Intelligent planning method for space monitoring of unmanned aerial vehicle

Abstract

The invention discloses an intelligent planning method for space monitoring of an unmanned aerial vehicle. The method comprises the steps that intelligent path allocation is realized by constructing a task demand priority matrix; the method comprises the following steps: firstly, collecting geographical, climate and environmental parameters of a monitoring area, quantifying regional complexity and color features of a monitoring target by combining high-resolution image data with a neural network model, and generating a priority matrix according to task importance, change frequency and risk level; a Dijkstra algorithm is adopted to plan an initial flight path giving consideration to priority and flight limitation, a path complexity index is calculated, and the index comprehensively considers a target priority weight, a task detouring coefficient and a path relaxation degree; and finally, dividing a monitoring area into height layers according to a path complexity index threshold value, performing height layer adjustment on the initial path, and generating a dynamic flight path containing height layer switching, thereby realizing efficient resource allocation and accurate risk prevention and control in a complex monitoring scene.

Description

Technical Field

[0001] The present invention relates to the field of electronic information technology, and in particular to an intelligent planning method for unmanned aerial vehicle space monitoring. Background Art

[0002] The development of intelligent planning methods for drone-based spatial monitoring stems from the limitations of traditional monitoring methods. In fields such as environmental monitoring, public safety, and architectural surveying and mapping, traditional manual inspections or fixed-device monitoring suffer from low efficiency, insufficient coverage, and slow response. While drones offer advantages such as maneuverability and wide coverage, path planning in complex environments remains challenging. For example, in GNSS-denied areas such as mountains and understory forests, traditional planning based on preset routes is susceptible to terrain obstruction, resulting in monitoring blind spots. Furthermore, multi-target monitoring tasks require a balance between efficiency and accuracy, making traditional static planning difficult to dynamically adapt to spatial complexity.

[0003] The invention patent application number 202110868376.4 discloses a method and system for intelligent planning of UAV spatial monitoring paths. The method comprises the following steps: obtaining a first monitoring target of a first UAV; obtaining first image data of the first monitoring target according to the first remote sensing monitoring module; constructing a first complexity detection model; inputting the first image data into the first complexity detection model, and obtaining a first complexity index according to the first complexity detection model; obtaining first spatial hierarchical information when the first complexity index is greater than or equal to a preset complexity index; and planning the spatial monitoring path of the first UAV based on the first spatial hierarchical information to obtain a first spatial planning path. This method solves the technical problems that the existing technology cannot plan the flight path of a UAV based on the spatial hierarchy, and the traditional path planning is inaccurate and unintelligent.

[0004] Existing technologies ignore the complex characteristics of the monitoring area and the differences in monitoring targets, resulting in low task execution efficiency and irrational resource allocation. In complex scenarios with diverse terrain, unpredictable weather, and varying importance and degree of change of monitoring targets, single-path planning is difficult to meet the needs. Summary of the Invention

[0005] This application provides an intelligent planning method for drone spatial monitoring, combines multi-dimensional parameters and image data, uses intelligent algorithms to quickly respond to changes in the monitoring area, optimizes the allocation of drone monitoring resources, and improves monitoring efficiency and accuracy.

[0006] This application provides an intelligent planning method for drone space monitoring, including:

[0007] S1, collects multi-dimensional parameters and image data of the monitoring area and generates a task requirement priority matrix;

[0008] S2, plans a preliminary flight path according to the priority matrix, obtains the mission detour coefficient and path slackness of each monitoring target according to the preliminary flight path, and calculates the path complexity index according to the mission detour coefficient and path slackness of each monitoring target;

[0009] S3 divides the monitoring area into altitude layers based on the path complexity index and generates a dynamic flight path based on the altitude layers; records the real-time trajectory of the drone and calculates the completion rate of the monitoring target; comprehensively adjusts the delay time and the deviation distance generated during the altitude layer switching process, and quantifies the sensitivity value corresponding to the layer adjustment; constructs an elastic model that reflects the dynamic adjustment of the drone's path planning strategy based on the sensitivity value, and generates candidate paths based on the elastic model; corrects the path complexity index according to the environmental dynamic adaptation coefficient of each candidate path, and adjusts the path preferentially.

[0010] Preferably, generating a task requirement priority matrix specifically includes:

[0011] The UAV control system sets the geographic scope of the area to be monitored and acquires its high-resolution image data. A complex target recognition model and an n×m matrix mission requirement priority matrix are constructed by combining the collected multi-dimensional parameters and image data. Based on the mission requirement priority matrix, a list of monitoring targets is generated, where n is the number of targets and m is the target weight. The matrix is filled in and sorted by weight priority information to form a mission requirement priority matrix.

