An unmanned aerial vehicle inspection method and system based on an unmanned aerial vehicle nest

Abstract

The application relates to the technical field of unmanned aerial vehicle inspection, in particular to an unmanned aerial vehicle inspection method and system based on an unmanned aerial vehicle nest, which comprises the following steps: collecting geographic coordinate data of a target inspection area; determining the position coordinates of an inspection target based on the geographic coordinate data; obtaining real-time spatial coordinates of an inspection unmanned aerial vehicle based on a UWB positioning base station in the unmanned aerial vehicle nest; performing path planning on the position coordinates and the real-time spatial coordinates based on an improved A* algorithm to obtain a first inspection path; inputting the first inspection path into a pre-constructed unmanned aerial vehicle dynamics model to obtain a first flight control parameter set; and controlling the inspection unmanned aerial vehicle to perform an inspection operation based on the first flight control parameter set. The inspection path and result data are uploaded to a cloud analysis platform, combined with big data analysis and machine learning technology, the development trend of defects can be predicted, data support is provided for subsequent maintenance and repair, and more intelligent decision-making is achieved.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) inspection technology, and specifically to a UAV inspection method and system based on UAV nests. Background Technology

[0002] Traditional methods typically rely on manual planning of inspection paths, which makes it difficult to consider real-time changes and complexities in the environment, and can easily lead to suboptimal paths or low efficiency. After inspection, traditional methods only provide the location and status of defects, lacking prediction and analysis of future development trends, and often rely on human experience for speculation. Traditional methods usually use preset paths and fixed flight control strategies, which are difficult to adapt to dynamic adjustments in real-time flight. The image data of traditional methods often comes from a single perspective, which may lead to missed detections or incomplete information, and data calibration is relatively cumbersome. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art and provide a drone inspection method and system based on drone nests.

[0004] The technical solution adopted to solve the above-mentioned technical problems is: a drone inspection method based on drone nests, comprising:

[0005] Collect geographic coordinate data of the target inspection area, determine the location coordinates of the inspection target based on the geographic coordinate data, obtain the real-time spatial coordinates of the inspection drone based on the UWB positioning base station inside the drone nest, and perform path planning on the location coordinates and real-time spatial coordinates based on the improved A* algorithm to obtain the first inspection path.

[0006] The first inspection path is input into a pre-built UAV dynamics model to obtain a first set of flight control parameters, and the inspection UAV is controlled to perform inspection operations based on the first set of flight control parameters.

[0007] Based on the airborne vision sensor of the inspection drone, multi-view image data is collected during the inspection process, and the multi-view image data is spatiotemporally synchronized and calibrated to obtain inspection result data, wherein the inspection result data includes the spatial location of defects and defect image data of the target inspection area.

[0008] A three-dimensional defect coordinate system is constructed based on the spatial location of the defect and the geographic coordinate data. The first inspection path is corrected based on the Bézier curve fitting algorithm and the three-dimensional defect coordinate system to obtain the second inspection path.

[0009] The second inspection path is input into the UAV dynamics model to obtain the second flight control parameter set. The return operation is performed based on the second flight control parameter set, and the second inspection path and inspection result data are uploaded to the cloud analysis platform to obtain an inspection analysis report containing defect development trend prediction.

[0010] Preferably, a path is planned based on the improved A* algorithm using the location coordinates and real-time spatial coordinates to obtain the first inspection path, including:

[0011] Based on the location coordinates and real-time spatial coordinates, a gridded map covering all potential flight paths of the target inspection area is constructed, wherein each grid of the gridded map represents the smallest unit area that the inspection UAV can fly.

[0012] The improved A* algorithm is used to calculate the comprehensive cost of all potential flight paths in the gridded map, wherein the comprehensive cost includes path length cost and path risk cost;

[0013] The first inspection path is obtained based on the comprehensive cost.

[0014] Preferably, the UAV dynamics model includes aerodynamic parameters, UAV mass parameters, and flight environment parameters. The inspection UAV is subjected to dynamic simulation by inputting the first inspection path to output the first set of flight control parameters.

