Digital supervision system and method based on unmanned aerial vehicle intelligent inspection and AI intelligent analysis

Abstract

The application is suitable for the field of intelligent monitoring technology, in particular to a digital monitoring system and method based on unmanned aerial vehicle (UAV) intelligent inspection and AI intelligent analysis, which comprises: acquiring a parameter set of a UAV inspection group; calculating redundant inspection resources of each UAV after entering a target area one by one based on the parameter set, which can quantitatively reflect the invalid input in the inspection process; determining the inspection sequence and task partition of the UAV based on the redundant inspection resources, and inversely driving the sequence and partition allocation through the degree of resource waste; inputting the sequence and partition data into a pre-trained path allocation machine learning model to generate accurate path allocation data with the help of the adaptive capacity of the model; and issuing the path allocation data and the order of entering the corresponding task inspection partition in the supervision target area to the UAVs in the UAV inspection group, thereby solving the problems of low resource utilization and incomplete supervision data caused by redundant collection, missing collection and missing inspection in fixed path inspection.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of intelligent monitoring, and particularly relates to a digital and intelligent supervision system and method based on unmanned aerial vehicle (UAV) intelligent inspection and AI intelligent analysis. BACKGROUND

[0002] The UAV inspection has become a key technical means in the fields of engineering construction supervision, power line operation and maintenance, oil and gas pipeline monitoring, and city infrastructure inspection, due to its high flexibility, wide coverage, high operation efficiency, and low safety risk. In the engineering supervision scene, the UAV can quickly obtain core supervision data such as image data, size information, and progress status of the target area, effectively making up for the shortcomings of traditional manual inspection, such as low efficiency, many blind spots, and data lag, and providing important data support for engineering quality control, safety hazard investigation, and progress management. As a development trend of the industry, the digital and intelligent supervision requires the deep integration of UAV inspection and intelligent analysis to realize the automation, dataization, and precision of the supervision process.

[0003] Currently, the application of UAV inspection in the supervision field mainly relies on the traditional operation mode. The technical personnel usually plan a fixed inspection path based on field experience, complete the data collection of the specified area by manually controlling the UAV, and then process the inspection data by manual comparison and simple software analysis to determine the key indicators in the engineering supervision, such as structural size deviation, construction progress matching degree, and safety hazard position. These methods can replace part of the manual inspection work to a certain extent, for example, reducing the amount of manual work in high-risk areas such as high-altitude structures and deep foundation pits, and improving the intuitiveness of the inspection data.

[0004] However, the existing UAV inspection application mode has obvious limitations. The traditional UAV inspection always relies on fixed planning and manual control, ignores the demand for multi-UAV collaborative inspection, and needs manual coordination of the operation sequence and area division of multiple UAVs, which not only consumes a large amount of labor cost, but also easily leads to low inspection efficiency, repeated collection of inspection data, and area missing problems due to unscientific coordination logic, thereby wasting equipment resources and affecting the completeness and accuracy of the supervision data. SUMMARY

[0005] The embodiments of the application provide a digital and intelligent supervision system and method based on UAV intelligent inspection and AI intelligent analysis, which can solve the problems in the background technology.

[0006] In a first aspect, the embodiments of the application provide a digital and intelligent supervision method based on UAV intelligent inspection and AI intelligent analysis, including:

[0007] obtaining a parameter set of a UAV inspection group;

[0008] Based on the parameter set of the unmanned aerial vehicle inspection group, a plurality of redundant inspection resources obtained after each unmanned aerial vehicle in the unmanned aerial vehicle inspection group enters the supervision target area one by one are calculated; wherein the redundant inspection resources are used to reflect the invalid inspection investment generated in the unmanned aerial vehicle inspection process, and the numerical size of the redundant inspection resources reflects the resource waste degree of the inspection process;

[0009] Based on a plurality of the redundant inspection resources, the order of each unmanned aerial vehicle in the unmanned aerial vehicle inspection group being included in the inspection sequence and the corresponding task inspection partition of each unmanned aerial vehicle are determined;

[0010] The order and the corresponding task inspection partition of each unmanned aerial vehicle are input into a preset path allocation model to obtain path allocation data of each unmanned aerial vehicle; wherein the path allocation model is a machine learning model obtained by pre-training;

[0011] The path allocation data and the order of entering the corresponding task inspection partition in the supervision target area in turn are issued to each unmanned aerial vehicle in the unmanned aerial vehicle inspection group.

[0012] The technical solution described above in the embodiments of the present application has at least the following technical effects:

[0013] The digital supervision method based on unmanned aerial vehicle intelligent inspection and AI intelligent analysis provided by the embodiments of the present application obtains the parameter set of the unmanned aerial vehicle inspection group, provides accurate data basis for subsequent inspection resource optimization and path planning, and avoids the problems of parameter ambiguity and poor adaptability in traditional manual experience planning from the source. Based on the parameter set, the redundant inspection resources of each unmanned aerial vehicle entering the target area one by one are calculated, which can quantitatively reflect the invalid input in the inspection process, accurately identify the resource waste points, and effectively solve the problems of low resource utilization and incomplete supervision data caused by redundant collection, missing collection and missing inspection in traditional fixed path inspection. Based on the redundant inspection resources, the inspection sequence and task partition of the unmanned aerial vehicle are determined, and the sequence and partition allocation are driven in reverse by the degree of resource waste, breaking the limitations of sequence confusion and partition overlap in traditional manual coordination of multi-machine operation, realizing the ordering and rationalization of multi-machine inspection, greatly reducing the human coordination cost of multi-machine cooperation, and improving the overall efficiency of inspection. The sequence and partition data are input into the pre-trained path allocation machine learning model, and the accurate path allocation data are generated by means of the adaptive ability of the model, avoiding the defects of rigid path and poor adaptability of traditional fixed path in complex supervision scene. The path allocation data and the order of entering the corresponding task inspection partition in the supervision target area are issued to each unmanned aerial vehicle in the unmanned aerial vehicle inspection group, which not only optimizes the inspection efficiency of single unmanned aerial vehicle, but also optimizes the operation system of the whole inspection group based on the dynamic adaptive ability of multi-machine cooperation logic and AI model, so as to improve the integrity and accuracy of engineering supervision data, ensure the safety and efficiency of the inspection process, reduce the dependence on manual experience in inspection, and comprehensively and systematically promote the transformation of engineering supervision from manual dominance to digitalization.

[0014] In a second aspect, the embodiments of the present application provide a digital supervision system based on unmanned aerial vehicle intelligent inspection and AI intelligent analysis, comprising:

[0015] The acquisition unit is configured to acquire a parameter set of an unmanned aerial vehicle inspection group.

[0016] The detection unit is configured to calculate a plurality of redundant inspection resources obtained after each unmanned aerial vehicle in the unmanned aerial vehicle inspection group enters a supervision target area one by one based on the parameter set of the unmanned aerial vehicle inspection group, wherein the redundant inspection resources are used to reflect invalid inspection input generated in the unmanned aerial vehicle inspection process, and the numerical value of the redundant inspection resources reflects the degree of resource waste in the inspection process.

[0017] The task unit is configured to determine the order of each unmanned aerial vehicle in the unmanned aerial vehicle inspection group being included in the inspection sequence and the corresponding task inspection partition of each unmanned aerial vehicle based on the plurality of redundant inspection resources.

[0018] a path unit configured to input the sequence and the task inspection subarea corresponding to each unmanned aerial vehicle into a preset path allocation model to obtain path allocation data of each unmanned aerial vehicle, wherein the path allocation model is a machine learning model trained in advance;

[0019] a distribution unit configured to distribute the path allocation data and the sequence of entering the corresponding task inspection subarea in the supervision target area to each unmanned aerial vehicle in the group of unmanned aerial vehicles.

[0020] In a third aspect, an embodiment of the present application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor, and the processor implements the method according to any one of the above aspects when executing the computer program.

[0021] In a fourth aspect, an embodiment of the present application provides a computer program product, which, when running on an electronic device, causes the electronic device to perform the method according to any one of the above aspects.

[0022] It can be understood that the beneficial effects of the above-mentioned second aspect to fourth aspect can be referred to the related description in the above aspects, which will not be repeated here. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0024] Figure 1 is a flowchart of a smart supervision method based on unmanned aerial vehicle smart inspection and AI intelligent analysis provided by an embodiment of the present application;

[0025] Figure 2 is a deployment schematic diagram of a smart supervision method based on unmanned aerial vehicle smart inspection and AI intelligent analysis provided by an embodiment of the present application;

[0026] Figure 3 is an inspection subarea and sequence schematic diagram of a smart supervision method based on unmanned aerial vehicle smart inspection and AI intelligent analysis provided by an embodiment of the present application;

[0027] Figure 4 is a structural schematic diagram of a smart supervision system based on unmanned aerial vehicle smart inspection and AI intelligent analysis provided by an embodiment of the present application;

[0028] Figure 5This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0029] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0030] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0031] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0032] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determination" or "if the described condition or event is detected" may be interpreted, depending on the context, as "once determination," "in response to determination," "once the described condition or event is detected," or "in response to the detection of the described condition or event."