[0012] Preferably, the calculation formula for the calculation path complexity index is as follows: CRI i Represents the path complexity index of the i-th monitoring target; RP i Indicates the task requirement weight priority of the i-th monitoring target; RNC i represents the mission detour coefficient of the i-th monitoring target; PS i represents the path relaxation of the i-th monitoring target, that is, the range of distance allowed to deviate from the preset path; N represents the total number of monitoring targets.

[0013] Preferably, the path complexity index RNC of the i-th monitoring target is i Indicates the degree to which the actual path deviates from the preset path due to terrain or obstacles. The actual required flight distance is the distance that the UAV actually flies to cover the monitoring target, and the direct flight distance is the straight-line distance from the starting point to the monitoring target. The path relaxation PS of the i-th monitoring target is i Indicates the range of distance allowed to deviate from the preset path, PS i =Basic relaxation × (1 + RNC i ), the basic slack represents the maximum base distance allowed to deviate from the preset path under ideal conditions.

[0014] Preferably, the optimal path adjustment specifically includes: adding the corrected path complexity indexes of the monitoring targets on the candidate paths to obtain a comprehensive path complexity, selecting the candidate path with the smallest comprehensive path complexity as the optimal path to adjust the dynamic flight path; wherein the calculation formula for the path complexity index of the monitoring targets on the corrected candidate paths is: CRIC i =CRI i ×(1+α×DEAC), CRI is the original path complexity index, α is the environmental sensitivity factor, and DEAC is the rate of change of the external environment.

[0015] Preferably, the external environment change rate includes: monitoring the terrain height change in real time through the laser radar, calculating the terrain undulation change rate per unit time, T is the sampling time window The number of wind speed and wind direction mutations is recorded by meteorological sensors, and the mutation frequency per unit time is calculated The external environment change rate DEAC=w is calculated based on the terrain undulation change rate and mutation frequency ΔH ×ΔH+w ΔW ×ΔW; where the weight coefficient w ΔH =0.5, w ΔW =0.5, T is the sampling time window, H t is the terrain height value of the previous unit time, H t+1 It is H t The height of the terrain at the next unit of time.

[0016] Preferably, the steps for selecting the minimum candidate path are as follows: extracting the time t of each altitude layer switching instruction from the UAV flight log 指令,i and the actual execution time t 执行,i , calculate the actual time taken to switch altitude levels n is the number of historical adjustments; compare the actual path after switching with the dynamic flight path, and calculate the actual path L caused by the switch 实际,i With the ideal path L 理想,i The Euclidean distance difference n is the historical adjustment times; RS and OPL are standardized to obtain RS norm and OPL norm , substitute the sensitivity value into the calculation formula S=(w RS ×RS norm +w OPL ×OPL norm )×CTC; where w RS and w OPLis the weight coefficient; based on this, the sensitivity threshold is set, the corresponding path strategy is matched, and K candidate paths with segmented altitude layer strategies are generated using Monte Carlo simulation.

[0017] Preferably, the real-time trajectory of the drone is recorded as follows: the drones are divided into groups according to the characteristics and objectives of the monitoring area, and each group is responsible for the monitoring task of a specific sub-area; the shortest path tree is generated based on the sub-area boundary to calculate the task volume and dynamically allocate tasks; the number of drones and tasks in the group are adjusted in real time, and the flight paths of the drones in the group are re-planned.

[0018] Preferably, each group is responsible for a specific sub-area monitoring task, including: combining the image data collected in the monitoring area with meteorological data to generate a three-dimensional environmental feature map; The monitoring area is divided into n sub-areas according to the number of images in the area. N 无人机 Assign independent groups to each sub-area for the total number of available drones; set area constraints for each sub-area Among them, the redundancy factor 1.3 is used to cover boundary overlap and emergency buffer; among them, N 监测目标 is the number of sub-area monitoring targets, A 子区域 is the area of the subregion, A 总面积 is the total area of the monitoring area.

[0019] Preferably, the dynamic allocation of tasks specifically includes: extracting the boundary information of each sub-area from the three-dimensional environmental feature map, using the monitoring targets in the sub-area as nodes, and using the Dijkstra algorithm to generate a shortest path tree covering all monitoring targets in the sub-area; calculating the task amount of the sub-area according to the length of the shortest path tree and the number of monitoring targets; P i is the i-th monitoring target, and I is the indicator function; a task list is established for each sub-area, and the specific values of the task volume of each sub-area are recorded.

