A Natural Resource Monitoring, Evidence Collection, and Scheduling Method Based on Tower UAVs and Their Nests

Abstract

This invention relates to the field of UAV path planning technology and discloses a natural resource monitoring and evidence-based scheduling method based on UAVs and their nests. The method includes: calculating the nest failure probability value using a Cox proportional hazards model; extracting performance parameters of heterogeneous UAV models; calculating and classifying the task complexity of map features; dividing the nest coverage area using nest reliability as a weighting factor and establishing a candidate nest mapping table; acquiring the map features to be taken over and calculating the takeover urgency when a nest fails; calculating the reachability time and return margin of candidate UAV models; generating a model-takeover task suitability matrix specific to the failure scenario; solving for the optimal allocation using a Hungarian algorithm with time window constraints; and generating differentiated failure takeover path schemes for heterogeneous UAV models. This invention solves the problem of unreasonable task takeover caused by nest failure in existing technologies.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) path planning technology, and more specifically, to a method for monitoring, evidence collection, and scheduling natural resources based on UAVs and their nests. Background Technology

[0002] In the application of the tower-based drone nest network for land change surveys and evidence collection, different types of heterogeneous drones were deployed, including long-endurance fixed-wing drones and highly maneuverable multi-rotor drones. These models exhibit significant differences in performance parameters such as endurance, flight speed, and hovering ability. The drone nests are deployed in outdoor environments, making them susceptible to failure due to factors such as lightning strikes, equipment aging, and communication outages.

[0003] When existing technologies divide the coverage area using weighted Voronoi diagrams, they assume that all nests are continuously available during mission execution and that the UAVs have the same performance parameters. When a nest fails, the emergency plan will transfer the mission of the failed nest to the nearest adjacent nest for takeover.

[0004] The above technical solutions have the following technical problems: First, the potential failure risk of the drone nest was not considered during the planning stage, and the coverage area division results are sensitive to single-point failures. Driving too many tasks on high-failure-risk drone nests exacerbates the scope of failure impact. Second, the emergency plan does not distinguish the performance differences of the takeover drone models. It may assign long-distance takeover tasks to multi-rotor drones, resulting in insufficient endurance and inability to return, or assign complex terrain tasks that require hovering and shooting to fixed-wing drones, resulting in the inability to complete the evidence collection. Third, the conventional model-task adaptability calculation is designed for normal patrol scenarios and does not consider the constraints of takeover distance and time urgency in emergency failure scenarios, making it difficult to meet the evidence collection time limit requirements for emergency map patches. Summary of the Invention

[0005] This invention provides a natural resource monitoring and evidence-gathering scheduling method based on tower drones and drone nests, solving the technical problem in related technologies that it is impossible to quickly schedule heterogeneous drones to take over tasks when the drone nest fails.

[0006] This invention provides a method for natural resource monitoring and evidence collection scheduling based on tower-mounted UAVs and their nests, including: Obtain historical operation logs, equipment service duration, recent maintenance records and current environmental monitoring data for each cell, and use a survival analysis model to calculate the cell failure probability value of each cell within the planned time window; Obtain a list of heterogeneous drone models deployed in each drone nest, extract the range, maximum flight speed, minimum turning radius and hovering ability of each model, and establish a model performance parameter table; Obtain the geometric features of the patch to be proven and the surrounding terrain data, calculate the patch task complexity for each patch, and divide the patches into complex tasks that require hovering for shooting and simple tasks that can be shot over. Nest reliability is used as a weighting factor in the weighted Voronoi diagram to divide the nest coverage area, the reachability of boundary patches to adjacent nests is calculated, and a candidate nest mapping table for boundary patches is established. When a fault signal is detected in a cell, the set of patches to be taken over within the coverage area of ​​the faulty cell is obtained, and the urgency of takeover is calculated based on the time limit for providing evidence for each patch to be taken over. Obtain the current availability status of drones in adjacent available nests in the candidate nest mapping table, and calculate the reach time and return margin to each patch to be taken over for each candidate drone type; Based on the urgency of takeover, the complexity of the map patch task, the reachability time, and the performance parameters of the candidate aircraft models, an aircraft model-takeover task adaptability matrix is ​​generated. The Hungarian algorithm with time window constraints is used to solve the model-takeover task fit matrix and output the optimal model-task allocation result. Based on the optimal allocation result, candidate aircraft type-differentiated takeover route segments are generated, and after time-series arrangement and conflict detection, fault takeover path schemes are output.

[0007] Furthermore, the survival analysis model is a Cox proportional hazards model. The inputs of the Cox proportional hazards model include: equipment service duration as a time variable, frequency of abnormal events in historical operation logs and number of fault repairs in recent maintenance records as covariates, and thunderstorm warning level and temperature and humidity anomaly degree in current environmental monitoring data as time-varying covariates. The output of the Cox proportional hazards model is the cumulative risk probability of each nest failing within the planned time window.

[0008] Furthermore, the nest failure probability value is calculated using the baseline survival function, the planning time window length, the values ​​of each covariate, and the regression coefficients obtained by fitting historical data. The baseline survival function refers to the probability that the nest will not fail within the planning time window under the baseline condition that all covariates are zero.