[0033] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0034] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0035] Unmanned aerial vehicle (UAV) inspection has become a key technology in the fields of engineering construction supervision, power line operation and maintenance, oil and gas pipeline monitoring, and urban infrastructure inspection, due to its high flexibility, wide coverage, high efficiency, and low safety risks. In the context of engineering supervision, UAVs can quickly obtain core supervision data such as image data, size information, and progress status of the target area, effectively addressing the shortcomings of traditional manual inspection such as low efficiency, many blind spots, and data lag. UAVs provide important data support for engineering quality control, safety hazard investigation, and progress management. As a trend in the industry, digital supervision requires the integration of UAV inspection and intelligent analysis to achieve automation, dataization, and precision in the supervision process.

[0036] Currently, the application of UAV inspection in the supervision field mainly relies on traditional operation modes. Technicians usually plan fixed inspection paths based on field experience, manually control UAVs to collect data in designated areas, and then process inspection data through manual comparison and simple software analysis to determine key indicators in engineering supervision, such as structural size deviation, construction progress matching degree, and safety hazard location. These methods can replace some manual inspection work to a certain extent, such as reducing the amount of manual work in high-risk areas (e.g., high-altitude structures and deep foundation pits) and improving the intuitiveness of inspection data.

[0037] However, the existing application of UAV inspection has obvious limitations. Traditional UAV inspection relies on fixed planning and manual control, ignoring the need for multi-UAV collaborative inspection. Manual coordination is required to determine the operation sequence and regional division of multiple UAVs, which not only consumes a large amount of labor costs but also leads to low inspection efficiency, repeated data collection, and regional omission due to unscientific coordination logic. This not only wastes equipment resources but also affects the completeness and accuracy of supervision data.

[0038] To solve the above problems, the embodiment of the present application provides a digital supervision system and method based on unmanned aerial vehicle intelligent inspection and AI intelligent analysis. In the method, the parameter set of the unmanned aerial vehicle inspection group is obtained to provide accurate data basis for subsequent inspection resource optimization and path planning, avoiding the problems of parameter ambiguity and poor adaptability in traditional manual experience planning. Based on the parameter set, the redundant inspection resources of each unmanned aerial vehicle entering the target area one by one are calculated, which can quantitatively reflect the invalid input in the inspection process, accurately identify the resource waste points, and effectively solve the problems of low resource utilization and incomplete supervision data caused by redundant collection, missing collection and missing inspection in traditional fixed path inspection. Based on the redundant inspection resources, the inspection sequence and task partition of the unmanned aerial vehicle are determined, and the sequence and partition allocation are driven in reverse by the degree of resource waste, breaking the limitations of sequence confusion and partition overlap in traditional manual coordination of multi-machine operation, realizing the order and rationalization of multi-machine inspection, greatly reducing the human coordination cost of multi-machine cooperation, and improving the overall efficiency of inspection. The sequence and partition data are input into the pre-trained path allocation machine learning model, and the accurate path allocation data are generated by the adaptive ability of the model, avoiding the defects of path rigidity and poor adaptability of traditional fixed path in complex supervision scenarios. The path allocation data and the order of entering the corresponding task inspection partition in the supervision target area are issued to each unmanned aerial vehicle in the unmanned aerial vehicle inspection group, which not only optimizes the inspection efficiency of a single unmanned aerial vehicle, but also optimizes the operation system of the entire inspection group based on the dynamic adaptive ability of multi-machine cooperation logic and AI model, thereby improving the completeness and accuracy of engineering supervision data, ensuring the safety and efficiency of the inspection process, reducing the dependence on manual experience, and comprehensively and systematically promoting the transformation of engineering supervision from manual dominance to digitalization.

[0039] The digital supervision method based on unmanned aerial vehicle intelligent inspection and AI intelligent analysis provided by the embodiment of the present application can be applied to an electronic device, and the electronic device is the execution subject of the digital supervision method based on unmanned aerial vehicle intelligent inspection and AI intelligent analysis provided by the embodiment of the present application. The specific type of the electronic device is not limited in the embodiment of the present application.

[0040] For example, the electronic device can be an ultra-mobile personal computer (UMPC), a netbook, a desktop computer, a computer, a laptop computer, a communication device, a computing device, a satellite wireless device, etc.

[0041] In order to better understand the digital supervision method based on unmanned aerial vehicle intelligent inspection and AI intelligent analysis provided by the embodiment of the present application, the specific implementation process of the digital supervision method based on unmanned aerial vehicle intelligent inspection and AI intelligent analysis provided by the embodiment of the present application is exemplarily introduced as follows.

[0042] Figure 1 A schematic flowchart of a digital supervision method based on unmanned aerial vehicle (UAV) intelligent inspection and AI intelligent analysis is shown, Figure 2 A running flowchart of a digital supervision method based on UAV intelligent inspection and AI intelligent analysis is shown. The digital supervision method based on UAV intelligent inspection and AI intelligent analysis includes:

[0043] S100, a parameter set of a UAV inspection group is acquired.

[0044] It can be understood that the UAV inspection group refers to a plurality of UAVs organized to perform an inspection task in a supervision target area. The UAV inspection group is not only a simple combination of several UAVs, but also includes parameter information related to group operation, such as the number of UAVs in the group, the flight speed range, the endurance time, the task load capacity, and the number and role allocation of each UAV in the group. The parameter set refers to a set of multi-dimensional parameters related to the operation of the UAV inspection group. The data in the parameter set can come from the hardware configuration of the UAV itself, the real-time state of the flight control system, and the environmental conditions of the supervision target area. Please refer to Figure 2 The parameter set can include the span of the multiple inspection subareas after the supervision target area is divided, the path distance between adjacent UAVs, the total length of the supervision target area, and can also include UAV battery capacity, maximum range, camera resolution, etc.

[0045] The parameter set of the UAV inspection group can be collected and integrated through a data interface or a sensor system. The API interface provided by the UAV management platform can be called to read the real-time state parameters of the UAV. At the same time, the map data of the supervision target area can be loaded from the task management database, and the inspection subarea span and the total length of the supervision target area can be numerically expressed. All the acquired parameters are stored in the memory according to the preset structure, such as parameter dictionary or multi-dimensional array, so as to be quickly called in the subsequent calculation process.

[0046] The reason for such design is that if the subsequent directly enters the inspection resource calculation without the parameter set as input, the spatial layout of the inspection task and the performance characteristics of the UAV cannot be accurately described, and the redundant inspection resource calculation cannot be effectively modeled. The completeness of the parameter set directly determines the accuracy of the redundant inspection resource calculation. The effect is that by acquiring the parameter set of the UAV inspection group, standardized input conditions can be provided for subsequent steps, so that the resource investment evaluation and redundant resource calculation process have a unified basis. This not only facilitates the subsequent path optimization and inspection subarea allocation, but also reduces the calculation deviation caused by missing or inconsistent parameters, thereby improving the efficiency and rationality of the entire UAV group inspection scheduling.

[0047] S200, based on the parameter set of the UAV inspection group, calculating a plurality of redundant inspection resources obtained after each UAV in the UAV inspection group enters the supervision target area one by one; wherein the redundant inspection resource is used to reflect the invalid inspection input generated in the UAV inspection process, and the value of the redundant inspection resource reflects the resource waste degree of the inspection process.

[0048] It can be understood that the redundant inspection resource refers to a quantitative index of invalid inspection input caused by repeated coverage, path overlap or uneven partition allocation in the process of the UAV group performing the inspection task of the supervision target area. The value of the redundant inspection resource reflects the resource waste degree, and the unit can be area, path length or corresponding calculation cost.

[0049] For all UAVs in the UAV group, the resource input situation of each UAV entering the task area can be simulated in turn combined with the description of the partition span, path spacing and total length of the target area in the parameter set, and the redundant part caused by spatial coverage waste and path repetition is calculated, so as to obtain the corresponding redundant inspection resource.

[0050] The reason for this is that if the redundant inspection resource is not quantified in group inspection, resource waste is easy to occur in task allocation, resulting in a decrease in overall inspection efficiency. By calculating the redundant inspection resource of each UAV one by one, the resource utilization rate can be intuitively reflected, and data support can be provided for subsequent task partition and inspection sequence optimization.

[0051] In one possible implementation, the parameter set includes the partition span of each inspection partition in the supervision target area, the path spacing between adjacent UAVs, and the total length of the supervision target area; S200, based on the parameter set of the UAV inspection group, calculating a plurality of redundant inspection resources obtained after each UAV in the UAV inspection group enters the supervision target area one by one, including:

[0052] S210, based on the partition span of each inspection partition in the supervision target area, the path spacing between adjacent UAVs, and the total length of the supervision target area, calculating the resource input evaluation value generated when each UAV in the UAV inspection group is assigned to any inspection partition as the first UAV.

[0053] It can be understood that the resource input evaluation value refers to the resource consumption measure calculated according to the span of the partition, the path distance between adjacent unmanned aerial vehicles and the total length of the supervision target area when the unmanned aerial vehicle is assigned to any inspection sub-area in the supervision target area. The resource input evaluation value reflects the spatial utilization efficiency and path planning cost of the unmanned aerial vehicle when performing the inspection task. The larger the value, the higher the resource waste. Calculating the resource input evaluation value means simulating the allocation of the unmanned aerial vehicle in different partitions using the parameter set of the supervision target area, and generating a quantitative index. For each unmanned aerial vehicle, each inspection sub-area is traversed, the spatial waste and path cost of the unmanned aerial vehicle when assigned to the sub-area are calculated, and then the two are combined to obtain the evaluation value.

