instructions in the same time dimension, further sorting is performed according to the spatial position relationship, the trajectory optimization instructions and the task redistribution instructions belonging to the same inspection sub-region are grouped together to ensure that the task adjustment and trajectory correction of the region are executed synchronously; if the instructions of different sub-regions are associated, such as the task redistribution of sub-region A affecting the cooperative work of sub-region B, the associated instructions are sorted according to the causal relationship, and the instructions causing the association are executed first; the sorted instructions are sequenced, each instruction is assigned a unique instruction number, the execution subject of the instruction, i.e. the corresponding nest or drone, the execution time is accurate to minutes, and the execution parameters include the specific flight path point coordinate, the adjusted flight speed, the specific content of task handover, etc.; the instructions related to trajectory optimization are integrated into a complete trajectory optimization instruction set and stored according to the inspection sub-region; the instructions related to task redistribution are integrated into a task redistribution instruction set and stored according to the nest identifier.

[0125] On the basis of the two instruction sets, verification rules for instruction execution are supplemented, such as state information to be fed back by the drone after execution, and retry mechanism when the instruction execution fails; these contents are integrated according to the structure of time sequence, spatial grouping, instruction set and verification rule, and finally a complete cooperative control instruction sequence is generated to ensure that each instruction in the sequence is clear and explicit, which can be directly received and executed by the nest master control unit.

[0126] In this embodiment, based on three-dimensional environmental situation data, the evaluation result is generated in combination with task priority and time planning. Whether it is environmental changes such as terrain mutation, weak signal area, or task emergency degree adjustment, it can be accurately captured through the evaluation result to avoid the disconnection between the instruction and the actual environment, and solve the problem of difficult timely access to external decision support in the environment without public network. The trajectory optimization strategy is used to correct the flight path in real time in response to environmental abnormalities, replace the flight path point and recalculate the path to avoid collision risk and communication blind area, and ensure that the unmanned aerial vehicle can still operate safely in a complex environment; the task reassignment strategy is based on the real-time resource dynamic adjustment of the nest to adjust the load, when the unmanned aerial vehicle of a nest is insufficient in power or in quantity, the task is quickly diverted to the nest with sufficient resources to avoid task interruption caused by single nest failure and ensure the continuity of inspection. In terms of resource utilization efficiency, the task reassignment strategy takes the real-time available quantity, battery power and task urgency as the basis for distribution to avoid the dual problems of resource idling and overload. For example, the unmanned aerial vehicle with long battery life is used to undertake long-distance tasks, and the emergency task occupies sufficient resources to maximize the efficiency of each unmanned aerial vehicle and each nest, especially in the scenario of resource scheduling in the environment without public network, which is difficult to rely on external coordination.

[0127] In a preferred embodiment of the present application, step S6, according to the cooperative control instruction sequence, a distributed control is performed, comprising:

[0128] Step S600, based on the cooperative control instruction sequence, the instruction is parsed according to the instruction priority identifier and the effective time stamp, the trajectory optimization instruction subset and the task reassignment instruction subset related to the nest are extracted, the local execution parameter set is generated combining the current unmanned aerial vehicle state information and the environmental perception data of the nest, the local execution parameter set includes the unmanned aerial vehicle control parameter, the task switching timing and the communication synchronization point, specifically including: the master control unit of each nest first receives the complete cooperative control instruction sequence, and starts the instruction parsing module to process the instructions in batches; when parsing, the priority identifier and the effective time stamp of each instruction are extracted first, the priority identifier is divided into three levels of emergency, important and regular, and the effective time stamp is accurate to seconds. All instructions are sorted by effective time stamp from near to far, and instructions with the same time stamp are arranged from high to low according to the priority identifier, forming an ordered instruction parsing queue to ensure that emergency instructions and instructions with near effective time are processed first.

[0129] In the instruction queue, the identifier information of the nest is matched one by one, and all instructions whose instruction receiving object is the current nest are screened out, which together constitute the trajectory optimization instruction subset and the task reassignment instruction subset related to the nest; the trajectory optimization instruction subset mainly marks the coordinates of the flight path points to be adjusted, the corresponding flight speed and height threshold; the task reassignment instruction subset clearly indicates the boundary of the newly added or transferred inspection sub-area, the task completion time limit and the associated unmanned aerial vehicle number.