[0015] Preferably, the first inspection path is input into a pre-built UAV dynamics model to obtain a first set of flight control parameters, including:

[0016] Input the coordinate sequence of the first inspection path into the UAV dynamics model;

[0017] Based on the aerodynamic parameters, UAV mass parameters, and flight environment parameters, the flight state of the inspection UAV on the first inspection path is simulated to obtain a first set of flight control parameters for the inspection UAV on the first inspection path, including optimal flight speed, flight altitude, pitch angle, roll angle, and yaw angle.

[0018] Preferably, the multi-view image data is subjected to spatiotemporal synchronization calibration to obtain inspection result data, including:

[0019] Based on the image acquisition timestamps of the airborne vision sensor and the geographic coordinate data, the multi-view image data is time-synchronized and spatially aligned to obtain synchronized and calibrated multi-view image data.

[0020] Image fusion processing is performed on the synchronized calibrated multi-view image data to obtain a panoramic image of the target inspection area;

[0021] Defect detection is performed on the panoramic image based on deep learning algorithms to obtain the defect area in the target inspection area;

[0022] The spatial location of the defect area and the corresponding defect image data are obtained to obtain the inspection result data.

[0023] Preferably, defect detection is performed on the panoramic image based on a deep learning algorithm to obtain the defect region in the target inspection area, including:

[0024] A sample image database containing multiple defect types is constructed, wherein each sample image in the sample image database is labeled with the corresponding defect type and defect severity;

[0025] The panoramic image is input into a pre-trained deep learning model, wherein the deep learning model adopts a convolutional neural network architecture and is trained using the sample image database;

[0026] Based on the deep learning model, feature extraction and classification are performed on the panoramic image to obtain the defect areas in the target inspection area and their corresponding defect types and degrees.

[0027] Preferably, the first inspection path is corrected based on the Bézier curve fitting algorithm and the three-dimensional defect coordinate system to obtain the second inspection path, including:

[0028] Based on the spatial location of the defect in the three-dimensional defect coordinate system, determine the key points that need to be corrected in the first inspection path;

[0029] The key points are fitted with curves based on the Bézier curve fitting algorithm to obtain the corrected inspection path segment.

[0030] The modified inspection path segment is then concatenated with the remaining path segments in the first inspection path to obtain the second inspection path.

[0031] Preferably, the second inspection path and inspection result data are uploaded to a cloud analysis platform to obtain an inspection analysis report that includes a prediction of defect development trends, including:

[0032] Upload the second inspection path and inspection result data to the cloud analysis platform;

[0033] The cloud-based analysis platform performs in-depth analysis of the inspection results data based on big data technology to obtain a defect development trend prediction model for the target inspection area.

[0034] Based on the defect development trend prediction model, the target inspection area is predicted to obtain an inspection analysis report that includes the defect development trend prediction.

[0035] Preferably, the cloud-based analysis platform performs in-depth analysis of the inspection results data based on big data technology to obtain a defect development trend prediction model for the target inspection area, including:

[0036] A defect development database is constructed based on historical inspection data and the inspection result data, wherein the defect development database contains historical development trends of various defect types;

[0037] The defect development database is trained using machine learning algorithms to obtain a defect development trend prediction model.

[0038] The technical solution adopted to solve the above-mentioned technical problems is: a drone inspection system based on drone nests, which is applicable to the aforementioned drone inspection method based on drone nests, including:

[0039] The path construction unit is used to collect geographic coordinate data of the target inspection area, determine the location coordinates of the inspection target based on the geographic coordinate data, obtain the real-time spatial coordinates of the inspection drone based on the UWB positioning base station inside the drone nest, and perform path planning on the location coordinates and real-time spatial coordinates based on the improved A* algorithm to obtain the first inspection path.