[0020] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0021] By collecting multi-dimensional parameters such as the monitoring area's geography, climate, and environment, as well as centimeter-level high-resolution imagery, a neural network model is constructed to analyze regional complexity and color characteristics. This assigns high weights to targets with high complexity, high risk, and high variability, generating a mission requirement priority matrix. Based on this matrix, the Dijkstra algorithm is used to plan a preliminary path that balances priorities and flight restrictions, and the path complexity index is calculated. The PCI threshold is used to divide the regional altitude layers, placing high-weight targets in the low-altitude layer and low-weight targets in the high-altitude layer. Ultimately, a dynamic flight path with altitude layer switching is generated, enabling efficient resource allocation and precise risk prevention and control in complex scenarios.

[0022] By building a dynamic path planning and sensitivity adaptation mechanism for drones, the coordinated optimization of flight efficiency and environmental adaptability is achieved. The core technology lies in: adjusting sensitivity values based on the flight log's quantitative hierarchy, dynamically evaluating system sensitivity through standardized response delay and path deviation; correcting path complexity based on real-time environmental data (terrain fluctuations, sudden weather changes), and generating flexible candidate paths; and ultimately selecting the optimal path based on the overall complexity, balancing mission completion efficiency and flight safety. This significantly improves the drone's autonomous decision-making capabilities and mission execution robustness in complex dynamic scenarios.

[0023] Through dynamic regional grouping, three-dimensional environment-integrated path planning, and real-time resource adaptive adjustment, the solution achieves efficient coordination and precise coverage of drone monitoring missions. This solution divides sub-regions based on target density and regional characteristics, dynamically adjusts path costs based on terrain and weather conditions, quantifies mission load, and dynamically allocates drone resources, ensuring priority protection in high-density areas and resource integration in low-density areas. It also optimizes flight trajectories using a path complexity index, effectively improving mission completion rates and resource utilization, significantly shortening monitoring response times in complex scenarios, and delivering both flexibility and robustness. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 The figure is a flow chart of an intelligent planning method for UAV space monitoring according to an embodiment of the present invention. DETAILED DESCRIPTION

[0025] To facilitate understanding of the present invention, the present application will be described more comprehensively below with reference to the relevant drawings; the drawings show preferred embodiments of the present invention, but the present invention can be implemented in many different forms and is not limited to the embodiments described herein; on the contrary, the purpose of providing these embodiments is to enable a more thorough and comprehensive understanding of the disclosed content of the present invention.

[0026] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this invention pertains; the terms used herein in the specification of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention; the term "and / or" used herein includes any and all combinations of one or more of the associated listed items.

[0027] Example 1: Figure 1 The present invention is a flowchart of an intelligent planning method for UAV space monitoring according to an embodiment of the present invention.

[0028] like Figure 1 As shown, a method for intelligent planning of UAV space monitoring includes the following steps:

[0029] S1, collects multi-dimensional parameters and image data of the monitoring area, determines the weight of the UAV monitoring target, and generates the mission requirement priority matrix.

[0030] Among them, multi-dimensional parameters are the geography, climate and environment of the monitoring area. These parameters include but are not limited to terrain height, slope, vegetation cover, meteorological conditions (such as wind speed, wind direction, temperature, humidity), etc.

[0031] Specifically, generate a task requirement priority matrix including:

[0032] The UAV control system is used to set the geographical area to be monitored and obtain high-resolution image data of the monitoring area.

[0033] Among them, high resolution needs to reach the centimeter level or even the millimeter level; image data should cover the terrain characteristics (topography, climate characteristics and surface coverage, etc.) and change areas (disaster-affected areas, ecological change areas and seasonal change areas, etc.) of the monitored area; image data requires adjacent images to have an overlap rate of 60%-80%.

[0034] Based on the collected multi-dimensional parameters and image data, a complex target recognition model is constructed to determine the weight of each monitoring target.

[0035] Among them, the complex target recognition model can analyze the complexity of image data, such as the complexity of regional features (such as mountains, plains, waters, etc.) and the complexity of color distribution features; the allocation of weights is a numerical indicator to measure the relative importance of monitoring targets. In the weight allocation standard, high weights are usually assigned to monitoring targets with the following characteristics: high complexity (monitoring targets have complex terrain, rich landform features or significant ecological diversity, requiring more refined image data and high-precision path planning), high importance (monitoring targets are crucial to the success of the overall monitoring task, or their monitoring results have an important impact on decision-making), high variability (monitoring targets undergo rapid changes, such as rapid urbanization, frequent natural disasters or ecological degradation, More frequent monitoring and higher data update frequency are required) and high risk (the monitoring target is located in a high-risk area, such as an earthquake zone, a flood-prone area or an ecologically sensitive area, and special attention is needed to prevent potential disasters or protect the ecological environment); low weights are usually assigned to monitoring targets with the following characteristics: low complexity (the monitoring target has a flat terrain, a single landform feature, and image data is relatively easy to obtain, and does not require excessively high path planning accuracy), low importance (the monitoring target has little impact on the overall monitoring task, and its monitoring results have limited contribution to decision-making), low variability (the monitoring target environment changes slowly, and frequent monitoring updates are not required) and low risk (the monitoring target is located in a low-risk area, and the possibility of disasters or ecological problems is low).