[0130] The current UAV state information of the local nest is retrieved, including the real-time position, remaining battery power, flight attitude, sensor working state, and progress of the executed task of each UAV, and the latest environmental perception data collected by the environmental sensors of the local nest, such as the current regional wind speed, visibility, and communication signal strength, are summarized. These real-time data are compared and calibrated with the screened instruction subset, for example, if the current battery power of the UAV specified in the instruction is insufficient, the maximum flight distance of the UAV is recalculated in combination with the remaining power, and the flight speed parameter in the instruction is fine-tuned accordingly.

[0131] The localized execution parameter set is generated based on the calibrated instruction data; wherein the UAV control parameters include the target waypoint sequence of each UAV, the flight speed range of each segment, the height adjustment gradient, and the hovering observation time length; the task switching timing specifies the time node at which the UAV switches from the current task to the new task, the position coordinates at the task handover, and the state confirmation process; and the communication synchronization point corresponds to the information exchange time between nests required in the instruction, and marks the specific content and format of the state data to be uploaded, to ensure the consistency of the information uploaded by each nest.

[0132] In step S601, each nest performs distributed control on the UAVs within the jurisdiction of the local nest based on the localized execution parameter set, including adjusting the flight trajectory, speed, and height parameters of the UAVs according to the trajectory optimization instruction subset, dynamically allocating the task target area of the UAVs according to the task reassignment instruction subset, and uploading the execution state information to the effective communication relay area in the communication coverage model in real time through the built-in communication equipment of the nest, to realize the sharing of state information between nests, including: the main control unit of each nest implements distributed control on the UAVs within the jurisdiction according to the localized execution parameter set. For trajectory adjustment, the main control unit converts the target waypoint sequence in the trajectory optimization instruction into a control signal recognizable by the UAV, and issues it in real time through the special communication link between the nest and the UAV; during the flight of the UAV, the deviation between its actual position and the target waypoint is continuously compared, and if the deviation exceeds 5 meters, a correction instruction is immediately sent to adjust the heading angle and flight speed of the UAV, to ensure that the flight trajectory conforms to the optimized path, and at the same time, the flight height is dynamically adjusted according to the height threshold in the parameter set and in combination with real-time terrain data, to avoid collision with the terrain or obstacles.

[0133] For task allocation, the master unit allocates specific task target areas to the drones in standby state according to the task re-allocation instruction subset; when allocating, the drones with sufficient battery capacity and good sensor performance are preferentially allocated with emergency tasks, and the boundary coordinates of the task area, key inspection points and data collection requirements are explicitly informed. If task handover is involved, that is, the drones of other nests need to assist in executing the tasks of the current nest or the drones of the current nest assist other areas, the master unit will send task switching instructions to the related drones 10 minutes in advance, synchronously inform the time and place of task handover and the interface nest identifier, and ensure smooth task connection.

[0134] During execution, the nest uploads real-time execution state information to the effective communication relay area marked in the communication coverage model through the built-in LoRa and ZigBee fusion communication module; the uploaded information includes the real-time position, battery capacity, task completion progress, current flight parameters and abnormal situations encountered by each drone of the nest, such as signal interruption and sensor failure, the information is encoded in a fixed format and updated every 30 seconds; at the same time, the nest receives state data uploaded by other nests to the relay area through the communication link, and the master unit analyzes these data in real time, and if it is found that there is a risk of operation conflict or resource shortage in the drones of the adjacent nest, the execution parameters of the current nest are adjusted in time to realize dynamic cooperation between nests.

[0135] When encountering sudden abnormalities such as loss of communication with the drone or failure of instruction execution, the nest master unit immediately starts the local emergency mechanism, schedules the standby drone to replace the task according to the abnormal handling plan in the parameter set, and quickly uploads the abnormal situation and processing measures to the relay area and synchronously to other related nests, so that each nest can not only complete its own task independently, but also realize global cooperation through state sharing, without relying on the public network to maintain the stable operation of the entire inspection system.