[0040] The inspection execution unit is used to input the first inspection path into a pre-built UAV dynamics model to obtain a first flight control parameter set, and control the inspection UAV to perform inspection operations based on the first flight control parameter set.

[0041] The inspection result unit is used to collect multi-view image data during the inspection process based on the airborne vision sensor of the inspection drone, and to perform spatiotemporal synchronization calibration on the multi-view image data to obtain inspection result data. The inspection result data includes the spatial location of defects and defect image data of the target inspection area.

[0042] The path optimization unit is used to construct a three-dimensional defect coordinate system based on the spatial location of the defect and the geographic coordinate data, and to modify the first inspection path based on the Bézier curve fitting algorithm and the three-dimensional defect coordinate system to obtain a second inspection path.

[0043] The report generation unit is used to input the second inspection path into the UAV dynamics model to obtain a second flight control parameter set, perform a return operation based on the second flight control parameter set, and upload the second inspection path and inspection result data to the cloud analysis platform to obtain an inspection analysis report containing defect development trend prediction.

[0044] The beneficial effects of the present invention are as follows: (1) The present invention can dynamically optimize the path and adjust the inspection path in real time to cope with changes in the environment by using the improved A* algorithm for path planning, thereby improving the inspection efficiency and the accuracy of the path; (2) The present invention can generate a flight control parameter set based on the UAV dynamics model, and can accurately control the flight of the inspection UAV according to the different requirements of the first inspection path and the second inspection path, thereby improving the stability and accuracy of the inspection operation; (3) The present invention can predict the development trend of defects by uploading the inspection path and result data to the cloud analysis platform and combining big data analysis and machine learning technology, thereby providing data support for subsequent maintenance and repair, and helping to achieve more intelligent decision-making. Attached Figure Description

[0045] Figure 1 This is a schematic diagram of the overall method steps in one embodiment of the present invention;

[0046] Figure 2 This is a schematic diagram of the overall system flow in one embodiment of the present invention; reference numerals: 1, path construction unit; 2, inspection execution unit; 3, inspection result unit; 4, path optimization unit; 5, report generation unit. Detailed Implementation

[0047] Example 1, as Figure 1 As shown, the present invention proposes a drone inspection method based on drone nests, comprising:

[0048] S1. Collect geographic coordinate data of the target inspection area, determine the location coordinates of the inspection target based on the geographic coordinate data, obtain the real-time spatial coordinates of the inspection drone based on the UWB positioning base station inside the drone nest, and perform path planning on the location coordinates and real-time spatial coordinates based on the improved A* algorithm to obtain the first inspection path.

[0049] S2. Input the first inspection path into the pre-built UAV dynamics model to obtain the first flight control parameter set, and control the inspection UAV to perform inspection operations based on the first flight control parameter set;

[0050] S3. Based on the airborne vision sensor of the inspection drone, collect multi-view image data during the inspection process, and perform spatiotemporal synchronization calibration on the multi-view image data to obtain inspection result data. The inspection result data includes the spatial location of defects and defect image data of the target inspection area.

[0051] S4. Construct a three-dimensional defect coordinate system based on the spatial location and geographic coordinate data of the defect, and correct the first inspection path based on the Bézier curve fitting algorithm and the three-dimensional defect coordinate system to obtain the second inspection path.

[0052] S5. Input the second inspection path into the UAV dynamics model to obtain the second flight control parameter set. Perform the return operation based on the second flight control parameter set, and upload the second inspection path and inspection result data to the cloud analysis platform to obtain an inspection analysis report containing defect development trend prediction.