[0036] Specifically, preprocess the collected multi-dimensional parameters and image data, including operations such as denoising, calibration, registration, etc. Extract geographical features (such as mountains, plains, waters) and color distribution features from the image data, and use image processing algorithms (edge detection) to extract feature vectors. Select a neural network to build a complex target recognition model, and use the preprocessed data and feature vectors to train the model. Set the weight allocation criteria, and assign weights to each monitoring target according to the analysis results of the complex target recognition model.

[0037] Generate a task requirement priority matrix according to the weights of the monitoring targets.

[0038] Specifically, generate a list of monitoring targets according to the weights of the monitoring targets, construct an n×m matrix framework, where n is the number of monitoring targets and m is the weight of the monitoring target. Fill in the corresponding priority information in the matrix according to the weight of each monitoring target, sort the priority information in the matrix, and optimize and adjust according to business requirements and resource constraints.

[0039] S2. Plan the initial flight path according to the multi-priority matrix, and calculate the path complexity index for the task bypass coefficient and path slack of each monitoring target.

[0040] Among them, the formula for calculating the path complexity index is:

[0041]

[0042] CRI i represents the path complexity index of the i-th monitoring target; RP i represents the task requirement priority weight of the i-th monitoring target (0 < RP ≤ 1, the RP of high-priority targets is close to 1, and the RP of low-priority targets is close to 0); RNC i represents the task bypass coefficient of the i-th monitoring target (0 ≤ RNC ≤ 1), that is, the degree of actual path deviation from the preset path due to terrain / obstacles. The actual flight distance is the distance that the UAV actually flies to cover the monitoring target, and the direct flight distance is the straight-line distance from the starting point to the monitoring target; PS i represents the path slack of the i-th monitoring target, that is, the allowable range distance of deviation from the preset path (the larger the PS, the higher the path flexibility), PS i = basic slack × (1 + RNC i ), the basic slack represents the maximum reference distance (unit: meter) allowed to deviate from the preset path under ideal conditions (without terrain / obstacle interference), and the value range is usually a fixed value (5 - 20 meters), which is determined by factors such as UAV sensor accuracy and task fault tolerance rate; N represents the total number of monitoring targets.

[0043] Specifically, a preliminary flight path is planned based on a multi-priority matrix, including:

[0044] According to the task requirement priority matrix generated in step S1, the priority order of each monitoring target is determined.

[0045] The Dijkstra algorithm is selected to generate the optimal or suboptimal path based on the priority of the monitoring target, image data, and the flight limitations of the drone (such as flight altitude, speed, battery life, etc.) to plan the preliminary flight path.

[0046] Based on the algorithm output, a preliminary flight path is generated, which should cover all monitoring targets as much as possible and give priority to high-priority targets.

[0047] S3, divides the monitoring area into altitude layers based on the path complexity index, and generates a dynamic flight path based on the altitude layers.

[0048] Among them, dividing the monitoring area into altitude layers means dividing the monitoring area into different altitude layers according to the path complexity index and the flight capabilities of the UAV (such as maximum flight altitude, minimum safe altitude, etc.). The division of altitude layers should take into account factors such as terrain undulations and changes in meteorological conditions.

[0049] Specifically, a tiered threshold is set, for example, to divide the system into three altitude layers: low-altitude layer (H1: 50-100 meters from the monitoring target): PCI ≥ 0.7, corresponding to high-complexity, high-priority monitoring targets; medium-altitude layer (H2: 150-250 meters from the monitoring target): 0.4 ≤ PCI < 0.7, corresponding to medium-complexity, medium-priority targets; high-altitude layer (H3: 300-400 meters from the monitoring target): PCI < 0.4, corresponding to low-complexity, low-priority targets. Based on the altitude layer division results, the initial flight path is adjusted to generate a dynamic flight path for the drone's altitude layer switching.