[0054] The span data of the partition of the supervision target area, the path distance between the unmanned aerial vehicles and the total length of the area can be loaded. Then, the scenario of the unmanned aerial vehicle as the first unmanned aerial vehicle entering the inspection sub-area is set. In each partition, the spatial waste degree is calculated based on the difference between the coverage span of the unmanned aerial vehicle and the span of the partition. Then, the path cost is calculated according to the ratio of the span of the partition to the total length of the area, combined with the path planning basic cost of the unmanned aerial vehicle. Finally, the spatial waste coefficient and the path planning cost are combined according to the weight to obtain the resource input evaluation value of the unmanned aerial vehicle in the partition. By traversing all partitions, multiple resource input evaluation values of the unmanned aerial vehicle in different partitions can be generated.

[0055] The reason for this is that the allocation strategy of the first unmanned aerial vehicle has a decisive influence on the task resource consumption of the entire inspection group. If the resource input of the first unmanned aerial vehicle is too large, it will directly lead to an increase in the redundancy of the subsequent unmanned aerial vehicle inspection resources. By calculating and quantifying the resource input of different partitions at this stage, a basis can be provided for subsequent analysis of redundant inspection resources and optimization of unmanned aerial vehicle scheduling. The effect is that the allocation scheme of the first unmanned aerial vehicle can be quantitatively evaluated before the inspection task has started, avoiding premature resource waste, thereby improving the execution efficiency and sub-area coverage balance of the overall inspection task.

[0056] Optionally, in S210, based on the span of each inspection sub-area in the supervision target area, the path distance between adjacent unmanned aerial vehicles, and the total length of the supervision target area, a resource input evaluation value generated when each unmanned aerial vehicle in the unmanned aerial vehicle inspection group is assigned as the first unmanned aerial vehicle to any inspection sub-area is calculated, including:

[0057] In S211, based on the span of each inspection sub-area in the supervision target area and the total length of the supervision target area, for each unmanned aerial vehicle in the unmanned aerial vehicle inspection group, a first resource input evaluation value generated when the unmanned aerial vehicle is assigned as the first unmanned aerial vehicle to any inspection sub-area is calculated.

[0058] It can be understood that the first resource investment evaluation value refers to the resource consumption value calculated according to parameters such as the span of the partition and the total length of the region when each unmanned aerial vehicle is assigned to any inspection partition as the first unmanned aerial vehicle in the supervision target region. This evaluation value is used to measure the degree of resource investment caused by the unmanned aerial vehicle as the first unmanned aerial vehicle entering a certain partition, and reflects the comprehensive performance of space utilization efficiency and path planning cost.

[0059] For each unmanned aerial vehicle, it can be assigned to any inspection partition as the first unmanned aerial vehicle, a coverage and path planning model is established in the partition, and then the resource consumption generated by the unmanned aerial vehicle is calculated based on parameters such as the span of the partition and the total length.

[0060] The span data of each inspection partition in the supervision target region and the total length of the supervision target region can be read. Then it is determined that a certain unmanned aerial vehicle is the first unmanned aerial vehicle entering a certain inspection partition. In the inspection direction, the coverage span of the unmanned aerial vehicle is compared with the span of the partition to obtain the space coverage efficiency. Then, combined with the proportion of the span of the partition to the total length, the path planning overhead required to enter the partition is calculated. The space coverage efficiency loss and the path planning overhead are weighted and synthesized to obtain the first resource investment evaluation value of the unmanned aerial vehicle in the partition. Finally, the above process is repeated to traverse all inspection partitions to generate multiple first resource investment evaluation values of the unmanned aerial vehicle in different partitions. The reason for this is that the task allocation of the first unmanned aerial vehicle often determines the subsequent scheduling of the entire inspection group. By quantitatively calculating the resource investment of each unmanned aerial vehicle in different partitions, the influence of the first unmanned aerial vehicle on the overall resource utilization under different task selections can be clearly reflected, thereby improving the inspection efficiency of the entire unmanned aerial vehicle group and reducing invalid investment.

[0061] Exemplarily, in S211, based on the span of each inspection partition in the supervision target region and the total length of the supervision target region, the first resource investment evaluation value generated when each unmanned aerial vehicle in the unmanned aerial vehicle inspection group is assigned to any inspection partition as the first unmanned aerial vehicle is calculated, including:

[0062] In S2111, for each unmanned aerial vehicle in the unmanned aerial vehicle inspection group, the coverage span of the unmanned aerial vehicle along the inspection direction is obtained.

[0063] It can be understood that the self-coverage span of the UAV along the inspection direction refers to the length of the space range in which the UAV can achieve effective data collection (such as clear image recognition and accurate sensor detection) along the preset inspection path direction (such as the horizontal flight direction or the vertical elevation direction) when performing the inspection task. The value is calculated by the inspection load performance (such as the field of view angle determined by the camera focal length and the detection radius of the laser radar) carried by the UAV and the actual flight height. For example, a UAV carrying a 24mm focal length camera has an effective coverage length of 45 meters along the inspection direction at a flight height of 80 meters, which is converted by a trigonometric function. The 45 meters is the self-coverage span. The flight control of the UAV can read the load model and flight height data, call the preset load performance database (storing the field of view / detection range parameters of different models of UAVs at different heights), and output the value of the self-coverage span along the inspection direction. The reason for doing so is that when calculating the spatial waste in the inspection partition, the actual coverage range of the UAV must be used as the benchmark. If this parameter is missing, it will not be possible to determine the matching relationship between the partition span and the coverage ability of the UAV, and thus it will not be possible to quantify the invalid consumption of spatial resources. This can provide accurate basic data for the subsequent calculation of the first spatial resource waste coefficient, ensure that the spatial waste evaluation does not deviate from the actual performance of the UAV, and avoid resource evaluation deviation caused by parameter estimation.

[0064] S2112, for any inspection partition, according to the preset space utilization rate evaluation rule, the first spatial resource waste coefficient between the self-coverage span of the UAV along the inspection direction and the partition span of the inspection partition when each UAV is assigned to the inspection partition as the first UAV.

[0065] It can be understood that the first space resource waste coefficient is a dimensionless index specially used to quantify the invalid consumption of space resources caused by the mismatch between the coverage span of the first unmanned aerial vehicle and the span of the inspection partition when the first unmanned aerial vehicle is allocated to the inspection partition, and the value range thereof is usually 0-1 (0 represents no waste, and 1 represents complete waste). For example, when the coverage span of the unmanned aerial vehicle is completely matched with the span of the inspection partition, the coefficient is 0, and when the coverage span of the unmanned aerial vehicle is only 50% of the span of the inspection partition, the coefficient can be 0.5. The absolute value |L-S| of the difference between the coverage span S of the unmanned aerial vehicle and the target inspection partition span L can be calculated, that is, the initial idle space value or the overlapping space value, and then a preset space utilization evaluation rule function (such as a linear function f(x)=|L-S| / L or a segmented function, when |L-S|≤5 meters, f(x)=0.1×|L-S| / L, and when |L-S|>5 meters, f(x)=0.5×|L-S| / L) is called to calculate the first space resource waste coefficient by substituting the difference into the function. The reason for this is that the essence of space waste is the mismatch between coverage capacity and partition demand, but the length difference cannot be directly used to fuse and calculate the path planning cost (non-length unit), and the coefficient needs to be converted into a unified dimension quantitative index to realize the additivity of space waste and path cost, provide the possibility for subsequent comprehensive calculation of resource investment evaluation value, and make the space waste degree of different partitions and different unmanned aerial vehicles comparable, which is convenient for screening the optimal allocation scheme

[0066] Exemplarily, S2112, for any inspection partition, according to the preset space utilization evaluation rule, the first space resource waste coefficient between the coverage span of the unmanned aerial vehicle along the inspection direction and the span of the inspection partition when each unmanned aerial vehicle is allocated to the inspection partition as the first unmanned aerial vehicle is calculated, including:

[0067] S21121, for any inspection partition, the difference between the span of the inspection partition and the coverage span of the unmanned aerial vehicle when each unmanned aerial vehicle is allocated to the inspection partition as the first unmanned aerial vehicle is calculated, to obtain the initial idle space value in the inspection partition.

[0068] It can be understood that the initial idle space value refers to the length of space not effectively covered by the first UAV in a certain inspection partition (when the UAV coverage span is less than the partition span), or the length of space beyond the partition boundary (when the UAV coverage span is greater than the partition span) after the first UAV is assigned to the inspection partition. The former is a positive value, and the latter is a negative value (the negative sign represents overlap waste). For example, when the partition span is 100 meters and the UAV coverage span is 70 meters, the initial idle space value is 30 meters. If the UAV coverage span is 120 meters, the initial idle space value is -20 meters. The partition span L of the target inspection partition and the self-coverage span S of the current UAV can be used to calculate the initial idle space value according to the formula: initial idle space value = L-S. If the result is positive, it is marked as idle waste, and if the result is negative, it is marked as overlap waste after taking the absolute value. The calculation result and the mark information are stored in the temporary database. The reason for this is that the initial idle space value is the original data of space waste, which directly reflects the deviation between the UAV coverage capability and the partition space demand. If the waste coefficient is directly calculated, the coefficient will lack actual data support and become an abstract index. The effect is to provide specific numerical basis for the calculation of the first space resource waste coefficient, making the physical meaning of the coefficient more explicit. At the same time, through the idle / overlap mark, it can assist the subsequent path optimization (such as adjusting the UAV flight boundary for overlap waste).