[0053] In this invention, geographic coordinate data refers to the location data of the target area, such as longitude, latitude, and altitude, obtained through GPS or other means; UWB refers to a broadband wireless communication technology widely used in positioning systems for accurately measuring the spatial position of drones; real-time spatial coordinates refer to the location coordinate data of the drone that is updated over time during flight; A* algorithm refers to an algorithm commonly used in path planning that finds the shortest path from the starting point to the target point through heuristic search; flight control parameter set refers to the various parameters used to control the flight of the drone; airborne vision sensor refers to a camera or other vision sensor mounted on the drone, used to collect image data of the ground or target area; three-dimensional defect coordinate system refers to a coordinate system that transforms the position of a defect into a three-dimensional space, used to represent the accurate position of the defect in space; Bézier curve fitting algorithm refers to a mathematical curve widely used for path smoothing and curve fitting, used to correct inspection paths, making the path smoother and more optimized; cloud analysis platform refers to a system platform that uploads inspection result data to a cloud server for processing, storage, and analysis, on which analysis such as defect development trend prediction can be performed.

[0054] Example 2: The UAV inspection method based on UAV nests proposed in this invention further includes the following features compared to Example 1:

[0055] A1. Based on the improved A* algorithm, path planning is performed on the location coordinates and real-time spatial coordinates to obtain the first inspection path, including:

[0056] A2. Construct a gridded map covering all potential flight paths of the target inspection area based on location coordinates and real-time spatial coordinates. Each grid of the gridded map represents the smallest unit area that the inspection drone can fly.

[0057] A3. Calculate the comprehensive cost of all potential flight paths in the gridded map based on the improved A* algorithm, where the comprehensive cost includes path length cost and path risk cost;

[0058] A4. Obtain the first inspection path based on the comprehensive cost.

[0059] In this embodiment, the gridded map refers to the flight space of the inspection area, which is used to simplify the complex environment and facilitate path planning algorithm processing; the comprehensive cost refers to the weighted sum of various factors that need to be considered in path planning, including path length cost and path risk cost; the first inspection path refers to the optimal path planned for the UAV after considering path length and risk cost.

[0060] In one optional embodiment, the UAV dynamics model includes aerodynamic parameters, UAV mass parameters, and flight environment parameters. The UAV is subjected to dynamic simulation by inputting a first inspection path to output a first set of flight control parameters.

[0061] It should be noted that the UAV dynamics model refers to the mathematical model of the UAV's motion behavior during flight. It describes the UAV's flight state and behavior under different conditions through different parameters (such as aerodynamic parameters, mass parameters, flight environment parameters, etc.). Aerodynamic parameters refer to the parameters of the UAV's interaction with the air, mainly including lift, drag, aerodynamic mass, wing lift coefficient, drag coefficient, etc. The first flight control parameter set refers to the set of parameters required for UAV control calculated based on the first inspection path and dynamic simulation results.

[0062] In an optional embodiment, the first inspection path is input into a pre-built UAV dynamics model to obtain a first set of flight control parameters, including:

[0063] B1. Input the coordinate sequence of the first inspection path into the UAV dynamics model;

[0064] B2. Based on aerodynamic parameters, UAV mass parameters, and flight environment parameters, the flight state of the inspection UAV on the first inspection path is simulated to obtain the first set of flight control parameters for the inspection UAV on the first inspection path, including optimal flight speed, flight altitude, pitch angle, roll angle, and yaw angle.

[0065] In an optional embodiment, spatiotemporal synchronization calibration is performed on the multi-view image data to obtain inspection result data, including:

[0066] C1. Based on the image acquisition timestamps and geographic coordinate data of the airborne vision sensor, perform time synchronization and spatial alignment on the multi-view image data to obtain synchronized and calibrated multi-view image data.

[0067] C2. Perform image fusion processing on the multi-view image data after synchronous calibration to obtain a panoramic image of the target inspection area;

[0068] C3. Defect detection is performed on panoramic images based on deep learning algorithms to obtain defect areas in the target inspection area;

[0069] C4. Obtain the spatial location of the defect area and the corresponding defect image data to obtain the inspection result data.