[0050] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0051] By collecting multi-dimensional parameters such as the monitoring area's geography, climate, and environment, as well as centimeter-level high-resolution imagery, a neural network model is constructed to analyze regional complexity and color characteristics. This assigns high weights to targets with high complexity, high risk, and high variability, generating a mission requirement priority matrix. Based on this matrix, the Dijkstra algorithm is used to plan a preliminary path that balances priorities and flight restrictions, and the path complexity index is calculated. The PCI threshold is used to divide the regional altitude layers, placing high-weight targets in the low-altitude layer and low-weight targets in the high-altitude layer. Ultimately, a dynamic flight path with altitude layer switching is generated, enabling efficient resource allocation and precise risk prevention and control in complex scenarios.

[0052] Example 2: In Example 1, when generating a dynamic flight path, path planning is mainly based on preset fixed rules or simple algorithms, without fully considering the impact of multi-dimensional factors such as the response characteristics of altitude layer switching during the flight of the UAV, the actual path deviation, and changes in the external dynamic environment on the sensitivity and complexity of the path. This fixed-mode path generation method is difficult to adapt flexibly when faced with complex and changing flight missions and environments, and may cause the UAV to have problems such as untimely response, large path deviation, and unreasonable energy consumption during flight, thereby affecting the completion efficiency and quality of the monitoring task. In order to improve the quality of dynamic flight path generation so that it can better meet actual flight needs and external environmental changes, Example 2 is specially proposed.

[0053] In some embodiments, when generating a dynamic flight path, step S3 further includes:

[0054] S31, records the real-time flight trajectory of the UAV and calculates the completion rate of the monitoring target.

[0055] The calculation formula for the completion rate of monitoring targets is:

[0056]

[0057] S32, the sensitivity value corresponding to the level adjustment is quantified by combining the adjustment delay time recorded in the UAV flight log and the distance the path deviates from the ideal state caused by the altitude layer switching.

[0058] Specifically, the command issuance time and actual execution time of each altitude switching are extracted from the UAV flight log, and the average adjustment delay time (the actual time (in seconds) from the UAV receiving the command to completing the altitude switching) is calculated:

[0059]

[0060] Where n is the number of historical adjustments.

[0061] Compare the actual flight path after each level change with the dynamic flight path and calculate the path deviation distance (the Euclidean distance difference (in meters) between the actual path and the ideal path caused by the level change):

[0062]

[0063] Where n is the number of historical adjustments.

[0064] RS and OPL were normalized to calculate the sensitivity value.

[0065] Specifically, let the minimum possible value of RS be RS min(For example, theoretically, the response speed of the drone cannot be infinitely fast. A minimum value can be set according to the actual situation, assumed to be 0 seconds here), and the maximum possible value is RS max (For example, a maximum delay time can be set according to the drone's performance and actual flight situation, assumed to be 10 seconds here). Then the normalized value of RS is RS norm , and the calculation formula is: The normalized RS norm The value range is between [0, 1]. The smaller the value, the faster the response speed and the higher the sensitivity. Let the minimum possible value of OPL be OPL min (For example, 0 meters, indicating that the actual path completely coincides with the dynamic path), and the maximum possible value is OPL max (For example, a maximum deviation distance can be set according to the flight environment and mission requirements, assumed to be 10 meters here). Then the normalized value of OPL is OPL norm The calculation formula is: The normalized OPL norm The value range is between [0, 1]. The smaller the value, the smaller the path deviation and the higher the sensitivity. According to the influence degrees of RS norm and OPL norm on the sensitivity, weights w RS and w OPL are assigned to them respectively, and w RS + w OPL = 1. For example, if it is considered that the response speed has a greater impact on the sensitivity, w RS can be set to 0.7 and w OPL to 0.3. The sensitivity value calculation formula is: S = (w RS × RS morm + w oPL × OPL norm ) × CTC. The value range of S is between [0, 1]. The smaller the value, the higher the sensitivity.

[0066] Set the sensitivity threshold according to the sensitivity value S.

[0067] For example, if S ≤ 0.3, it is defined as low sensitivity (frequent switching is allowed), and the layer cutting response ≤ 1 second; if 0.3 < S ≤ 0.7, it is defined as medium sensitivity, and the layer cutting response is 1 - 3 seconds; if S > 0.7, it is defined as high sensitivity (frequent switching is prohibited), and the layer cutting response ≥ 3 seconds. <00​​​​Specifically, according to the sensitivity threshold matching path strategy, K candidate paths are generated using Monte Carlo simulation. Each path contains a segmented altitude layer strategy. For example, 20 paths are generated under low sensitivity, and 5 paths are generated under high sensitivity. Each path contains a segmented altitude layer strategy (such as "low altitude - high altitude - low altitude").