[0069] S21122, according to the preset space utilization evaluation rule, the initial idle space value is converted into the first space resource waste coefficient; wherein the first space resource waste coefficient is positively correlated with the initial idle space value.

[0070] It can be understood that the preset space utilization evaluation rule is prepared in advance according to the accuracy requirement of the inspection task and the resource constraint condition, which is a mathematical mapping rule for converting the initial idle space value (or overlap space value) into the first space resource waste coefficient. Common rules include linear mapping rule (coefficient proportional to space value), step mapping rule (space value in different intervals corresponds to fixed coefficient), exponential mapping rule (coefficient rapidly increases when space value exceeds threshold), etc.

[0071] For example, a certain power inspection task requires strict control of idle space. The preset rule can be that when the space value is less than or equal to 10 meters, the coefficient = 0.2 x space value / partition span; when the space value is greater than 10 meters, the coefficient = 0.6 x space value / partition span. The configuration file of the preset rule (determining the mapping function type and parameters) can be read to extract the initial idle space value V and the corresponding partition span L from the temporary database. If V is positive (idle waste), directly substitute the mapping function; if V is negative (overlap waste), take the absolute value |V| and substitute the mapping function to calculate the first space resource waste coefficient. At the same time, record the rule parameters (such as interval threshold, coefficient weight) in the conversion process for tracing back. The reason for this is that different inspection scenarios have different tolerances for space waste (such as low tolerance for overlap waste in pipeline inspection and slightly higher tolerance for idle waste in grassland inspection). A unified conversion rule cannot meet the diversified needs, and the preset rule can achieve scene adaptation. The effect is that the first space resource waste coefficient is more in line with the actual task requirements, avoiding overestimation or underestimation caused by general rules. At the same time, the setting of the positive correlation between the first space resource waste coefficient and the initial idle space value ensures that the more serious the space waste, the larger the coefficient, which intuitively reflects the degree of resource waste.

[0072] S2113, calculate the path planning cost required for each unmanned aerial vehicle to enter the inspection partition when the unmanned aerial vehicle is assigned to the inspection partition as the first unmanned aerial vehicle; wherein the path planning cost is determined based on the total length of the supervision target area and the partition span of the inspection partition.

[0073] It can be understood that the path planning cost required for the UAV to enter the inspection subzone refers to the resource consumption value consumed when the UAV flies from the task take-off point (or the previous task node) to the target inspection subzone and plans a path in the subzone that meets the inspection accuracy requirements. The unit can be set as energy consumption unit (such as kilowatt-hour), time unit (such as minute) or calculation resource unit (such as floating point operation times) according to the demand. For example, the flight energy consumption of a UAV from the take-off point to the subzone is 0.3 kilowatt-hour, and the calculation consumption of the subzone path planning is 0.1 calculation unit, and the total path planning cost is 0.4 (after weighting). The total length T of the target area and the target subzone span L can be used to extract the path planning basic cost value C (i.e. the path resource consumption in unit space) of the UAV from the UAV performance database, calculate the subzone space ratio R = L / T, and then according to the preset cost calculation model (such as cost = C / R, the smaller R is, the more dispersed the subzone is, the more complex the path planning is, and the higher the cost is), the preliminary cost is calculated by substituting C and R. If the UAV needs to cross multiple subzones to reach the target subzone, the cross-region flight cost (calculated according to the subzone distance) needs to be additionally superimposed to obtain the total path planning cost. The reason for this is that path planning is the core link of the UAV to perform the inspection task, and its resource consumption (such as energy consumption, time) directly affects the overall inspection efficiency. If only the space waste is evaluated and the path cost is ignored, the resource investment evaluation value will be one-sided (such as a subzone with small space waste but high path planning cost, and the overall resource investment is actually larger); its effect is to realize the resource evaluation in two dimensions, so that the resource investment evaluation value is more comprehensive, and the introduction of the subzone space ratio ensures that the path cost is related to the location characteristics of the subzone in the overall area, avoiding the cost estimation that is out of the actual scene.

[0074] S2113, the path planning cost required for the UAV to enter the inspection subzone when each UAV is allocated to the inspection subzone as the first UAV includes:

[0075] S21131, obtain the path planning basic cost value of each UAV.

[0076] It can be understood that the path planning basic cost value of the unmanned aerial vehicle refers to the benchmark resource consumption value required by the unmanned aerial vehicle to complete unit length (usually 1 kilometer) path planning in a standard test environment (such as no shelter, flat terrain, fixed flight speed 15 m / s), which is determined by the hardware performance (such as power system efficiency, flight control chip computing power) of the unmanned aerial vehicle and the path planning algorithm (such as A algorithm, RRT algorithm). For example, a certain unmanned aerial vehicle using high-efficiency motor and optimized algorithm has a path planning basic cost value of 0.2 energy consumption units per kilometer, while an unmanned aerial vehicle using ordinary motor and basic algorithm has a value of 0.5 energy consumption units per kilometer. The theoretical basic cost value can be obtained through the technical manual provided by the unmanned aerial vehicle manufacturer, and then a plurality of (at least 10) path planning experiments are performed in the pre-standard test environment, the actual resource consumption (such as energy consumption, calculation time) of each experiment is recorded, the average value is calculated and compared with the theoretical value, if the deviation is ≤5%, the theoretical value is adopted; if the deviation is >5%, the experimental average value is taken as the final path planning basic cost value, which is stored in the unmanned aerial vehicle performance database and associated with the unmanned aerial vehicle number for calling. The reason for this is that the theoretical value provided by the manufacturer may deviate from the actual use environment, and if the theoretical value is directly used, it will lead to inaccurate path planning cost calculation, which will affect the reliability of the resource investment evaluation value. The effect is to ensure that the path planning basic cost value is consistent with the actual performance of the unmanned aerial vehicle, providing benchmark data for accurate calculation of subsequent path planning cost, and avoiding the chain evaluation error caused by the deviation of the basic data.

[0077] S21132, calculate the ratio of the span of the inspection sub-area to the total length of the supervision target area when each unmanned aerial vehicle is assigned to the inspection sub-area as the first unmanned aerial vehicle, to obtain the sub-area space proportion of the inspection sub-area.

[0078] It can be understood that the partition space ratio of the inspection partition refers to the ratio of the span of a certain inspection partition to the total length of the supervision target area. This ratio is a dimensionless value, and the value range is between 0 and 1. The larger the ratio, the higher the proportion of the partition in the overall area, and the more concentrated the space. The smaller the ratio, the more dispersed the partition. For example, the total length of the supervision target area is 2000 meters, and the span of a certain partition is 400 meters, the partition space ratio is 400 / 2000=0.2, and the span of another partition is 1000 meters, the ratio is 0.5. The reason for doing this is that the partition space ratio directly reflects the spatial concentration of the partition, and the spatial concentration is a key factor affecting the complexity of path planning (dispersed partitions require more turning and obstacle avoidance operations, and the path cost is higher). If the partition space ratio is missing, the path planning cost cannot be combined with the actual spatial characteristics of the partition, and a scenario adjustment factor is provided for subsequent path planning cost calculation, ensuring that the path cost reflects the difference in different concentration partitions for the same UAV, avoiding one-size-fits-all cost estimation, and improving evaluation accuracy.

[0079] S21133, based on the partition space ratio of the inspection partition and the path planning base cost value of each UAV, the path planning cost required for each UAV to enter the inspection partition is obtained.

[0080] It can be understood that the path planning cost is calculated based on the partition space ratio of the inspection partition and the path planning base cost value of the UAV. The core is to adjust the base cost through the partition space ratio to obtain the resource consumption value that fits the actual scenario of the partition. The path planning base cost value C of the current UAV can be extracted from the UAV performance database, and the space ratio R of the target partition can be extracted from the temporary database. Through the cost calculation formula path planning cost=C / R (when R is smaller, the denominator is smaller, and the cost is larger, which conforms to the actual situation that the path planning of dispersed partitions is more complex), if the task requires additional consideration of the distance between the partition and the take-off and landing point, a distance correction term can be added to the formula (such as cost=C / R+k×D, D is the distance between the center of the partition and the take-off and landing point, and k is the distance weight coefficient). After calculation, the result is stored in association with the UAV number and the partition number. The reason for doing this is that the path planning base cost value is the unit cost in a standard environment, while the spatial concentration of the actual partition is different, and the complexity of path planning is also different. If the base cost value is used directly, the cost of dispersed partitions will be underestimated (the actual consumption is much higher than the base value), and the cost of concentrated partitions will be overestimated. The calculation of the path planning cost realizes the scenario-based correction of the base cost value, which is more consistent with the resource consumption of the UAV when actually performing the task. The introduction of the correction term can further improve the accuracy of the cost calculation, and provide a guarantee for the comprehensiveness of the subsequent resource investment evaluation value.

[0081] S2114, the first space resource waste coefficient is weighted and summed with the path planning cost to obtain a first resource input evaluation value generated when each UAV is assigned to the inspection partition as the first UAV.