[0070] It should be noted that image acquisition timestamps refer to the precise time record of each image acquisition. Timestamps help mark the specific acquisition time of an image, thus enabling time synchronization of images during processing. Time synchronization refers to aligning image data from different perspectives according to their acquisition time, ensuring that images captured by multiple sensors are consistent in time. Spatial alignment refers to aligning images captured from different perspectives spatially using geographic coordinate data, mapping images from different locations (viewpoints) to the same coordinate system to form unified spatial data. Image fusion refers to combining image data from multiple perspectives to form a panoramic image with higher information density and a wider field of view. Deep learning algorithms refer to convolutional neural networks (CNNs), etc., used to automatically extract features from image data and automatically detect defect areas in images. Defect detection refers to identifying potential defect areas in an image, such as cracks, corrosion, and damage. The spatial location of a defect area refers to the actual location of the defect in three-dimensional space, usually accurately located by combining image data with geographic coordinate data. Defect image data refers to image segments or regions containing defect information.

[0071] In an optional embodiment, defect detection is performed on the panoramic image based on a deep learning algorithm to obtain defect regions in the target inspection area, including:

[0072] D1. Construct a sample image database containing multiple defect types, where each sample image in the database is labeled with its corresponding defect type and defect severity.

[0073] D2. Input the panoramic image into a pre-trained deep learning model, wherein the deep learning model adopts a convolutional neural network architecture and is trained through a sample image database;

[0074] D3. Based on a deep learning model, feature extraction and classification are performed on the panoramic image to obtain the defect areas in the target inspection area and their corresponding defect types and degrees.

[0075] It should be noted that the sample image database refers to a repository containing a large number of image datasets, which are used for training deep learning models; feature extraction refers to the deep learning model extracting features from the input panoramic image that help determine the content of the image through methods such as convolutional layers.

[0076] In an optional embodiment, the first inspection path is modified based on a Bézier curve fitting algorithm and a three-dimensional defect coordinate system to obtain a second inspection path, including:

[0077] E1. Based on the spatial location of the defect in the three-dimensional defect coordinate system, determine the key points that need to be corrected in the first inspection path;

[0078] E2. Based on the Bézier curve fitting algorithm, curve fitting is performed on the key points to obtain the corrected inspection path segment.

[0079] E3. The corrected inspection path segment is spliced ​​with the remaining path segments in the first inspection path to obtain the second inspection path.

[0080] It should be noted that key points refer to special locations that need to be corrected or adjusted. These locations are determined based on the spatial location of the defects and the inspection path. The second inspection path is a new inspection path formed by splicing the corrected inspection path segment with other unmodified path segments in the first inspection path.

[0081] In an optional embodiment, the second inspection path and inspection result data are uploaded to a cloud analysis platform to obtain an inspection analysis report that includes a defect development trend prediction, including:

[0082] F1. Upload the second inspection path and inspection result data to the cloud analysis platform;

[0083] F2. The cloud-based analysis platform uses big data technology to conduct in-depth analysis of inspection results data to obtain a predictive model of defect development trends in the target inspection area.

[0084] F3. Based on the defect development trend prediction model, predict the target inspection area to obtain an inspection analysis report that includes the defect development trend prediction.

[0085] It should be noted that big data technology refers to a technology for processing, analyzing, and storing large-scale datasets, which usually involves tools and methods such as distributed computing, data mining, and machine learning; a defect development trend prediction model is a model that uses data mining or machine learning algorithms to predict the future development trend and pattern of defects by analyzing historical inspection data and current defect status. Such models can usually give the possible evolution of defects in the future.

[0086] In an optional embodiment, the cloud-based analytics platform performs in-depth analysis of the inspection results data based on big data technology to obtain a defect development trend prediction model for the target inspection area, including:

[0087] G1. Construct a defect development database based on historical inspection data and inspection result data. The defect development database contains historical development trends of various defect types.

[0088] G2. Train the defect development database based on machine learning algorithms to obtain a defect development trend prediction model.

[0089] It should be noted that historical inspection data refers to defect data collected from all inspection activities over a past period; the defect development database is a systematic database that stores historical defect data, which includes the development trends of various defect types over different time periods; and machine learning algorithms refer to models trained by analyzing historical inspection data that can predict future defect occurrence and development trends.