[0009] Furthermore, the geometric features of the patch include the patch area, the patch shape index, and the patch boundary tortuosity; the surrounding terrain data includes terrain undulation, vegetation occlusion rate, and building distribution density; the patch shape index refers to the ratio of the patch perimeter to the perimeter of a circle of the same area; the terrain undulation refers to the difference between the maximum and minimum elevation values ​​within the area where the patch is located.

[0010] Furthermore, the criteria for classifying the complexity of the map patch task are as follows: when the shape index of the map patch exceeds the shape threshold, or the terrain undulation exceeds the undulation threshold, or the vegetation occlusion rate exceeds the occlusion threshold, the map patch is classified as a complex task; otherwise, it is classified as a simple task.

[0011] Furthermore, the nest reliability is the complement of the nest failure probability value; the weight of the weighted Voronoi diagram is composed of nest reliability and UAV endurance; the boundary patch refers to the patch located in the boundary area of ​​the coverage area of ​​adjacent nests; the accessibility is determined by judging whether the round-trip distance from the boundary patch to the adjacent nest is less than the endurance of the UAV deployed in the adjacent nest.

[0012] Furthermore, the urgency of takeover is calculated by the reciprocal of the difference between the evidence submission deadline of the plot to be taken over and the current time; the reachability time is calculated by dividing the distance between the plot to be taken over and the nest by the maximum flight speed of the candidate aircraft; and the return trip margin is calculated by subtracting the round-trip distance and the mileage consumed in mission execution from the range of the candidate aircraft.

[0013] Furthermore, the elements of the aircraft type-takeover task adaptability matrix are calculated by weighted summation of the normalized values ​​of takeover urgency, task complexity matching degree, reachability time, and return margin; wherein, the task complexity matching degree is one when the task to be taken over is a complex task and the candidate aircraft type has hovering capability, one when the task to be taken over is a simple task, and zero in other cases; the weight coefficient of the takeover urgency dimension is set to be greater than the weight coefficients of other dimensions.

[0014] Furthermore, the Hungarian algorithm with time window constraints includes the following steps: The patches to be taken over are sorted in descending order of urgency, and all candidate models and patches to be taken over are initialized to an unmatched state. Perform row and column reduction operations on the model-takeover task compatibility matrix; Starting with the map patch with the highest urgency of takeover, candidate models with zero fit and meeting the time window constraints are searched for each map patch to be taken over in turn; If a candidate model that meets the conditions is found, a matching relationship is established; if not, the search is repeated after expanding the set of zero elements by adjusting the dual variable. When all the patches to be taken over have been matched or there are no feasible candidate models for the remaining patches to be taken over, output the current matching result; The time window constraint is the time when the candidate aircraft arrives at the patch to be taken over plus the task execution time not exceeding the evidence submission deadline of the patch to be taken over.

[0015] This invention provides a natural resource monitoring, evidence collection, and scheduling system based on tower-mounted unmanned aerial vehicles (UAVs) and their nests, comprising: The nest failure probability calculation module is used to obtain the historical operation logs, equipment service duration, recent maintenance records and current environmental monitoring data of each nest, and use the survival analysis model to calculate the nest failure probability value of each nest within the planned time window; The model performance extraction module is used to obtain a list of heterogeneous drone models deployed in each drone nest, extract the range, maximum flight speed, minimum turning radius and hovering ability of each model, and establish a model performance parameter table; The image patch classification module is used to obtain the geometric features of the image patches to be proven and the surrounding terrain data, calculate the image patch task complexity for each image patch, and classify the image patches into complex tasks that require hovering for shooting and simple tasks that can be shot over. The coverage area partitioning module is used to partition the nest coverage area by using nest reliability as a weighting factor in the weighted Voronoi diagram, calculate the reachability of boundary patches to adjacent nests, and establish a candidate nest mapping table for boundary patches. The takeover urgency calculation module is used to obtain the set of takeover patches within the coverage area of ​​the faulty nest when a nest fault signal is detected, and to calculate the takeover urgency based on the evidence collection time limit for each takeover patch. The candidate drone evaluation module is used to obtain the current availability status of drones in adjacent available drone nests in the candidate drone nest mapping table, and calculate the reachability time and return margin to each plot to be taken over for each candidate drone model. The adaptability matrix generation module is used to generate an aircraft type-takeover task adaptability matrix based on takeover urgency, map task complexity, reachability time, and candidate aircraft performance parameters. The task allocation module is used to solve the machine type-takeover task fit matrix using the Hungarian algorithm with time window constraints, and output the optimal allocation result of machine type-task. The path generation module is used to generate candidate aircraft type differentiated takeover route segments based on the optimal allocation results, and outputs fault takeover path schemes after time sequence arrangement and conflict detection.