[0082] It can be understood that the resource input of the two dimensions of space waste and path consumption can be integrated to obtain a single dimension of the first resource input evaluation value, the unit of the first resource input evaluation value is consistent with the path planning cost (such as energy consumption unit, time unit), and the larger the value represents the more serious the resource waste of the first UAV assigned to the partition, for example, the first space resource waste coefficient is 0.3 (dimensionless, which needs to be converted to a value with the same dimension as the cost first), the path planning cost is 0.5 energy consumption unit, and the weights are 0.4 and 0.6 respectively, then the first resource input evaluation value = 0.3 x 0.4 x K + 0.5 x 0.6 (K is the dimension conversion coefficient of the coefficient and the cost, and here it is assumed that K = 1) = 0.12 + 0.3 = 0.42 energy consumption unit. The space coefficient weight W1 and the path cost weight W2 (W1 + W2 = 1) can be obtained by reading the preset weight configuration file (the weight is set according to the task priority, such as space utilization priority, space coefficient weight 0.6, path cost weight 0.4; path efficiency priority, then the weight is opposite), the first space resource waste coefficient F and the path planning cost C are extracted from the temporary database, if F is a dimensionless value, multiplied by the dimension conversion coefficient K (K takes the value equal to the average path planning cost under the current task, to ensure that F x W1 x K and C x 2 are consistent in dimension), then according to the formula: the first resource input evaluation value = F x W1 x K + C x W2, after the calculation is completed, it is stored in the evaluation result database and associated with the UAV and partition information. The reason for this is that space waste and path cost are two core factors affecting the resource input of the first UAV, and separate evaluation of either factor cannot fully reflect the overall resource waste situation (such as a small space waste but a high path cost in a certain partition, the overall input is still large), and a comprehensive evaluation needs to be achieved by weighted summation; the effect is to obtain a quantitative index that can fully reflect the resource input of the first UAV, which can be directly used for comparison between different partitions and different UAVs, and provides a unified standard for subsequent selection of the optimal UAV-partition combination.

[0083] S2115, all inspection partitions are traversed to obtain a plurality of first resource input evaluation values corresponding to each UAV assigned to different inspection partitions as the first UAV.

[0084] It can be understood that for a single unmanned aerial vehicle, the resource input evaluation value of each patrol partition is calculated one by one to form a partition-evaluation value mapping set of the unmanned aerial vehicle, for example, after a certain unmanned aerial vehicle traverses three partitions, the mapping relationship of partition 1: 0.42 energy consumption units, partition 2: 0.35 energy consumption units, and partition 3: 0.5 energy consumption units is obtained. The partition list can be formed by the number and span information of all patrol partitions, the evaluation value set of the unmanned aerial vehicle is initialized to be empty, then a partition is selected from the partition list as a target partition, the first resource input evaluation value calculation logic is called to obtain the first resource input evaluation value corresponding to the partition, the partition number-evaluation value key-value pair is added to the evaluation value set, and the process is repeated until all partitions in the partition list are processed. Finally, the evaluation value set is sorted in ascending order of evaluation value (to facilitate subsequent quick selection of the optimal partition) and stored in the evaluation database dedicated to the unmanned aerial vehicle. The reason for this is that the span and spatial location of different patrol partitions are different, and the spatial waste and path cost of the same unmanned aerial vehicle in different partitions are different. If only the evaluation value of a single partition is calculated, a better partition allocation scheme may be missed. The effect of multiple first resource input evaluation values is to provide full- partition coverage resource evaluation data for each unmanned aerial vehicle. Through sorting, the optimal candidate partition (the partition with the smallest evaluation value) of the unmanned aerial vehicle can be quickly located, and at the same time, rich data support is provided for subsequent unmanned aerial vehicle-partition matching optimization at the group level, avoiding overall resource waste caused by local optimization.

[0085] In S212, based on the path distance between adjacent unmanned aerial vehicles, for each remaining unmanned aerial vehicle, a second resource input evaluation value is calculated when the unmanned aerial vehicle is assigned to any remaining patrol partition as the next sequential unmanned aerial vehicle, until all unmanned aerial vehicles are assigned; wherein the second resource input evaluation value is the resource input degree of any unmanned aerial vehicle except the first unmanned aerial vehicle assigned to any remaining patrol partition.

[0086] It can be understood that the second resource input evaluation value is a quantitative index of the resource input of each remaining unmanned aerial vehicle except the first unmanned aerial vehicle when it is assigned to any remaining patrol partition (i.e., all unmanned aerial vehicles are arranged in full) under the premise of considering the influence of the path distance of the assigned unmanned aerial vehicle. The core difference between this index and the first resource input evaluation value is that the influence factor of the path distance of adjacent unmanned aerial vehicles is introduced, for example, the path distance requirement of the assigned unmanned aerial vehicle in partition A is 20 meters, and when the current unmanned aerial vehicle is assigned to adjacent partition B, if the actual distance is only 15 meters, there will be 5 meters of overlapping waste, which will be included in the second resource input evaluation value.

[0087] The preset reasonable path distance D0 between adjacent unmanned aerial vehicles can be extracted from the parameter set, the path position information (such as path center coordinates) of the allocated unmanned aerial vehicles can be extracted from the path database of the allocated unmanned aerial vehicles, the remaining unmanned aerial vehicle list and the remaining partition list are initialized, then one unmanned aerial vehicle is selected from the remaining unmanned aerial vehicle list as a current calculation object, one partition is selected from the remaining partition list as a target partition in turn, the actual distance D between the path of the current unmanned aerial vehicle and the path of the allocated unmanned aerial vehicle is calculated, if D < D0, the overlapping space value (D0-D) is calculated and converted into a second space resource waste coefficient; if D > D0, the idle space value (D-D0) is calculated and converted into a second space resource waste coefficient, and the path planning cost is calculated by combining the span of the partition and the total length of the region, the second space resource waste coefficient and the path planning cost are weighted and summed to obtain a second resource input evaluation value, the unmanned aerial vehicle-partition-evaluation value is stored, and the combination with the smallest evaluation value is selected to complete the allocation of the current unmanned aerial vehicle (the unmanned aerial vehicle is removed from the remaining list, and the partition is removed from the remaining list), and the above process is repeated until the remaining unmanned aerial vehicle list is empty. The reason for this is that after the first unmanned aerial vehicle is allocated, if the path of the subsequent unmanned aerial vehicle is not reasonably spaced from the path of the allocated unmanned aerial vehicle, new resource waste (overlap or idle) will be generated, and if the spacing influence is ignored, the overall redundant resource evaluation will be low, which cannot reflect the actual inspection scene. Through the dynamic update of the remaining list, the problem of repeated partition allocation or no partition available for the unmanned aerial vehicle is avoided, and finally a preliminary partition-evaluation value matching relationship of all unmanned aerial vehicles is formed.

[0088] Illustratively, in S212, based on the path distance between adjacent unmanned aerial vehicles, for each remaining unmanned aerial vehicle, a second resource input evaluation value generated when the unmanned aerial vehicle is allocated to any remaining inspection partition as the next sequential unmanned aerial vehicle is calculated until all unmanned aerial vehicles are allocated, including:

[0089] In S2121, based on the path distance between adjacent unmanned aerial vehicles, for each remaining unmanned aerial vehicle, a second resource input evaluation value generated when the unmanned aerial vehicle is allocated to any remaining inspection partition as the next sequential unmanned aerial vehicle is calculated.

[0090] It can be understood that calculating the second resource input evaluation value of the remaining drones based on the path spacing between adjacent drones. The core is to incorporate the rationality of the path spacing into the resource waste evaluation to ensure the path coordination between the subsequently allocated drones and the already allocated drones. The path center coordinates of all the already allocated drones (such as the x-axis coordinates along the inspection direction) can be extracted from the path database of the already allocated drones, and the coverage range of the already allocated paths (such as the coordinates ± the drone coverage span / 2) can be determined. The preset reasonable path spacing D0 (that is, the minimum reasonable distance between the path centers of adjacent drones, which needs to be greater than half of the sum of the coverage spans of two drones to avoid overlap) can be read from the parameter set. For the current remaining drones and the target remaining partition, calculate the expected path center coordinates of the drone if it is allocated to the target partition (determined according to the partition center position and the inspection path planning rules), and calculate the distances D1, D2... D between this expected coordinate and all the already allocated path center coordinates n and select the minimum distance D min (the minimum distance is most likely to cause overlap or idleness). If D min < D0, calculate the overlap space value = (D0 - D min ) × the drone coverage width (the coverage range along the vertical inspection direction), and convert it into the second space resource waste coefficient; if D min > D0, calculate the idle space value = (D min - D0) × the drone coverage width, and convert it into the second space resource waste coefficient. At the same time, calculate the path planning cost of this partition according to the logic of S2113. Finally, weighted sum the second space resource waste coefficient and the path planning cost according to the preset weight to obtain the second resource input evaluation value. The reason for doing this is that the path spacing between adjacent drones is the key factor for generating redundant resources in group inspection (too small spacing causes overlap waste, too large spacing causes idle waste). If only referring to the parameters of the partition itself and ignoring the already allocated paths, it will cause conflicts between the newly allocated drones and the existing paths, exacerbating the overall resource waste; the effect is to ensure that the second resource input evaluation value can reflect the impact of individual allocation on the group as a whole, avoid isolated evaluation of the resource input of a single drone, and lay a foundation for minimizing the overall redundant resources

[0091] S2122. Repeat the calculation process of the second resource input evaluation value until all drones are allocated