[0136] As shown in Figure 2 The embodiment of the present application also proposes a multi-nest cooperative unmanned aerial vehicle autonomous flight and inspection system in a public network-free communication area, which comprises:

[0137] The processing module is used for discretizing the terrain data into continuous terrain grid units with uniform spatial resolution by collecting digital terrain data and nest distribution data of the target inspection area, and calculating the coverage ability coefficient of each nest to the terrain grid based on the nest position and communication parameters to form a nest-terrain grid coverage relationship matrix; the elevation change gradient of the terrain undulation feature and the nest coverage ability distribution data are extracted from the nest-terrain grid coverage relationship matrix, the inspection area is dynamically divided into non-overlapping and completely covered inspection sub-areas, and the final inspection zoning scheme is generated by dynamically adjusting the area size and shape complexity of each sub-area;

[0138] A construction module is configured to construct a communication coverage model based on the patrol partition scheme and the nest resource state, generate a cooperative scheduling plan according to the communication coverage model, and perform three-dimensional path optimization calculation by each nest according to the cooperative scheduling plan to obtain flight trajectory data.

[0139] A control module is configured to process the flight trajectory data to obtain a three-dimensional environment situation data set, generate a cooperative control instruction sequence based on the three-dimensional environment situation data set, wherein the cooperative control instruction sequence comprises a trajectory optimization strategy and a task re-allocation strategy, and perform distributed control according to the cooperative control instruction sequence.

[0140] The above describes the preferred embodiments of the present application. It should be noted that, for those skilled in the art, without departing from the principles of the present application, a number of improvements and refinements can be made, which should also be considered as the protection scope of the present application.

Claims

1. A method for autonomous flight patrol of a UAV in a public network communication area multi-machine nest cooperation, characterized in that, The method comprises: Step S1, by collecting the digital terrain data and nest distribution data of the target inspection area, discretizing the terrain data into continuous terrain grid units with uniform spatial resolution, and calculating the coverage ability coefficient of each nest to the terrain grid based on the nest location and communication parameters, forming a nest-terrain grid coverage relationship matrix; extracting the elevation change gradient of the terrain relief feature and the nest coverage ability distribution data from the nest-terrain grid coverage relationship matrix, dynamically segmenting the inspection area into non-overlapping and complete inspection sub-areas, and generating the final inspection zoning scheme by dynamically adjusting the area size and shape complexity of each sub-area, including: Collecting the digital terrain data of the target inspection area, and obtaining the nest distribution data of each nest, the nest distribution data including the geographic location coordinates of each nest, the communication equipment performance parameters and the number of schedulable unmanned aerial vehicles state information; Based on the digital terrain data, it is discretized into a set of continuous terrain grid units with uniform spatial resolution; based on the geographic location coordinates and communication equipment performance parameters in the nest distribution data, the communication coverage distance weight and the coverage ability coefficient of each nest to each terrain grid unit under the influence of terrain obstruction are calculated, forming a nest-terrain grid unit coverage relationship matrix, including: based on the digital terrain data and the nest distribution data, the digital terrain data is discretized; according to the spatial resolution, the entire target inspection area is divided into a set of continuous and non-overlapping terrain grid units, each grid unit is assigned a unique identification number, and the three-dimensional spatial coordinate elevation information of the center point of each grid unit, the terrain slope value and the surface attached object type and other core data are extracted, the discretization conversion of the terrain data is completed; the coverage ability coefficient of each nest to each terrain grid unit is calculated, and the calculation basis of the communication distance weight is determined; the straight line spatial distance between the geographic location coordinates of the nest provided by the global positioning system and the center point coordinates of the grid unit is calculated; The terrain obstruction influence coefficient is calculated, the ray tracing method is used to simulate the communication signal propagation path between the nest and the grid unit, a virtual ray is emitted from the nest position to the center point of the grid unit, and the elevation information of all terrain grid units through which the ray passes is checked one by one; if the ray is not blocked by the elevation of other terrain grid units throughout the journey, that is, the elevation of all points on the ray trajectory is lower than the elevation of the line connecting the nest and the target grid unit, the terrain obstruction influence coefficient is 1; if there is an obstruction, the number of grid units in the obstruction area and the difference between the highest obstruction elevation and the line elevation are counted, the more the number of obstruction grid units, the larger the difference, and the smaller the obstruction influence coefficient, which can be reduced to 0.05 at the lowest; the communication distance weight of each nest to the corresponding grid unit is multiplied by the terrain obstruction influence coefficient to obtain the coverage ability coefficient of the nest to the grid unit; taking the nest identification as the row index and the terrain grid unit identification as the column index, the coverage ability coefficients of all nests to all grid units are sequentially filled in to form a complete nest-terrain grid unit coverage relationship matrix, each value in the matrix represents the actual coverage ability of the corresponding nest to the corresponding grid unit. Step S2, based on the inspection zoning scheme and the nest resource state, a communication coverage model is constructed, including: based on the inspection zoning scheme and the nest resource state, the theoretical communication coverage radius of each nest in the three-dimensional space is calculated, and the actual communication coverage range of each nest under the influence of the terrain obstruction is determined; based on the actual communication coverage range of each nest, the communication coverage area of each nest is modeled as a conical body model with the geographic location coordinates of the nest as the vertex, the actual communication coverage radius as the generatrix length, and the terrain fluctuation angle as the opening angle; the intersection region volume between the conical body models corresponding to adjacent nests in the three-dimensional space position is calculated, and when the intersection region volume is greater than a preset minimum communication relay volume threshold, the corresponding intersection region is determined as an effective communication relay region; each conical body model communication coverage space is divided into multiple horizontal communication layers along the vertical height direction; according to the signal strength attenuation degree and data transmission bandwidth demand in the height interval of each horizontal communication layer, a corresponding communication priority level and dynamic bandwidth allocation strategy are allocated to each horizontal communication layer; based on the effective communication relay region, the communication priority level and the dynamic bandwidth allocation strategy, a communication coverage model is constructed; according to the communication coverage model, a cooperative scheduling plan is generated; Step S3, according to the cooperative scheduling plan, each nest respectively performs three-dimensional track optimization calculation to obtain flight trajectory data; Step S4, processing the flight trajectory data to obtain a three-dimensional environmental situation data set; Step S5, based on the three-dimensional environmental situation data set, a cooperative control instruction sequence is generated, which includes a trajectory optimization strategy and a task reassignment strategy; Step S6, according to the cooperative control instruction sequence, distributed control is performed.