[0016] The beneficial effects of this invention are as follows: This invention quantifies nest failure risk using a survival analysis model and uses nest reliability as a weighting factor for coverage area partitioning, enabling high-failure-risk nests to obtain smaller coverage areas, reducing their workload and minimizing the impact of failures. By synchronously establishing a candidate nest mapping table for boundary patches during the planning phase, a pre-set candidate resource pool is provided for failure response, shortening the failure response time. A dedicated aircraft type-takeover task fit matrix is ​​constructed for failure scenarios, incorporating takeover urgency and patch task complexity as independent dimensions into the fit calculation, ensuring that the fit assessment reflects time constraints and task characteristic constraints in emergency scenarios. The Hungarian algorithm with time window constraints is used to solve the allocation problem, pruning and eliminating allocation schemes that do not meet the evidence submission deadline, ensuring that all patches to be taken over can complete evidence submission before the deadline. By generating aircraft-differentiated takeover flight path segments, fixed-wing UAVs use high-curvature flight paths to perform long-distance simple tasks, while multi-rotor UAVs use precise hovering paths to perform complex tasks, achieving reasonable scheduling of heterogeneous UAV resources in failure takeover scenarios. Attached Figure Description

[0017] Figure 1 This is a flowchart of the natural resource monitoring, evidence collection, and scheduling method based on tower drones and drone nests according to the present invention; Figure 2 This is a comparative analysis chart of the failure probability and reliability of the machine nest in this invention; Figure 3 This is a comparison chart of the performance parameters of the heterogeneous UAV models of this invention; Figure 4 This is a task complexity feature distribution map of the map to be taken over according to the present invention; Figure 5 This is a sorting diagram of the urgency of the patch to be connected according to the present invention; Figure 6 This is a feasibility assessment diagram of the return trip margin for candidate models of the present invention; Figure 7 This is a heat map showing the model-to-task compatibility of the present invention. Detailed Implementation

[0018] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, some features described in the examples may be combined in other examples.

[0019] At least one embodiment of the present invention discloses a natural resource monitoring and evidence-gathering scheduling method based on tower drones and drone nests, such as... Figure 1 As shown, it includes the following steps: Step 1: Computer nest failure probability value.

[0020] The historical operation logs, equipment service duration, recent maintenance records, and current environmental monitoring data of each nest are obtained, and the Cox proportional hazards model is used to calculate the nest failure probability value of each nest within the planned time window.

[0021] Furthermore, the inputs to the Cox proportional hazards model are the service life of the equipment in the cell as a time variable, the frequency of abnormal events in the historical operation log and the number of fault repairs in the recent maintenance records as covariates, and the thunderstorm warning level and the degree of temperature and humidity anomalies in the current environmental monitoring data as time-varying covariates; the output of the Cox proportional hazards model is the cumulative risk probability of each cell failing within the planned time window.

[0022] Furthermore, the nest The probability value of the cellular failure Calculated using the following formula:

[0023] in, As the baseline survival function, To plan the length of the time window, For the nest The values ​​of each covariate, These are the regression coefficients obtained by fitting historical data. The total number of covariates. This represents an exponential function.

[0024] Furthermore, the length of the aforementioned planning time window This refers to the time span from the current moment to the end of the current mission planning cycle, in hours; the aforementioned baseline survival function This refers to the nest's time-dependent behavior under the baseline condition that all covariates are zero. The probability that no failure will occur.

[0025] Furthermore, during the training of the Cox proportional hazards model, historical nest covariate data are used as input samples, and the labels of actual failures are used as supervision signals. The maximum likelihood estimation method is used to fit the regression coefficients.

[0026] Step 2: Extract the performance parameters of the drone model.

[0027] Obtain a list of heterogeneous drone models deployed in each drone nest, extract performance parameters such as range, maximum flight speed, minimum turning radius and hovering ability for each model, and establish a model performance parameter table.

[0028] Furthermore, the hovering capability mentioned above refers to the level of a drone's ability to maintain a stationary flight state at a designated location. The hovering capability level of a fixed-wing drone is zero, indicating that it does not have hovering capability. The hovering capability level of a multi-rotor drone is divided into multiple discrete levels based on its wind resistance level and positioning accuracy.

[0029] Furthermore, the aforementioned minimum turning radius refers to the minimum radius of curvature required for a drone to change its course while maintaining its flight speed. The minimum turning radius of a fixed-wing drone is usually greater than that of a multi-rotor drone.

[0030] Step 3: Calculate the task complexity of the polygons and classify the polygons.

[0031] Obtain the geometric features of the patch to be used for evidence and the surrounding terrain data. Calculate the task complexity of each patch based on the geometric features and surrounding terrain data, and divide the patches into complex tasks that require hovering for shooting and simple tasks that can be shot over.

[0032] Furthermore, the geometric features of the aforementioned patches include parameters such as patch area, patch shape index, and patch boundary tortuosity; the surrounding terrain data includes parameters such as terrain undulation, vegetation occlusion rate, and building distribution density of the area where the patch is located.

[0033] Furthermore, the aforementioned patch shape index refers to the ratio of the patch perimeter to the perimeter of a circle of the same area, used to quantify the irregularity of the patch boundary. The larger the ratio, the more irregular the patch shape.

[0034] Furthermore, the aforementioned topographic relief refers to the difference between the maximum and minimum elevation values ​​within the area where the map patch is located, used to quantify the vertical variation of the terrain.

[0035] Furthermore, the criteria for classifying the complexity of a patch task are as follows: when the patch shape index exceeds the shape threshold, or the terrain undulation exceeds the undulation threshold, or the vegetation occlusion rate exceeds the occlusion threshold, the patch is determined to require hovering photography to obtain multi-angle evidence images and is classified as a complex task; otherwise, it is classified as a simple task that can be completed by flying over and taking photos.