[0070] Among them, for low sensitivity (S ≤ 0.3), the unmonitored target with the highest priority is preferentially selected, allowing high-frequency layer switching, and the path contains more altitude segments (such as "low - high - low - high"); for high sensitivity (S > 0.7), conservative paths are preferably generated to reduce altitude layer switching; for medium sensitivity (0.3 < S ≤ 0.7), stable flight is maintained, layer switching is restricted, and the path is mainly at a single altitude layer (such as low altitude or high altitude throughout the process).

[0071] S34. Calculate the dynamic environment adaptation coefficient of each candidate path and correct the path complexity index.

[0072] Specifically, the terrain altitude change is monitored in real time by lidar, and the terrain undulation change rate (ΔH) within a unit time is calculated. T is the sampling time window (10 seconds), H t is the terrain altitude value of the previous unit time, and H t+1 is H t the terrain altitude of the next unit time:

[0073]

[0074] The number of sudden changes in wind speed and wind direction is recorded by a meteorological sensor, and the mutation frequency (ΔW) within a unit time is calculated:

[0075]

[0076] Based on the terrain undulation change rate and the mutation frequency, the external environment change rate is calculated:

[0077] DEAC = w ΔH ×ΔH + w ΔW ×ΔW

[0078] Among them, the weight coefficient w ΔH = 0.5, w ΔW = 0.5 (can be adjusted according to the mission type).

[0079] According to the external environment change rate, correct the path complexity index of the monitored target on the candidate path:

[0080] CRIC i = CRI i ×(1 + α×DEAC)

[0081] Wherein, CRI is the original path complexity index, α=0.5 (environmental sensitivity factor).

[0082] S35, calculating the comprehensive path complexity of the candidate paths based on the corrected path complexity index, selecting the optimal path and adjusting the dynamic flight path.

[0083] Specifically, the corrected path complexity indexes of the monitored targets calculated on the candidate paths are added together to obtain the comprehensive path complexity, and the candidate path with the smallest comprehensive path complexity is selected as the optimal path to adjust the dynamic flight path.

[0084] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0085] By building a dynamic path planning and sensitivity adaptation mechanism for drones, the coordinated optimization of flight efficiency and environmental adaptability is achieved. The core technology lies in: adjusting sensitivity values based on the flight log's quantitative hierarchy, dynamically evaluating system sensitivity through standardized response delay and path deviation; correcting path complexity based on real-time environmental data (terrain fluctuations, sudden weather changes), and generating flexible candidate paths; and ultimately selecting the optimal path based on the overall complexity, balancing mission completion efficiency and flight safety. This significantly improves the drone's autonomous decision-making capabilities and mission execution robustness in complex dynamic scenarios.

[0086] Example 3: The fixed grouping and task allocation mode of Example 2 cannot adapt to dynamic changes quickly. When the task volume suddenly increases, the corresponding grouped drones may be delayed due to insufficient resources; when the task volume suddenly decreases, the drones will be idle and wasted. Flight path planning also does not fully consider the impact of the dynamic environment on the complexity of the path. Under a unified plan, drones may face safety risks or fail to reach the target point efficiently, and it is difficult to meet the differentiated requirements of task difficulty and urgency in different sub-areas, reducing the efficiency and quality of monitoring task execution. In order to improve the adaptability, efficiency and accuracy of monitoring tasks in complex scenarios, it is necessary to optimize and improve Example 2, and Example 3 is generated.

[0087] In some embodiments, in step S31, recording the real-time flight trajectory of the drone further includes:

[0088] S311, based on the characteristics of the monitoring area and the monitoring targets, the monitoring area is divided into several groups of drones, each group is responsible for monitoring targets in a specific sub-area.

[0089] Specifically, the image data of the monitoring area is combined with meteorological data (wind speed, visibility fluctuation frequency) to generate a three-dimensional environmental feature map. N 监测目标 is the number of sub-area monitoring targets, A 子区域is the area of the sub-region) and regional characteristics, the monitoring area is divided into n sub-regions ( N 无人机 is the total number of available drones), and each sub-area is assigned an independent group.

[0090] Apply area constraints to each subregion: Among them, A 总面积 is the total area of the monitoring area; the redundancy factor of 1.3 is used to cover the boundary overlap (20%) and the emergency buffer zone (10%), ensuring that the drone can still complete the monitoring task under complex terrain or weather conditions. For example: the total monitoring area is 100km 2 , monitoring target density 12 / km 2 , divided into 4 groups, the area of a single sub-area is ≤32.5km 2 , high-density areas (such as cities) are preferentially assigned to independent groups, and low-density areas (such as water areas) can be combined into groups.

[0091] S312: Generate a shortest path tree covering the sub-region according to the sub-region boundary constraints, and calculate the sub-region task volume.