[0092] It can be understood that the partition allocation of all drones (full permutation allocation of drones and inspection partitions) can be realized by performing the process of selecting a remaining drone, calculating a remaining partition evaluation value, determining an optimal allocation combination, and updating a remaining list through a loop, for example, after the first loop allocates one drone and one partition, there are 4 drones and 2 partitions left, after the second loop, there are 3 drones and 1 partition left, and so on until the fifth loop (at this time, there may be multiple drones allocated to the same partition, which needs to be selected in order according to the evaluation value) completes all allocations. The loop counter (records the number of allocations) and the allocation result list can be initialized, and the loop termination condition is set as the remaining drone list being empty, then the loop is entered: a drone is randomly selected from the remaining drone list (or selected in order according to the drone number), the logic of S2121 is called to calculate the second resource input evaluation value of the drone in all remaining partitions, the drone-partition combination with the minimum evaluation value is selected, the combination is added to the allocation result list, the drone is removed from the remaining drone list, if the remaining capacity of the partition (such as a partition can accommodate a maximum of 2 drones) is 0, the partition is removed from the remaining partition list, the loop counter is incremented by 1, if the loop counter exceeds the preset maximum value (such as twice the total number of drones, to avoid infinite loops) and the termination condition is still not met, an exception handling mechanism (such as recalculating the evaluation value) is triggered, until the remaining drone list is empty. The reason for this is that a single calculation can only complete the allocation of one drone, while group inspection requires all drones to have a clear partition, if the calculation is not repeated, some drones will not be allocated, affecting the overall inspection task execution; the effect of this is to form a complete drone-partition allocation result list, in which each drone has a corresponding partition and resource input evaluation value, and all allocations take into account the path impact of the allocated drones, providing a complete data basis for determining the final inspection sequence.

[0093] S213, summing the first resource input evaluation value and all second resource input evaluation values to obtain the resource input evaluation value generated when each drone is allocated to any inspection partition as the first drone.

[0094] It can be understood that the first resource input evaluation value refers to a comprehensive input index obtained by weighting the space waste coefficient and the path planning cost when the drone is allocated to any inspection partition as the first drone.

[0095] The second resource input evaluation value refers to an input index calculated according to factors such as the path distance between adjacent drones when each remaining drone enters the remaining partition in turn, representing the resource consumption of non-first drones entering the inspection task.

[0096] The comprehensive investment of the first unmanned aerial vehicle and the subsequent unmanned aerial vehicle in the inspection task can be calculated to form a whole input quantitative value. By reading the first resource input evaluation value of each unmanned aerial vehicle when it is assigned as the first unmanned aerial vehicle, then calling all second resource input evaluation values of other unmanned aerial vehicles in the corresponding sequence, then performing numerical accumulation operation, adding the first resource input evaluation value and the second resource input evaluation value of each subsequent unmanned aerial vehicle one by one, and finally outputting the sum result as the comprehensive resource input evaluation value of the current unmanned aerial vehicle under the condition of the partition. The reason for this is that the evaluation value of the first unmanned aerial vehicle alone cannot reflect the total resource consumption brought by the whole allocation, and the resource input of the subsequent unmanned aerial vehicle must be considered to obtain the complete input situation. The effect of the resource input evaluation value is to truly represent the total resource consumption caused by the first unmanned aerial vehicle in the whole inspection partition and the remaining allocation, thereby laying a numerical foundation for the subsequent extraction of redundant inspection resources.

[0097] S220, each resource input evaluation value is determined as a redundant inspection resource, obtaining a plurality of redundant inspection resources obtained by each unmanned aerial vehicle in the unmanned aerial vehicle inspection group entering the supervision target area one by one.

[0098] It can be understood that the resource input evaluation value here refers to the complete numerical result formed after the aforementioned weighting, difference and path calculation steps, which is used to reflect the overall input situation of the unmanned aerial vehicle under a specific task allocation. The redundant inspection resource is an index defined based on the resource input evaluation value, which is used to represent unnecessary or inefficient resource occupation in the inspection process. The larger the value, the more inefficient the input in the inspection process. The resource input evaluation value corresponding to all unmanned aerial vehicles can be directly mapped to a numerical set of redundant resources to form a data set that can reflect the overall redundancy of the inspection group.

[0099] The resource input evaluation value of each unmanned aerial vehicle can be read, then the preset mapping rule is called to confirm the resource input evaluation value as a redundant inspection resource value, then the redundant inspection resource set corresponding to the number of unmanned aerial vehicles is generated one by one according to the order of the unmanned aerial vehicles entering the supervision target area, and finally all redundant inspection resources are output as a complete set for the input of the subsequent task sorting and path allocation model.

[0100] The reason for this is that the resource input evaluation value is converted into a redundant inspection resource, which can uniformly quantify the inefficient resource occupation in the inspection process, so that the computer can compare, sort and optimize the redundancy in the subsequent steps. The effect is to form a directly available redundant inspection resource data set, which provides a basis for the sequence optimization of the unmanned aerial vehicle and the allocation of the task partition, which is beneficial to reduce repeated coverage and resource waste, thereby improving the overall efficiency of the inspection group.

[0101] S300, based on the plurality of redundant inspection resources, determine the order of each unmanned aerial vehicle in the unmanned aerial vehicle inspection group to be included in the inspection sequence and the task inspection partition corresponding to each unmanned aerial vehicle.

[0102] It can be understood that the redundant inspection resources corresponding to all unmanned aerial vehicles can be sorted, generally arranged from the smallest to the largest value, reflecting which unmanned aerial vehicles have the least redundancy and the highest efficiency when performing tasks. Subsequently, according to the sorting result, the unmanned aerial vehicle corresponding to the redundant resource with the smallest value is preferentially included in the inspection sequence, and the task inspection partition to be performed is determined. This process is repeated for each unmanned aerial vehicle until the order and inspection partition of all unmanned aerial vehicles are determined. By arranging the task order based on the principle of minimum redundancy, the repeated coverage and resource waste can be minimized, and the overall execution efficiency of the inspection group can be optimized. At the same time, the generated unmanned aerial vehicle inspection sequence and partition correspondence also provides clear input basis for the subsequent path allocation model, ensuring that path planning can be based on the optimized unmanned aerial vehicle sequence.

[0103] In one possible implementation, S300, based on the plurality of redundant inspection resources, determine the order of each unmanned aerial vehicle in the unmanned aerial vehicle inspection group to be included in the inspection sequence and the task inspection partition corresponding to each unmanned aerial vehicle, comprising:

[0104] S310, based on all redundant inspection resources from small to large, select the order of the unmanned aerial vehicle corresponding to the redundant inspection resource with the smallest value and the inspection partition as the order of the unmanned aerial vehicle to be included in the inspection sequence and the task inspection partition corresponding to each unmanned aerial vehicle.

[0105] It can be understood that all the redundant inspection resources corresponding to the unmanned aerial vehicles can be sorted in ascending order of value by comparison algorithm or sorting algorithm. The smaller the value, the less invalid or inefficient resource occupation the unmanned aerial vehicle produces during task execution, i.e. the higher the inspection efficiency. After sorting, the redundant inspection resource with the smallest value is identified, and the order of the unmanned aerial vehicle corresponding to the redundant inspection resource and the inspection partition are determined as the starting element of the current optimal inspection sequence. Please refer to Figure 3 According to the number of unmanned aerial vehicles and the distribution of task partitions, the unmanned aerial vehicles corresponding to the remaining redundant inspection resources after sorting and their inspection partitions are sequentially included in the inspection sequence to form a complete unmanned aerial vehicle inspection sequence list and partition correspondence. The purpose of this is to arrange the unmanned aerial vehicle with the least redundancy to perform the inspection task first, maximize the reduction of resource waste and repeated coverage, and make the entire inspection group have the highest efficiency when completing the task. At the same time, this sorting result provides clear input for the subsequent path allocation model, so that path planning can be based on the optimized unmanned aerial vehicle sequence, ensuring that each unmanned aerial vehicle covers the specified inspection partition in the optimal order, thereby improving the overall resource utilization and work efficiency of the inspection group.

[0106] S400, input the task inspection partition corresponding to each unmanned aerial vehicle in turn into a preset path allocation model to obtain path allocation data of each unmanned aerial vehicle; wherein the path allocation model is a machine learning model trained in advance.

[0107] It can be understood that the specific path data of each unmanned aerial vehicle can be generated by the path allocation model using the determined unmanned aerial vehicle inspection sequence and the corresponding task partition, thereby providing an executable flight route and scheduling scheme for actual inspection. The path allocation model is a machine learning model trained in advance, and its training process and application logic can be described in detail as follows:

[0108] In the model training stage, a training data set can be constructed in advance. The training data can include the execution sequence of each unmanned aerial vehicle in the historical inspection task, the allocated task inspection partition position, the actual flight path coordinates, the inspection completion time, the energy consumption data, and the environmental constraint conditions (such as wind speed, obstacle distribution, airspace restriction, etc.). The training data can come from actual inspection records, simulation or artificially designed reference task sequences. Each training sample can be represented as a corresponding relationship between an input vector and an output vector, wherein the input vector includes the unmanned aerial vehicle sequence, the inspection partition position and other task constraints, and the output vector is the coordinate sequence of the optimal flight path and related parameters.