[0036] Furthermore, the aforementioned shape threshold is determined according to the criteria for judging irregular patches in the technical specifications for land change surveys, with a value range of 1.5 to 2.0; the aforementioned undulation threshold is determined according to the slope range of hilly terrain in the terrain classification standards, with a value range of 30 meters to 50 meters; the aforementioned occlusion threshold is determined according to the visibility index in the image acquisition quality requirements, with a value range of 40% to 60%.

[0037] Step 4: Divide the nest coverage area and establish a candidate nest mapping table.

[0038] Nest reliability is used as an additional weighting factor in the weighted Voronoi diagram to divide the nest coverage area, obtain the boundary patches of each nest coverage area, calculate the reachability of the boundary patches to the corresponding nests in adjacent nest coverage areas, and establish a candidate nest mapping table for the boundary patches.

[0039] Furthermore, the aforementioned cell reliability refers to the complement of the cell failure probability value, i.e. ,in, For the nest Nest reliability, The value is the probability of a fault in the cell calculated in step 1.

[0040] Furthermore, the weights of the aforementioned weighted Voronoi diagram are composed of both nest reliability and drone endurance. The higher the nest reliability and the longer the endurance of the deployed drones, the greater the weight the nest receives in the nest coverage area division, and the wider the corresponding nest coverage area.

[0041] Furthermore, the aforementioned boundary patches refer to patches located in the boundary area between adjacent nest coverage areas. The distance of these boundary patches to the nests in their respective nest coverage areas is similar to the distance to the nests in adjacent nest coverage areas, and the boundary patches are feasible to be taken over by adjacent nests.

[0042] Furthermore, the aforementioned accessibility refers to the effective service capability from the boundary patch to the adjacent nest, which is determined by calculating the straight-line distance from the boundary patch to the adjacent nest. Obtain the range of each aircraft type deployed in adjacent nests. If a model meets the requirements If the round-trip distance is less than the range, then the boundary patch is determined to be reachable from the adjacent nest, and the adjacent nest and the corresponding reachable aircraft type are recorded in the boundary patch's alternative nest mapping table.

[0043] Furthermore, to pre-define a candidate resource pool before a failure occurs and shorten the failure response time, a candidate nest mapping table is synchronously established and persistently stored during the nest coverage area partitioning phase. When the nest status or UAV availability changes, the affected boundary patch entries are incrementally updated to avoid the computational overhead of rebuilding the entire table.

[0044] Step 5: Obtain the patch to be taken over and calculate the urgency of takeover.

[0045] When a nest fault signal is detected, the set of patches to be taken over within the coverage area of ​​the faulty nest is obtained, and the urgency of takeover of each patch is calculated based on the evidence collection time limit for each patch.

[0046] Furthermore, the aforementioned urgency of takeover refers to the required speed of takeover response for the patch to be taken over after a fault occurs, and is calculated using the following formula:

[0047] in, Map to be taken over The urgency of the takeover Map to be taken over The deadline for submitting evidence, The current time is displayed in hours. A higher urgency value indicates that the area to be taken over is closer to the deadline for providing evidence, and therefore requires priority in takeover allocation.

[0048] Step 6: Calculate the reachability and return margin of the candidate aircraft models.

[0049] Obtain the current availability status of drones in adjacent available drone nests in the candidate drone nest mapping table, and calculate the reach time and return margin to each plot to be taken over for each candidate drone type.

[0050] Furthermore, the aforementioned reachable time refers to the estimated flight time from the candidate aircraft's takeoff from its respective nest to the patch to be taken over, calculated by dividing the distance between the patch to be taken over and the nest by the maximum flight speed of the candidate aircraft.

[0051] Furthermore, the aforementioned return range refers to the remaining endurance of the candidate aircraft after completing the evidence collection task for the areas to be taken over and returning to the nest, calculated using the following formula:

[0052] in, For the nest Candidate models For the plots to be taken over The return journey margin, Candidate models The driving range, For the nest Map to be taken over distance, Map to be taken over The mileage consumed during task execution is measured in kilometers. When this is the case, it indicates that the candidate aircraft model has the endurance to complete the takeover mission of the area to be taken over.

[0053] Furthermore, the aforementioned tasks consume mileage. This refers to the flight distance of a drone while performing a photography mission over a plot of land to be taken over. For simple missions, the flight distance is the straight-line distance of flying over the plot of land to be taken over; for complex missions, the flight distance is the sum of the distances traveled between multiple hovering points.

[0054] Step 7: Generate the model-takeover task adaptation matrix for the fault scenario.

[0055] Based on the urgency of takeover, the complexity of the map task, the reachability time, and the performance parameters of the candidate models, a model-takeover task adaptability matrix is ​​generated specifically for the fault scenario.

[0056] Furthermore, the elements of the aforementioned model-takeover task adaptability matrix... Indicates nest Candidate models Treatment of takeover map The degree of takeover compatibility is calculated using the following formula:

[0057] in, Map to be taken over The normalized value of the urgency of the takeover; Candidate models Treatment of takeover map The matching degree of the task complexity of the map patch, when the map patch to be taken over For complex tasks and candidate models The value is one when hovering capability is available, and one when the image to be taken over is... The value is 1 for simple tasks and 0 for other cases. This is the normalized value for reachability time; This is the normalized value of the return trip margin; The weight coefficients for each dimension satisfy... .