2. The method of claim 1, wherein the method further comprises: According to the communication coverage model, a cooperative scheduling plan is generated, including: Based on the distribution of the effective communication relay region in the communication coverage model, the communication connection relationship between each inspection sub-region is identified, and a communication topology network between the nests is constructed; According to the connection strength and bandwidth capacity of each node in the communication topology network, a master nest and a standby relay nest are allocated to each inspection sub-region, forming a hierarchical cooperative control architecture; According to the cooperative control architecture, combined with the number of unmanned aerial vehicles, battery endurance time and task urgency in the nest resource state, a specific inspection task sequence and take-off time window are allocated to each nest; According to the inspection task sequence and the take-off time window, the time synchronization point of the cooperative operation of unmanned aerial vehicles between adjacent nests is calculated; according to the time synchronization point of the cooperative operation of unmanned aerial vehicles between adjacent nests, the terrain complexity and the communication signal stability, the inspection speed and data transmission frequency of each nest are dynamically adjusted, and a cooperative scheduling plan is generated.

3. The method of claim 2, wherein the method further comprises: The cooperative scheduling plan includes a task allocation scheme, a time schedule, a communication relay path, and an exception handling plan.

4. The method of claim 3, wherein the method further comprises: According to the cooperative scheduling plan, each nest respectively performs three-dimensional track optimization calculation to obtain flight trajectory data, including: Based on the task allocation scheme and the time schedule table contained in the cooperative scheduling plan, the boundary coordinate set of the sub-region responsible for inspection by each nest and the corresponding priority sequence are extracted, and the initial search space range for three-dimensional track optimization calculation is determined in combination with the performance parameters of the unmanned aerial vehicle in the nest resource state; Based on the initial search space range, the inspection sub-region is divided into multiple continuous triangular element grids in three-dimensional space, and the vertex coordinates of each triangular element grid are determined by the boundary points of the sub-region and the terrain elevation data, forming a three-dimensional triangular grid model covering the entire inspection sub-region; Based on the three-dimensional triangular grid model, the geometric center point coordinates of each triangular element grid are taken as the ideal hovering observation positions of the unmanned aerial vehicle; according to the communication quality level of the corresponding ideal hovering observation position in the communication coverage model, the flight height weight parameter of the geometric center point is dynamically adjusted to generate a preliminary set of track points; Based on the preliminary set of track points, the connection order between adjacent track points is optimized and sorted to generate a flight path segment; a backup observation position is added near the key track point of the flight path segment, and finally the flight trajectory data containing the complete flight trajectory coordinate sequence, flight speed parameter and observation time length configuration are obtained.