[0109] Feature engineering processing is performed on the training data. The features include the spatial coordinates (two-dimensional or three-dimensional) of the inspection partition, the unmanned aerial vehicle takeoff point position, the range limit, the energy consumption index, the historical redundant resource value, the task priority, etc. These features are standardized, normalized or encoded for processing, so that the machine learning model can efficiently understand and utilize them.

[0110] The model selection can use reinforcement learning, deep neural network (DNN), graph neural network (GNN) or hybrid optimization algorithm. The reinforcement learning model optimizes the path by defining a reward function, for example, the reward can be the maximization of the efficiency of covering all inspection points, the minimization of total energy consumption, the minimization of redundant resources, etc. In the training process, the model continuously iterates and adjusts the strategy, so that the output path sequence can maximize the task efficiency under the given input (unmanned aerial vehicle sequence + inspection partition).

[0111] The training process can be roughly divided into the following steps: first, divide the training data into a training set and a validation set; second, the model is iterated on the training set for multiple rounds, and the model parameters are updated according to the loss function (such as path length error, time delay error, energy consumption error) in each round; then, evaluate on the validation set to detect the generalization ability of the model to unseen tasks; finally, according to the evaluation result, adjust the parameters for optimization until the model converges and meets the performance indicators.

[0112] After the training is completed, the path allocation model can receive the UAV inspection sequence and the inspection subarea as input, automatically generate the path allocation data of each UAV, including the flight path coordinate sequence, the estimated flight time, the energy consumption estimation, and the possible midway stopover or obstacle avoidance point, and realize the accurate and efficient inspection task.

[0113] S500, the path allocation data and the order of entering the corresponding task inspection subarea in the supervision target area in turn for each UAV in the UAV inspection group are issued.

[0114] It can be understood that the path allocation data and the order of entering the corresponding task inspection subarea in the supervision target area in turn for each UAV can be converted into the key connection action of the UAV executable instruction through a preset protocol or instruction set, a customized structured instruction package for each UAV. The path allocation data can include the complete coordinate sequence of the take-off point-inspection point-landing point, the recommended flight speed of each section, the inspection point stay time, and the safety constraints (such as the no-fly section and the minimum height), the order clearly indicates the absolute take-off time (such as 10:00 take-off) or the relative time sequence (such as starting 5 minutes after the previous UAV arrives at the subarea), and the instruction format is adapted to the UAV navigation system (such as JSON, manufacturer-defined protocol), without the need for secondary conversion; the instruction package-UAV one-to-one mapping can be established through the unique UAV identifier (such as the device number), which not only makes each UAV clear when to fly, which road to fly, and which subarea to be responsible for, avoids execution confusion, but also provides instruction basis for subsequent flight monitoring and abnormal adjustment, ensures that the entire inspection group executes in order according to the low-redundancy and high-efficiency preset scheme, optimizes not only the inspection efficiency of a single UAV, but also the operation system of the entire inspection group, thereby improving the completeness and accuracy of the engineering supervision data, ensuring the safety and efficiency of the inspection process, reducing the dependence on human experience in inspection, and comprehensively and systematically promoting the transformation of engineering supervision from human-led to digital-driven.

[0115] Optionally, the UAV is a multi-spectral imaging inspection UAV, and the subarea spans of the inspection subareas in the supervision target area are not completely the same, and the path distances between the UAVs are not completely the same.

[0116] It can be understood that the UAVs in the UAV group are multi-spectral imaging inspection UAVs, for example, the inspection UAV can be a 6-band multi-spectral camera agricultural inspection machine, an electric power inspection machine with insulator spectral recognition, an ecological inspection machine with red edge band collection, an industrial inspection machine with spectral monitoring, and various multi-spectral imaging inspection UAVs. Please refer to Figure 3The partition spans of the inspection sub-areas in the target area of supervision are not completely the same, for example, if the target area is 10,000 mu of farmland, the span of a continuous wheat planting area can be 60-80 meters, the span of a vegetable greenhouse concentrated area can be 10-15 meters, and the span of an orchard planting area can be 25-35 meters; the path distances between the unmanned aerial vehicles are not completely the same; the path distances between the unmanned aerial vehicles are not completely the same.

[0117] Corresponding to the digital supervision method based on unmanned aerial vehicle intelligent inspection and AI intelligent analysis of the above embodiments, the embodiments of the present application also provide a digital supervision system based on unmanned aerial vehicle intelligent inspection and AI intelligent analysis. Each unit of the system can implement each step of the digital supervision method based on unmanned aerial vehicle intelligent inspection and AI intelligent analysis. Figure 4 The structure block diagram of the digital supervision system based on unmanned aerial vehicle intelligent inspection and AI intelligent analysis provided by the embodiments of the present application is shown, and only the parts related to the embodiments of the present application are shown for ease of illustration.

[0118] Referring to Figure 4 The digital supervision system based on unmanned aerial vehicle intelligent inspection and AI intelligent analysis includes:

[0119] The acquisition unit is configured to acquire a parameter set of the unmanned aerial vehicle inspection group.

[0120] The detection unit is configured to calculate, based on the parameter set of the unmanned aerial vehicle inspection group, a plurality of redundant inspection resources obtained after each unmanned aerial vehicle in the unmanned aerial vehicle inspection group enters the target area of supervision one by one. The redundant inspection resources are used to reflect the invalid inspection investment generated in the unmanned aerial vehicle inspection process, and the numerical value of the redundant inspection resources reflects the degree of resource waste in the inspection process.

[0121] The task unit is configured to determine, based on the plurality of redundant inspection resources, the order of each unmanned aerial vehicle in the unmanned aerial vehicle inspection group in the inspection sequence and the corresponding task inspection sub-area of each unmanned aerial vehicle.

[0122] The path unit is configured to input the order and the corresponding task inspection sub-area of each unmanned aerial vehicle into a preset path allocation model to obtain path allocation data of the each unmanned aerial vehicle. The path allocation model is a machine learning model trained in advance.

[0123] The distribution unit is configured to distribute the path allocation data and the order of entering the corresponding task inspection sub-area in the target area of supervision to each unmanned aerial vehicle in the unmanned aerial vehicle inspection group.

[0124] It should be noted that the information interaction, execution process and the like between the above systems / units are based on the same concept as the method embodiments of the present application, and the specific functions and technical effects brought by them can be referred to the method embodiments part, which will not be repeated here.

[0125] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the above division of functional units and modules is exemplified, and in actual application, the above functions can be completed by different functional units and modules according to needs, that is, the internal structure of the system is divided into different functional units or modules to complete all or part of the above described functions. Each functional unit and module in the embodiment can be integrated in one processing unit, or each unit and module can be physically independent, or two or more unit modules can be integrated in one unit. The integrated unit can be realized in the form of hardware or software. In addition, the specific name of each functional unit and module is only for easy distinction, and does not limit the protection scope of the present application. The specific working process of the unit and module in the system can refer to the corresponding process in the foregoing method embodiments, which will not be repeated here.

[0126] The present application also provides an electronic device, Figure 5 The structure schematic diagram of the electronic device provided by an embodiment of the present application is shown in the figure. Figure 5 As shown in the figure, the electronic device 6 of this embodiment includes at least one processor 60 (only one is shown in the figure), at least one memory 61 (only one is shown in the figure) and a computer program 62 stored in the at least one memory 61 and executable on the at least one processor 60. When the processor 60 executes the computer program 62, the electronic device 6 realizes the steps in any of the above aluminum alloy casting defect prediction method embodiments, or the electronic device 6 realizes the functions of the units in the above system embodiments. Figure 5 Figure 5 Exemplarily, the computer program 62 can be divided into one or more units, which are stored in the memory 61 and executed by the processor 60 to complete the present application. The one or more units can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program 62 in the electronic device 6.

[0127] Exemplarily, the computer program 62 can be divided into one or more units, which are stored in the memory 61 and executed by the processor 60 to complete the present application. The one or more units can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program 62 in the electronic device 6.

[0128] The electronic device 6 can be a desktop computer, a notebook computer, a palm computer, a cloud server and other computing devices or terminal devices. The electronic device can include, but is not limited to, a processor 60 and a memory 61. Those skilled in the art can understand that, Figure 5 ​The electronic device 6 is merely an example and does not limit the electronic device 6, which can include more or fewer components than shown, or combine some components, or have different components, such as an input / output device, a network access device, a bus, and the like.

[0129] The processor 60 can be a central processing unit (CPU), and can also be other general-purpose processors, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, and the like. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0130] The memory 61 can be an internal storage unit of the electronic device 6, such as a hard disk or a memory of the electronic device 6 in some embodiments. The memory 61 can also be an external storage device of the electronic device 6, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, and the like, in other embodiments. Further, the memory 61 can include both an internal storage unit and an external storage device of the electronic device 6. The memory 61 is used to store an operating system, an application program, a boot loader, data, and other programs, such as program codes of the computer program, and the like. The memory 61 can also be used to temporarily store data that has been output or will be output.

[0131] The embodiments of the present application also provide a computer readable storage medium, which stores a computer program. The computer program is executed by a processor to implement the steps in any of the above method embodiments.

[0132] The embodiments of the present application provide a computer program product. When the computer program product is run on an electronic device, the electronic device implements the steps in any of the above method embodiments.