[0058] Furthermore, to highlight the importance of time urgency in emergency response scenarios, the weighting coefficient for the takeover urgency dimension is adjusted. Set the value to be greater than the weight coefficients of other dimensions so that the map patches to be taken over that are closer to the deadline for providing evidence can be given priority in allocation.

[0059] Furthermore, the normalized value of the aforementioned takeover urgency By assessing the urgency of taking over each patch of map to be taken over Dividing by the maximum takeover urgency value among all the plots to be taken over, the above achievable time normalized value is obtained. By candidate models Map to be taken over The reachability time is obtained by dividing the reachability time of all candidate aircraft types to the patch to be taken over; the above return margin normalized value. By using return trip margin Divide by candidate models range get.

[0060] Step 8: Solve the model-takeover task adaptation matrix and output the model-task allocation results.

[0061] The Hungarian algorithm with time window constraints is used to solve the aircraft type-takeover task suitability matrix, which assigns long-distance simple tasks to long-haul aircraft types and short-distance complex tasks to high-mobility aircraft types, and outputs the optimal aircraft type-task allocation result.

[0062] Furthermore, the input to the aforementioned Hungarian algorithm with time window constraints is the aircraft type-takeover task fit matrix and the time window constraints for each target map patch. The output of the Hungarian algorithm with time window constraints is the one-to-one optimal matching result between candidate aircraft types and target map patches. The Hungarian algorithm with time window constraints adds a time window feasibility check to the standard Hungarian algorithm, verifying whether the candidate assignments meet the time window constraints each time an augmenting path is attempted to be extended.

[0063] Furthermore, the time window constraint is: the time when the candidate aircraft arrives at the patch to be taken over plus the task execution time shall not exceed the evidence submission deadline of the patch to be taken over, i.e. ,in, For the current moment, For reachable time, Map to be taken over The estimated execution time of the task. The time units are all hours, representing the deadline for providing evidence. Allocation schemes that do not meet the time window constraints are pruned and eliminated during the iteration of the Hungarian algorithm with time window constraints.

[0064] Furthermore, the estimated execution time of the aforementioned tasks This refers to the time required for the drone to complete the shooting task above the area to be taken over. For simple tasks, the estimated mission execution time is the shooting time during the flight over the area to be taken over; for complex tasks, the estimated mission execution time includes the hovering shooting time at multiple hovering points and the movement time between hovering points.

[0065] Furthermore, the specific steps of the Hungarian algorithm with time window constraints are as follows: Step 8.1: Sort the patches to be taken over in descending order of urgency, and initialize all candidate models and patches to be taken over to an unmatched state; Step 8.2: Perform row and column reduction operations on the model-takeover task adaptation matrix to ensure that each row has at least one zero element and each column has at least one zero element; Step 8.3: Starting with the patch to be taken over with the highest urgency, find candidate models with zero fit and that meet the time window constraints for each patch to be taken over in turn; Step 8.4: If a candidate model that meets the conditions is found, establish a matching relationship; if not, expand the set of zero elements by adjusting the dual variable and repeat step 8.3. Step 8.5: When all the patches to be taken over have been matched or there are no feasible candidate models for the remaining patches to be taken over, output the current matching result as the optimal allocation scheme.

[0066] Furthermore, the above-mentioned row and column reduction operation refers to subtracting the minimum value of each row of the aircraft type-takeover task fit matrix, and then subtracting the minimum value of each column. Through the row and column reduction operation, the allocation problem is transformed into the problem of finding the optimal match at the zero element position of the aircraft type-takeover task fit matrix.

[0067] Furthermore, the dual variables mentioned above refer to the auxiliary variables maintained for each patch to be taken over and each candidate aircraft type in the Hungarian algorithm with time window constraints. By adjusting the values ​​of the dual variables, new zero elements can be generated in the aircraft type-takeover task fit matrix, thereby expanding the search space for feasible matching.

[0068] Furthermore, to improve the solution efficiency, before executing the Hungarian algorithm with time window constraints, the patches to be taken over are sorted in descending order of takeover urgency, and matching candidate models are found first for patches with high takeover urgency, reducing the number of conflict adjustments in subsequent iterations.

[0069] Step 9: Generate a fault takeover path scheme for heterogeneous machine types.

[0070] Based on the optimal allocation result, differentiated takeover route fragments for candidate aircraft types are generated. Fixed-wing UAVs adopt high curvature flight paths, while multi-rotor UAVs adopt precise hovering paths. After time-series arrangement and conflict detection, differentiated fault takeover path schemes for heterogeneous aircraft types are output.

[0071] Furthermore, the aforementioned high curvature flight path refers to a continuous flight path with a curvature radius of not less than the minimum turning radius of a fixed-wing UAV, which, when passing over the area to be managed, is used to obliquely capture and collect evidence images along the long axis of the area to be managed.

[0072] Furthermore, the aforementioned precise hovering path refers to a segmented path in which a multi-rotor UAV flies to the designated location of the patch to be taken over, enters a hovering state, takes multi-directional photos according to a preset angle sequence, and then flies to the next location.

[0073] Furthermore, the aforementioned timing arrangement refers to staggering the takeoff times of multiple drones taking off from the same nest to avoid nest operation conflicts caused by simultaneous takeoffs and landings.

[0074] Furthermore, the aforementioned conflict detection refers to checking whether there is spatial overlap between the flight paths of UAVs taking off from different nests. When a conflict is detected, the potential flight conflict is eliminated by adjusting the flight altitude layer or inserting a waiting segment.