[0092] Specifically, the boundary information of each sub-area is extracted from the three-dimensional environmental feature map, and the monitoring targets in the sub-area are used as nodes to generate the shortest path tree covering all monitoring targets in the sub-area using the Dijkstra algorithm or the Prim algorithm. The task volume of the sub-area is calculated based on the length of the shortest path tree and the number of monitoring targets. P i is the i-th monitoring target, and I is the indicator function (1 if the target is within the sub-region, 0 otherwise). A task list is created for each sub-region, recording the specific values of the task volume for each sub-region.

[0093] S313, dynamically allocate sub-area task volumes and adjust the number of drones and task allocation within the formation in real time.

[0094] Specifically, the system collects real-time drone flight data and the number and status of monitored targets, calculates the remaining workload in each sub-area, and dynamically assesses the workload for each sub-area based on this real-time monitoring data. It determines whether the workload in each sub-area exceeds or falls short of expectations, and whether there is an imbalance in task distribution. Based on the workload assessment results, it formulates a grouping adjustment strategy. It also updates the table that maps groups to sub-areas, recording relevant information after the grouping adjustments.

[0095] Among them, task load assessment indicators include task completion rate (completed tasks / total tasks) and task progress deviation (the difference between actual task progress and planned task progress). When the task completion rate falls below a certain threshold (such as 70%) or the task progress deviation exceeds a certain range (such as ±20%), it is considered that there is a problem with the task load in the sub-region and adjustments are required. The group adjustment strategy is that if the task load in a sub-region is too large and exceeds the processing capacity of the drones in its group, drone support can be deployed from other groups with smaller task loads. If the task load in a sub-region is small and drones are idle, some drones can be deployed to other groups with larger task loads.

[0096] S314: Re-plan the flight paths of the drones within the formation based on the implementation adjustment information.

[0097] Specifically, the ground control center collects relevant information about the grouping adjustments, including the number of drones within the group, their locations, task assignments, and the latest sub-area boundaries and mission requirements. This information is collated and analyzed to determine the drones and sub-areas requiring rerouting. The path complexity index is calculated and corrected for the monitoring targets within the new sub-areas. The optimal path is then selected and adjusted to serve as the drone flight path for the new sub-area.

[0098] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0099] Through dynamic regional grouping, three-dimensional environment-integrated path planning, and real-time resource adaptive adjustment, the solution achieves efficient coordination and precise coverage of drone monitoring missions. This solution divides sub-regions based on target density and regional characteristics, dynamically adjusts path costs based on terrain and weather conditions, quantifies mission load, and dynamically allocates drone resources, ensuring priority protection in high-density areas and resource integration in low-density areas. It also optimizes flight trajectories using a path complexity index, effectively improving mission completion rates and resource utilization, significantly shortening monitoring response times in complex scenarios, and delivering both flexibility and robustness.

[0100] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Various modifications and variations are readily apparent to those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. An intelligent planning method for UAV space monitoring, characterized in that: include: S1, collects multi-dimensional parameters and image data of the monitoring area and generates a task requirement priority matrix; S2, plans a preliminary flight path according to the priority matrix, obtains the mission detour coefficient and path slackness of each monitoring target according to the preliminary flight path, and calculates the path complexity index according to the mission detour coefficient and path slackness of each monitoring target; S3, divides the monitoring area into altitude layers based on the path complexity index, generates a dynamic flight path based on the altitude layers; records the real-time trajectory of the drone and calculates the completion rate of the monitoring target; Comprehensively adjust the delay time and the deviation distance generated during the altitude layer switching process, and quantify the sensitivity value corresponding to the layer adjustment; Based on the sensitive values, an elastic model that reflects the dynamic adjustment path planning strategy of the UAV is constructed, and candidate paths are generated based on the elastic model. According to the environmental dynamic adaptation coefficient of each candidate path, the path complexity index is corrected and the path is adjusted optimally.

2. The intelligent planning method for UAV space monitoring according to claim 1, characterized in that: The generation of the task requirement priority matrix specifically includes: setting the geographical scope of the area to be monitored through the unmanned aerial vehicle control system and obtaining its high-resolution image data; combining the collected multi-dimensional parameters and image data to construct a complex target recognition model and an n×m matrix task requirement priority matrix; based on the task requirement priority matrix, generating a monitoring target list; wherein n is the number of targets and m is the target weight; filling the matrix with priority information of the weights and sorting them to form the task requirement priority matrix.

3. The intelligent planning method for UAV space monitoring according to claim 1, characterized in that: The calculation formula for the path complexity index is as follows: CRI i Represents the path complexity index of the i-th monitoring target; RP i Indicates the task requirement weight priority of the i-th monitoring target; RNC i represents the mission detour coefficient of the i-th monitoring target; PS i represents the path relaxation of the i-th monitoring target, that is, the range of distance allowed to deviate from the preset path; N represents the total number of monitoring targets.