[0090] Example 3, as Figure 2 As shown, the present invention proposes a drone inspection system based on drone nests, which is applicable to a drone inspection method based on drone nests, including:

[0091] Path construction unit 1 is used to collect geographic coordinate data of the target inspection area, determine the location coordinates of the inspection target based on the geographic coordinate data, obtain the real-time spatial coordinates of the inspection drone based on the UWB positioning base station inside the drone nest, and perform path planning on the location coordinates and real-time spatial coordinates based on the improved A* algorithm to obtain the first inspection path.

[0092] Inspection execution unit 2 is used to input the first inspection path into the pre-built UAV dynamics model to obtain the first flight control parameter set, and control the inspection UAV to perform inspection operations based on the first flight control parameter set;

[0093] Inspection result unit 3 is used to collect multi-view image data during the inspection process based on the airborne vision sensor of the inspection drone, and to perform spatiotemporal synchronization calibration on the multi-view image data to obtain inspection result data. The inspection result data includes the spatial location of defects and defect image data of the target inspection area.

[0094] The path optimization unit 4 is used to construct a three-dimensional defect coordinate system based on the spatial location and geographic coordinate data of the defect, and to correct the first inspection path based on the Bézier curve fitting algorithm and the three-dimensional defect coordinate system to obtain the second inspection path.

[0095] The report generation unit 5 is used to input the second inspection path into the UAV dynamics model to obtain the second flight control parameter set, perform the return operation based on the second flight control parameter set, and upload the second inspection path and inspection result data to the cloud analysis platform to obtain an inspection analysis report containing defect development trend prediction.

[0096] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. A drone inspection method based on drone nests, characterized in that, include: Collect geographic coordinate data of the target inspection area, determine the location coordinates of the inspection target based on the geographic coordinate data, obtain the real-time spatial coordinates of the inspection drone based on the UWB positioning base station inside the drone's nest, and based on the improved A... The algorithm performs path planning on the location coordinates and real-time spatial coordinates to obtain the first inspection path; The first inspection path is input into a pre-built UAV dynamics model to obtain a first set of flight control parameters, and the inspection UAV is controlled to perform inspection operations based on the first set of flight control parameters. Based on the airborne vision sensor of the inspection drone, multi-view image data is collected during the inspection process, and the multi-view image data is spatiotemporally synchronized and calibrated to obtain inspection result data, wherein the inspection result data includes the spatial location of defects and defect image data of the target inspection area. A three-dimensional defect coordinate system is constructed based on the spatial location of the defect and the geographic coordinate data. The first inspection path is corrected based on the Bézier curve fitting algorithm and the three-dimensional defect coordinate system to obtain the second inspection path. The second inspection path is input into the UAV dynamics model to obtain the second flight control parameter set. The return operation is performed based on the second flight control parameter set, and the second inspection path and inspection result data are uploaded to the cloud analysis platform to obtain an inspection analysis report containing defect development trend prediction. Based on improved A The algorithm performs path planning on the location coordinates and real-time spatial coordinates to obtain the first inspection path, including: Based on the location coordinates and real-time spatial coordinates, a gridded map covering all potential flight paths of the target inspection area is constructed, wherein each grid of the gridded map represents the smallest unit area that the inspection UAV can fly. Based on the improved A The algorithm calculates the comprehensive cost of all potential flight paths in the gridded map, wherein the comprehensive cost includes path length cost and path risk cost; The first inspection path is obtained based on the comprehensive cost; The UAV dynamics model includes aerodynamic parameters, UAV mass parameters, and flight environment parameters. By inputting the first inspection path, the inspection UAV is subjected to dynamic simulation to output the first set of flight control parameters.