[0075] Furthermore, in order to fully utilize the long-range endurance advantage of fixed-wing UAVs, multiple spatially adjacent long-distance simple task take-over patches are linked together into a single-flight multi-point overflight route when the take-over route segment is generated, reducing the number of take-offs and landings of fixed-wing UAVs and increasing the number of tasks covered by a single flight.

[0076] This implementation quantifies nest failure risk using a survival analysis model and uses nest reliability as a weighting factor for nest coverage area partitioning. This results in nests with high failure risk having a smaller nest coverage area during the nest coverage area partitioning stage, thereby reducing the workload of nests with high failure risk. When a nest with high failure risk fails, the number of patches to be taken over is reduced accordingly, thus narrowing the scope of the failure's impact.

[0077] This implementation provides a pre-set candidate resource pool for fault response by synchronously establishing a candidate nest mapping table for boundary patches during the nest coverage area division stage. When a nest fault occurs, the candidate nest mapping table can be directly queried to obtain candidate takeover nests and reachable candidate models, avoiding the delay overhead of traversing all nests to calculate reachability after a fault.

[0078] This implementation constructs a model-takeover task adaptability matrix specifically for fault scenarios, incorporating takeover urgency and map task complexity as independent dimensions into the adaptability calculation. This allows the adaptability assessment to reflect the impact of time constraints and task characteristic constraints in emergency scenarios, overcoming the problem that conventional model-task adaptability matrices only consider parameters for normal patrol scenarios and are not applicable to emergency scenarios.

[0079] This implementation uses the Hungarian algorithm with time window constraints to solve the allocation problem. During the iteration process of the Hungarian algorithm with time window constraints, allocation schemes that do not meet the evidence submission deadline are pruned and eliminated. This ensures that all the map patches to be taken over in the output optimal allocation result can complete the evidence submission before the deadline, avoiding the time limit exceeding the deadline that may be caused by the allocation first and verification later mode.

[0080] This implementation method generates differentiated takeover flight path segments for candidate aircraft models, enabling fixed-wing UAVs to use high-curvature flight paths that conform to their flight characteristics when performing simple long-distance tasks, and multi-rotor UAVs to use precise hovering paths for multi-angle shooting when performing complex tasks. This fully leverages the performance advantages of each candidate aircraft model and achieves reasonable scheduling of heterogeneous UAV resources in fault takeover scenarios.

[0081] A provincial-level natural resources survey center deployed a network of 12 iron towers within its jurisdiction for evidence collection during the 20XX land use change survey. The jurisdiction covers approximately 3,600 square kilometers, encompassing plains, hills, and mountains. Each tower deployed a hybrid type of tower. to The No. 1 aircraft nest is equipped with long-endurance fixed-wing UAVs (with a range of 120 kilometers), mainly responsible for plain areas; to The No. 1 aircraft nest is equipped with highly mobile multi-rotor UAVs (with a range of 45 kilometers), mainly responsible for hilly and mountainous areas; to The No. 1 nest is equipped with a hybrid aircraft type.

[0082] On the morning of January 22, 20XX, the monitoring system detected... Due to a lightning strike, the communication module of drone nest #1 failed. Within the nest's coverage area, there are 23 map patches awaiting takeover verification, of which 8 are considered emergency patches with less than 6 hours remaining before the verification deadline. The system needs to quickly allocate these patches to heterogeneous drones in adjacent available nests to complete the takeover verification process.

[0083] Example of implementing core step 1: Computer nest failure probability value.

[0084] System Acquisition No. 1 nest and its adjacent nests , , Based on historical operating data, the probability of failure for each nest in the next 24 hours is calculated using the Cox proportional hazards model.

[0085] Table 1 Input data for step 1:

[0086] Table 2 Output data for step 1:

[0087] for The calculation process for the failure probability value of the No. 1 machine nest is as follows:

[0088] The regression coefficients are summed. We obtain this by weighting the covariates:

[0089] Figure 2 The comparison of failure probability and reliability of different cell types is presented.

[0090] Example of implementing core step 2: Extracting the performance parameters of the drone model.

[0091] The system extracts the performance parameters of drone models deployed in adjacent available drone nests from the model management database and establishes a model performance parameter table.

[0092] Table 3 Output data for step 2:

[0093] Figure 3 It shows a comparison of the performance parameters of different drone models.

[0094] Implementation example of core step 3: Calculate the complexity of the mosaic task and classify the mosaics.

[0095] System Acquisition The geometric features and surrounding terrain data of 23 map patches awaiting takeover within the coverage area of ​​the No. 1 aircraft nest were analyzed to calculate task complexity and classify them. Four emergency map patches were used as examples: Table 4 shows the input and output data for step 3:

[0096] Classification criteria (shape threshold = 1.6, undulation threshold = 35 meters, occlusion threshold = 45%): Shape index 1.3 < 1.6, undulation 18 < 35, occlusion 25% < 45%, classified as a simple task. The task is classified as complex if the fluctuation is 42 > 35 and the occlusion rate is 58% > 45%. The shape index is 1.8 > 1.6, indicating a complex task. All metrics did not exceed the threshold, therefore the task was classified as simple. Figure 4 The distribution of the complexity characteristics of the map patches to be taken over is shown.