5. The method of claim 4, wherein the method further comprises: The flight trajectory data is processed to obtain a three-dimensional environmental situation data set, including: Based on the complete flight trajectory coordinate sequence contained in the flight trajectory data, the environmental perception data continuously collected by the unmanned aerial vehicle during the inspection process is extracted, and the environmental perception data is spatio-temporally aligned according to the flight time stamp to form a three-dimensional data matrix; Based on the three-dimensional data matrix, the flight trajectory is divided into multiple continuous time interval segments on the horizontal projection plane, each time interval segment corresponding to the flight path between two adjacent track points on the flight trajectory, and the environmental perception data in each time interval segment is projected onto the vertical plane of the flight path to form multiple trapezoidal data regions; The coverage area value and data density distribution characteristics of the trapezoidal data region in three-dimensional space are calculated; the coverage area value is converted into the confidence weight of the environmental situation to generate the environmental situation feature vector corresponding to each trapezoidal data region; The environmental situation feature vectors are weighted and fused to identify the spatial correlation and temporal continuity between different trapezoidal data regions, and a three-dimensional environmental situation data set is constructed.

6. The method of claim 5, wherein the method further comprises: Based on the three-dimensional environmental situation data set, a cooperative control instruction sequence is generated, which contains a trajectory optimization strategy and a task reallocation strategy, including: Based on the three-dimensional environmental situation data set, spatial distribution data are extracted; according to the spatial distribution data, the task priority sequence and the time schedule table in the cooperative scheduling plan are combined to generate an environmental situation evaluation result; Based on the environmental situation evaluation result, the flight trajectory data is corrected in real time to generate a trajectory optimization strategy; based on the environmental situation evaluation result, the real-time number of available unmanned aerial vehicles, the remaining battery capacity and the task urgency in the nest resource state are combined to calculate the task load reallocation proportion of each inspection sub-region, and a task reallocation strategy is generated; Based on the trajectory optimization strategy and the task reassignment strategy, the instruction serialization processing is performed according to the time sequence and the spatial position relationship, and a cooperative control instruction sequence containing a complete trajectory optimization instruction set and a task reassignment instruction set is generated.

7. The method of claim 6, wherein the method further comprises: According to the cooperative control instruction sequence, distributed control is performed, including: Each nest performs instruction analysis according to the instruction priority identifier and the effective time stamp based on the cooperative control instruction sequence, extracts a trajectory optimization instruction subset and a task reassignment instruction subset related to the nest, and generates a localized execution parameter set in combination with the current unmanned aerial vehicle state information and the environment perception data of the nest, wherein the localized execution parameter set contains unmanned aerial vehicle control parameters, task switching timing and communication synchronization points. Each nest performs distributed control on the unmanned aerial vehicles within the jurisdiction of the nest based on the localized execution parameter set, including adjusting the flight trajectory, speed and height parameters of the unmanned aerial vehicles according to the trajectory optimization instruction subset, dynamically assigning the task target area of the unmanned aerial vehicles according to the task reassignment instruction subset, and uploading the execution state information to the effective communication relay area in the communication coverage model in real time through the communication equipment built in the nest, to realize the sharing of state information between nests.

8. An autonomous unmanned aerial vehicle (UAV) patrol system with multi-nest collaboration in areas without public network communication, wherein the system implements the method as described in any one of claims 1 to 7, characterized in that, It includes: The processing module is configured to discretize the terrain data into continuous terrain grid units with a uniform spatial resolution by collecting digital terrain data and nest distribution data of the target inspection area, and calculate the coverage ability coefficient of each nest on the terrain grid based on the nest position and communication parameters to form a nest-terrain grid coverage relationship matrix; extract the elevation change gradient of the terrain undulation feature and the nest coverage ability distribution data from the nest-terrain grid coverage relationship matrix, dynamically segment the inspection area into non-overlapping and complete inspection sub-areas, and generate the final inspection zoning scheme by dynamically adjusting the area size and shape complexity of each sub-area. The construction module is configured to construct a communication coverage model based on the inspection zoning scheme and the nest resource state, generate a cooperative scheduling plan according to the communication coverage model, and perform three-dimensional trajectory optimization calculation by each nest according to the cooperative scheduling plan to obtain flight trajectory data. The control module is configured to process the flight trajectory data to obtain a three-dimensional environmental situation data set, generate a cooperative control instruction sequence based on the three-dimensional environmental situation data set, and perform distributed control according to the cooperative control instruction sequence.

Images