4. The intelligent planning method for UAV space monitoring according to claim 3, characterized in that: The path complexity index RNC of the i-th monitoring target i Indicates the degree to which the actual path deviates from the preset path due to terrain or obstacles. The actual required flight distance is the distance that the UAV actually flies to cover the monitoring target, and the direct flight distance is the straight-line distance from the starting point to the monitoring target. The path relaxation PS of the i-th monitoring target is i Indicates the range of distance allowed to deviate from the preset path, PS i =Basic relaxation × (1 + RNC i ), the basic slack represents the maximum base distance allowed to deviate from the preset path under ideal conditions.

5. The intelligent planning method for UAV space monitoring according to claim 1, characterized in that: The optimal path adjustment specifically includes: adding the corrected path complexity indexes of the monitoring targets calculated on the candidate paths to obtain a comprehensive path complexity, selecting the candidate path with the smallest comprehensive path complexity as the optimal path to adjust the dynamic flight path; wherein the calculation formula for the corrected path complexity index of the monitoring targets on the candidate paths is: CRIC i =CRI i ×(1+α×DEAC), CRI is the original path complexity index, α is the environmental sensitivity factor, and DEAC is the rate of change of the external environment.

6. The intelligent planning method for UAV space monitoring according to claim 5, characterized in that: The steps for obtaining the rate of change of the external environment are as follows: monitor the change of terrain height in real time through the laser radar, and calculate the rate of change of terrain undulation per unit time. The number of wind speed and wind direction mutations is recorded by meteorological sensors, and the mutation frequency per unit time is calculated The external environment change rate DEAC=w is calculated based on the terrain undulation change rate and mutation frequency ΔH ×ΔH+w ΔW ×ΔW; where the weight coefficient w ΔH =0.5, w ΔW =0.5, T is the sampling time window, H t is the terrain height value of the previous unit time, H t+1 It is H t The height of the terrain at the last unit of time.

7. The intelligent planning method for UAV space monitoring according to claim 5, characterized in that: The steps for selecting the minimum candidate path are as follows: extract the time t of each altitude layer switching instruction from the UAV flight log 指令,i and the actual execution time t 执行,i , calculate the actual time taken to switch altitude levels n is the number of historical adjustments; compare the actual path after switching with the dynamic flight path, and calculate the actual path L caused by the switch 实际,i With the ideal path L 理想,i The Euclidean distance difference n is the historical adjustment times; RS and OPL are standardized to obtain RS norm and OPL norm , substitute the sensitivity value into the calculation formula: S=(w RS ×RS norm +w OPL ×OPL norm )×CTC; where w RS and w OPL is the weight coefficient; based on this, the sensitivity threshold is set, the corresponding path strategy is matched, and K candidate paths with segmented altitude layer strategies are generated using Monte Carlo simulation.

8. The intelligent planning method for UAV space monitoring according to claim 1, characterized in that: The real-time trajectory of drones is recorded as follows: drones are divided into groups according to the characteristics and objectives of the monitoring area, and each group is responsible for monitoring tasks in a specific sub-area; the shortest path tree is generated based on the sub-area boundary to calculate the task volume and dynamically allocate tasks; the number of drones and tasks in the group are adjusted in real time, and the flight paths of the drones in the group are replanned.

9. The intelligent planning method for UAV space monitoring according to claim 8, characterized in that: Each group is responsible for monitoring tasks in a specific sub-area, including: combining the image data collected in the monitoring area with meteorological data to generate a three-dimensional environmental feature map; The monitoring area is divided into n sub-areas according to the number of images in the area. N 无人机 Assign independent groups to each sub-area for the total number of available drones; set area constraints for each sub-area Among them, the redundancy factor 1.3 is used to cover boundary overlap and emergency buffer; among them, N 监测目标 is the number of sub-area monitoring targets, A 子区域 is the area of the subregion, A 总面积 is the total area of the monitoring area.

10. The intelligent planning method for UAV space monitoring according to claim 8, characterized in that: The dynamic allocation of tasks specifically includes: extracting the boundary information of each sub-area from the three-dimensional environmental feature map, using the monitoring targets in the sub-area as nodes, and using the Dijkstra algorithm to generate a shortest path tree covering all monitoring targets in the sub-area; calculating the task volume of the sub-area based on the length of the shortest path tree and the number of monitoring targets. P i is the i-th monitoring target, and I is the indicator function; a task list is established for each sub-area, and the specific values of the task volume of each sub-area are recorded.

Images