[0097] Implementation examples of core steps 4 and 5: Establishing an alternative mapping table and calculating the urgency of takeover.

[0098] The system has established a candidate nest mapping table for boundary patches during nest coverage area delineation. When a nest is detected... When the No. 1 unit experiences a fault (current time is 10:30 on January 22, 20XX), the system extracts the map patch to be taken over and calculates the urgency of the takeover.

[0099] Table 5 shows the output data for steps 4 and 5:

[0100] Example of taking over urgency calculation ( ):

[0101] Figure 5 The document displays the urgency of taking over the map patches to be taken over and the remaining time.

[0102] Example of implementing core step 6: Calculate the reachability time and return margin of candidate aircraft models.

[0103] The system retrieves candidate nests from the alternative nest mapping table and calculates the reachability time and return trip margin from each candidate aircraft type to the area to be taken over. Taking the image as an example: Table 6 Output data for step 6:

[0104] Example of calculating return margin ( Nest model):

[0105] because Nest The aircraft model has a negative return range (-2.4 km), indicating that it does not have the endurance to complete the takeover mission for this map patch and will be excluded from subsequent compatibility calculations.

[0106] Figure 6 The return trip capacity of the candidate aircraft models is displayed.

[0107] Implementation examples of core steps 7 and 8: generating the fitness matrix and solving for the optimal allocation.

[0108] The system generates an aircraft type-takeover task adaptability matrix based on takeover urgency, task complexity, reachability time, and return margin, and solves it using a Hungarian algorithm with time window constraints. The weighting coefficients are set to... , , , .

[0109] Table 7 shows the fitness matrix and allocation results for steps 7 and 8:

[0110] Image - Fit calculation process: Takeover urgency normalization:

[0111] Task complexity matching degree: (Complex tasks with hovering capability) Normalized reachability time:

[0112] Return trip margin normalization:

[0113]

[0114] - right The compatibility is 0 because the hovering capability level of the fixed-wing drone is 0, which makes it unable to complete the hovering shooting requirements of complex tasks. Even if other indicators are good, they are excluded.

[0115] Figure 7 This demonstrates the compatibility between the aircraft model and the takeover mission.

[0116] The embodiments of the present invention have been described above. However, the embodiments are not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make more equivalent embodiments under the guidance of the present embodiments, and all of them are within the protection scope of the present embodiments.

Claims

1. A method for monitoring, evidence collection, and scheduling natural resources based on tower-mounted unmanned aerial vehicles (UAVs) and their nests, characterized in that, Includes the following steps: Obtain historical operation logs, equipment service duration, recent maintenance records and current environmental monitoring data for each cell, and use a survival analysis model to calculate the cell failure probability value of each cell within the planned time window; Obtain a list of heterogeneous drone models deployed in each drone nest, extract the range, maximum flight speed, minimum turning radius and hovering ability of each model, and establish a model performance parameter table; Obtain the geometric features of the patch to be proven and the surrounding terrain data, calculate the patch task complexity for each patch, and divide the patches into complex tasks that require hovering for shooting and simple tasks that can be shot over. Nest reliability is used as a weighting factor in the weighted Voronoi diagram to divide the nest coverage area, the reachability of boundary patches to adjacent nests is calculated, and a candidate nest mapping table for boundary patches is established. When a fault signal is detected in a cell, the set of patches to be taken over within the coverage area of ​​the faulty cell is obtained, and the urgency of takeover is calculated based on the time limit for providing evidence for each patch to be taken over. Obtain the current availability status of drones in adjacent available nests in the candidate nest mapping table, and calculate the reach time and return margin to each patch to be taken over for each candidate drone type; Based on the urgency of takeover, the complexity of the map patch task, the reachability time, and the performance parameters of the candidate aircraft models, an aircraft model-takeover task adaptability matrix is ​​generated. The Hungarian algorithm with time window constraints is used to solve the model-takeover task fit matrix and output the optimal model-task allocation result. Based on the optimal allocation result, candidate aircraft type-differentiated takeover route segments are generated, and after time-series arrangement and conflict detection, a fault takeover path scheme is output.

2. The method for natural resource monitoring, evidence collection, and scheduling based on tower-mounted UAVs and their nests as described in claim 1, characterized in that, The survival analysis model is a Cox proportional hazards model. The inputs of the Cox proportional hazards model include: equipment service duration as a time variable, frequency of abnormal events in historical operation logs and number of fault repairs in recent maintenance records as covariates, and thunderstorm warning level and temperature and humidity anomaly degree in current environmental monitoring data as time-varying covariates. The output of the Cox proportional hazards model is the cumulative risk probability of each nest failing within the planned time window.

3. The natural resource monitoring, evidence collection, and scheduling method based on tower-mounted UAVs and their nests as described in claim 2, characterized in that... The nest failure probability value is calculated using the baseline survival function, the planning time window length, the values ​​of each covariate, and the regression coefficients obtained by fitting historical data. The baseline survival function refers to the probability that the nest will not fail within the planning time window under the baseline condition that all covariates are zero.