[0133] The integrated unit, if implemented in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, the present application can implement all or part of the processes in the above-mentioned embodiment methods through a computer program to instruct the relevant hardware to complete, and the computer program can be stored in a computer readable storage medium. When the processor executes the computer program, the steps of each method embodiment described above can be implemented. The computer program includes computer program code, which can be in the form of source code, object code, executable files or some intermediate forms. The computer readable medium can at least include any entity or device capable of carrying the computer program code to the electronic device, recording medium, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium. For example, U disk, mobile hard disk, magnetic disk or optical disk, etc. In some jurisdictions, according to legislation and patent practice, the computer readable medium can not be an electrical carrier signal and a telecommunication signal.

[0134] In the above embodiments, the description of each embodiment has its own focus, and the parts not described or recorded in detail in a certain embodiment can be referred to the relevant description of other embodiments.

[0135] Those skilled in the art can appreciate that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0136] In the embodiments provided by the present application, it should be understood that the disclosed digital supervision system / electronic device and method based on unmanned aerial vehicle intelligent inspection and AI intelligent analysis can be implemented in other ways. For example, the above-described digital supervision system / electronic device based on unmanned aerial vehicle intelligent inspection and AI intelligent analysis embodiment is only illustrative, for example, the division of the unit is only a logical function division, and actual implementation can have another division mode, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point, the coupling or direct coupling or communication connection between the displayed or discussed each other can be through some interface, device or unit indirect coupling or communication connection, which can be electrical, mechanical or other forms.

[0137] The units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may also be distributed to multiple network units. Part or all of the units can be selected to achieve the purpose of the embodiment scheme according to actual needs.

[0138] The above embodiments are only used to illustrate the technical solutions of the present application, but not limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that the technical solutions recorded in the foregoing embodiments can still be modified, or some technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A digital supervision method based on unmanned aerial vehicle (UAV) intelligent inspection and AI intelligent analysis, characterized in that, The method comprises: acquiring a parameter set of a UAV inspection group; based on the parameter set of the UAV inspection group, calculating a plurality of redundant inspection resources obtained after each UAV in the UAV inspection group enters a supervision target area one by one; wherein the redundant inspection resources are used to reflect invalid inspection investment generated in the UAV inspection process, and the numerical value of the redundant inspection resources reflects the degree of resource waste in the inspection process; based on the plurality of redundant inspection resources, determining the order of each UAV in the UAV inspection group in the inspection sequence and the corresponding task inspection partition of each UAV; inputting the order and the corresponding task inspection partition of each UAV into a preset path allocation model to obtain path allocation data of each UAV; wherein the path allocation model is a machine learning model trained in advance; issuing the path allocation data and the order of entering the corresponding task inspection partition in the supervision target area to each UAV in the UAV inspection group; the parameter set comprises the partition span of each inspection partition in the supervision target area, the path spacing between adjacent UAVs, and the total length of the supervision target area; based on the parameter set of the UAV inspection group, calculating a plurality of redundant inspection resources obtained after each UAV in the UAV inspection group enters a supervision target area one by one, comprising: based on the partition span of each inspection partition in the supervision target area, the path spacing between adjacent UAVs, and the total length of the supervision target area, calculating the resource investment evaluation value generated when each UAV in the UAV inspection group is assigned to any inspection partition as the first UAV; determining each resource investment evaluation value as a redundant inspection resource to obtain a plurality of redundant inspection resources obtained after each UAV in the UAV inspection group enters a supervision target area one by one; the calculation of the resource investment evaluation value generated when each UAV in the UAV inspection group is assigned to any inspection partition as the first UAV based on the partition span of each inspection partition in the supervision target area, the path spacing between adjacent UAVs, and the total length of the supervision target area, comprises: based on the partition span of each inspection partition in the supervision target area and the total length of the supervision target area, calculating, for each UAV in the UAV inspection group, a first resource investment evaluation value generated when the UAV is assigned to any inspection partition as the first UAV; based on the path spacing between adjacent UAVs, calculating, for each remaining UAV, a second resource investment evaluation value generated when the UAV is assigned to any remaining inspection partition as the next order UAV until all UAVs are assigned; wherein the second resource investment evaluation value is the resource investment degree of any UAV except the first UAV assigned to any remaining inspection partition. Summing up the first resource input evaluation value and all the second resource input evaluation values, the generated resource input evaluation value of each UAV as the first UAV assigned to any of the inspection sub-zones is obtained. 2.The digital supervision method based on unmanned aerial vehicle (UAV) intelligent inspection and AI intelligent analysis of claim 1, wherein The first resource input evaluation value generated when each UAV in the UAV inspection group is assigned as the first UAV to any of the inspection sub-zones in the supervision target area is calculated based on the sub-zone span of each inspection sub-zone in the supervision target area and the total length of the supervision target area, and includes: For each UAV in the UAV inspection group, the self-coverage span of the UAV in the inspection direction is obtained. For any inspection sub-zone, according to a preset space utilization evaluation rule, a first space resource waste coefficient between the self-coverage span of each UAV in the inspection direction and the sub-zone span of the inspection sub-zone when the UAV is assigned as the first UAV to the inspection sub-zone is calculated. The path planning cost required for the UAV to enter the inspection sub-zone when each UAV is assigned as the first UAV to the inspection sub-zone is calculated, wherein the path planning cost is determined based on the total length of the supervision target area and the sub-zone span of the inspection sub-zone. The first space resource waste coefficient and the path planning cost are weighted and summed to obtain the first resource input evaluation value generated when each UAV is assigned as the first UAV to the inspection sub-zone. All inspection sub-zones are traversed to obtain a plurality of first resource input evaluation values corresponding to each UAV being assigned as the first UAV to different inspection sub-zones. 3.The digital supervision method based on UAV intelligent inspection and AI intelligent analysis of claim 2, wherein, The first space resource waste coefficient between the self-coverage span of each UAV in the inspection direction and the sub-zone span of the inspection sub-zone when the UAV is assigned as the first UAV to the inspection sub-zone is calculated for any of the inspection sub-zones according to a preset space utilization evaluation rule, and includes: For any inspection sub-zone, the difference between the sub-zone span of the inspection sub-zone and the self-coverage span of the UAV when each UAV is assigned as the first UAV to the inspection sub-zone is calculated to obtain an initial idle space value in the inspection sub-zone. According to a preset space utilization evaluation rule, the initial idle space value is converted into a first space resource waste coefficient; wherein the first space resource waste coefficient is positively correlated with the initial idle space value. 4.The digital supervision method based on UAV intelligent inspection and AI intelligent analysis of claim 2, wherein, The path planning cost required for the UAV to enter the inspection sub-zone when each UAV is assigned as the first UAV to the inspection sub-zone is calculated, and includes: The path planning basic cost value of each UAV is obtained. The ratio of the sub-zone span of the inspection sub-zone to the total length of the supervision target area when each UAV is assigned as the first UAV to the inspection sub-zone is calculated to obtain the sub-zone space proportion of the inspection sub-zone. Based on the sub-zone space proportion of the inspection sub-zone and the path planning basic cost value of each UAV, the path planning cost required for each UAV to enter the inspection sub-zone is obtained. 5.The digital supervision method based on UAV intelligent inspection and AI intelligent analysis of claim 1, wherein, The second resource input evaluation value generated when each of the remaining drones is assigned to any remaining inspection sub-area as the next sequential drone is calculated based on the path distance between adjacent drones, until all drones are assigned. The second resource input evaluation value generated when each of the remaining drones is assigned to any remaining inspection sub-area as the next sequential drone is calculated based on the path distance between adjacent drones. The second resource input evaluation value calculation process is repeated until all drones are assigned. 6.The digital supervision method based on UAV intelligent inspection and AI intelligent analysis according to claim 1, wherein, The order of each drone in the drone inspection group in the inspection sequence and the corresponding task inspection sub-area of each drone are determined based on the plurality of redundant inspection resources. The order of each drone in the inspection sequence and the corresponding task inspection sub-area of each drone are selected based on the smallest redundant inspection resource value from small to large. 7.The digital supervision method based on UAV intelligent inspection and AI intelligent analysis of claim 1, wherein, The drones are inspection drones with multi-spectral imaging function, and the sub-area spans of each inspection sub-area in the target area are not completely the same, and the path distances between the drones are not completely the same.

8. A digital supervision system based on unmanned aerial vehicle intelligent inspection and AI intelligent analysis, characterized in that, The digital supervision system based on drone intelligent inspection and AI intelligent analysis for implementing the method of any one of claims 1 to 7 comprises: An acquisition unit is configured to acquire a parameter set of a drone inspection group. A detection unit is configured to calculate a plurality of redundant inspection resources obtained after each drone in the drone inspection group enters a target area for supervision based on the parameter set of the drone inspection group. The redundant inspection resource reflects the invalid inspection input generated during the drone inspection process, and the numerical value of the redundant inspection resource reflects the resource waste degree of the inspection process. A task unit is configured to determine the order of each drone in the drone inspection group in the inspection sequence and the corresponding task inspection sub-area of each drone based on a plurality of redundant inspection resources. A path unit is configured to input the order and the corresponding task inspection sub-area of each drone into a preset path allocation model to obtain path allocation data of each drone. The path allocation model is a machine learning model trained in advance. A distribution unit is configured to distribute the path allocation data and the order of entering the corresponding task inspection sub-area in the target area for supervision to each drone in the drone inspection group.

Images