4. The method for monitoring, evidence collection, and scheduling natural resources based on tower-mounted UAVs and their nests as described in claim 1, characterized in that, The geometric features of the map patch include the patch area, the patch shape index, and the tortuosity of the patch boundary; the surrounding terrain data includes the terrain relief, vegetation occlusion rate, and building distribution density; the patch shape index is the ratio of the perimeter of the patch to the perimeter of a circle of the same area; the terrain relief is the difference between the maximum and minimum elevation values ​​within the area where the patch is located.

5. The natural resource monitoring, evidence collection, and scheduling method based on tower-mounted UAVs and their nests as described in claim 4, characterized in that, The criteria for classifying the complexity of the map patch task are as follows: when the shape index of the map patch exceeds the shape threshold, or the terrain undulation exceeds the undulation threshold, or the vegetation occlusion rate exceeds the occlusion threshold, the map patch is classified as a complex task; otherwise, it is classified as a simple task.

6. The method for natural resource monitoring, evidence collection, and scheduling based on tower-mounted UAVs and their nests as described in claim 1, characterized in that, The nest reliability is the complement of the nest failure probability value; the weight of the weighted Voronoi diagram is composed of nest reliability and UAV endurance; the boundary patch refers to the patch located in the boundary area of ​​the coverage area of ​​adjacent nests; the accessibility is determined by judging whether the round-trip distance from the boundary patch to the adjacent nest is less than the endurance of the UAV deployed in the adjacent nest.

7. The method for natural resource monitoring, evidence collection, and scheduling based on tower-mounted UAVs and their nests as described in claim 1, characterized in that, The urgency of takeover is calculated by the reciprocal of the difference between the evidence submission deadline of the plot to be taken over and the current time; the reachability time is calculated by dividing the distance between the plot to be taken over and the nest by the maximum flight speed of the candidate aircraft; the return trip margin is calculated by subtracting the round-trip distance and the mileage consumed in mission execution from the range of the candidate aircraft.

8. The method for natural resource monitoring, evidence collection, and scheduling based on tower-mounted UAVs and their nests as described in claim 1, characterized in that, The elements of the aircraft type-takeover task adaptability matrix are calculated by weighted summation of the takeover urgency normalized value, the map task complexity matching degree, the reachability time normalized value, and the return margin normalized value. Among them, the map task complexity matching degree is one when the map to be taken over is a complex task and the candidate aircraft type has hovering capability, one when the map to be taken over is a simple task, and zero in other cases. The weight coefficient of the takeover urgency dimension is set to be greater than the weight coefficients of other dimensions.

9. The method for natural resource monitoring, evidence collection, and scheduling based on tower-mounted UAVs and their nests as described in claim 1, characterized in that, The Hungarian algorithm with time window constraints includes the following steps: The patches to be taken over are sorted in descending order of urgency, and all candidate models and patches to be taken over are initialized to an unmatched state. Perform row and column reduction operations on the model-takeover task compatibility matrix; Starting with the map patch with the highest urgency of takeover, candidate models with zero fit and meeting the time window constraints are searched for each map patch to be taken over in turn; If a candidate model that meets the conditions is found, a matching relationship is established; if not, the search is repeated after expanding the set of zero elements by adjusting the dual variable. When all the patches to be taken over have been matched or there are no feasible candidate models for the remaining patches to be taken over, output the current matching result; The time window constraint is the time when the candidate aircraft arrives at the patch to be taken over plus the task execution time not exceeding the evidence submission deadline of the patch to be taken over.

10. A natural resource monitoring and evidence collection scheduling system based on tower-mounted UAVs and their nests, used to execute the natural resource monitoring and evidence collection scheduling method based on tower-mounted UAVs and their nests as described in any one of claims 1 to 9, characterized in that, include: The nest failure probability calculation module is used to obtain the historical operation logs, equipment service duration, recent maintenance records and current environmental monitoring data of each nest, and use the survival analysis model to calculate the nest failure probability value of each nest within the planned time window; The model performance extraction module is used to obtain a list of heterogeneous drone models deployed in each drone nest, extract the range, maximum flight speed, minimum turning radius and hovering ability of each model, and establish a model performance parameter table; The image patch classification module is used to obtain the geometric features of the image patches to be proven and the surrounding terrain data, calculate the image patch task complexity for each image patch, and classify the image patches into complex tasks that require hovering for shooting and simple tasks that can be shot over. The coverage area partitioning module is used to partition the nest coverage area by using nest reliability as a weighting factor in the weighted Voronoi diagram, calculate the reachability of boundary patches to adjacent nests, and establish a candidate nest mapping table for boundary patches. The takeover urgency calculation module is used to obtain the set of takeover patches within the coverage area of ​​the faulty nest when a nest fault signal is detected, and to calculate the takeover urgency based on the evidence collection time limit for each takeover patch. The candidate drone evaluation module is used to obtain the current availability status of drones in adjacent available drone nests in the candidate drone nest mapping table, and calculate the reachability time and return margin to each plot to be taken over for each candidate drone model. The adaptability matrix generation module is used to generate an aircraft type-takeover task adaptability matrix based on takeover urgency, map task complexity, reachability time, and candidate aircraft performance parameters. The task allocation module is used to solve the machine type-takeover task fit matrix using the Hungarian algorithm with time window constraints, and output the optimal allocation result of machine type-task. The path generation module is used to generate candidate aircraft type differentiated takeover route segments based on the optimal allocation results, and outputs fault takeover path schemes after time sequence arrangement and conflict detection.

Images