2. The UAV inspection method based on UAV nests according to claim 1, characterized in that, The first inspection path is input into a pre-built UAV dynamics model to obtain a first set of flight control parameters, including: Input the coordinate sequence of the first inspection path into the UAV dynamics model; Based on the aerodynamic parameters, UAV mass parameters, and flight environment parameters, the flight state of the inspection UAV on the first inspection path is simulated to obtain a first set of flight control parameters for the inspection UAV on the first inspection path, including optimal flight speed, flight altitude, pitch angle, roll angle, and yaw angle.

3. The UAV inspection method based on UAV nests according to claim 2, characterized in that, The multi-view image data is spatiotemporally synchronized and calibrated to obtain inspection result data, including: Based on the image acquisition timestamps of the airborne vision sensor and the geographic coordinate data, the multi-view image data is time-synchronized and spatially aligned to obtain synchronized and calibrated multi-view image data. Image fusion processing is performed on the synchronized calibrated multi-view image data to obtain a panoramic image of the target inspection area; Defect detection is performed on the panoramic image based on deep learning algorithms to obtain the defect area in the target inspection area; The spatial location of the defect area and the corresponding defect image data are obtained to obtain the inspection result data.

4. The UAV inspection method based on UAV nests according to claim 3, characterized in that, Defect detection is performed on the panoramic image based on a deep learning algorithm to obtain the defect region in the target inspection area, including: A sample image database containing multiple defect types is constructed, wherein each sample image in the sample image database is labeled with the corresponding defect type and defect severity; The panoramic image is input into a pre-trained deep learning model, wherein the deep learning model adopts a convolutional neural network architecture and is trained using the sample image database; Based on the deep learning model, feature extraction and classification are performed on the panoramic image to obtain the defect areas in the target inspection area and their corresponding defect types and degrees.

5. The UAV inspection method based on UAV nests according to claim 4, characterized in that, The first inspection path is corrected based on the Bézier curve fitting algorithm and the three-dimensional defect coordinate system to obtain the second inspection path, including: Based on the spatial location of the defect in the three-dimensional defect coordinate system, determine the key points that need to be corrected in the first inspection path; The key points are fitted with curves based on the Bézier curve fitting algorithm to obtain the corrected inspection path segment. The modified inspection path segment is then concatenated with the remaining path segments in the first inspection path to obtain the second inspection path.

6. The UAV inspection method based on UAV nests according to claim 5, characterized in that, The second inspection path and inspection result data are uploaded to the cloud analysis platform to obtain an inspection analysis report that includes defect development trend prediction, including: Upload the second inspection path and inspection result data to the cloud analysis platform; The cloud-based analysis platform performs in-depth analysis of the inspection results data based on big data technology to obtain a defect development trend prediction model for the target inspection area. Based on the defect development trend prediction model, the target inspection area is predicted to obtain an inspection analysis report that includes the defect development trend prediction.

7. The UAV inspection method based on UAV nests according to claim 6, characterized in that, The cloud-based analytics platform performs in-depth analysis of the inspection results data based on big data technology to obtain a defect development trend prediction model for the target inspection area, including: A defect development database is constructed based on historical inspection data and the inspection result data, wherein the defect development database contains historical development trends of various defect types; The defect development database is trained using machine learning algorithms to obtain a defect development trend prediction model.

8. A drone inspection system based on drone nests, applicable to the drone inspection method based on drone nests as described in any one of claims 1-7, characterized in that, include: Path construction unit (1), the path construction unit (1) is used to collect geographic coordinate data of the target inspection area, determine the location coordinates of the inspection target based on the geographic coordinate data, obtain the real-time spatial coordinates of the inspection drone based on the UWB positioning base station inside the drone nest, and based on the improved A The algorithm performs path planning on the location coordinates and real-time spatial coordinates to obtain the first inspection path; Inspection execution unit (2), the inspection execution unit (2) is used to input the first inspection path into the pre-built UAV dynamics model to obtain the first flight control parameter set, and control the inspection UAV to perform inspection operations based on the first flight control parameter set; Inspection result unit (3) is used to collect multi-view image data during the inspection process based on the airborne vision sensor of the inspection drone, and to perform spatiotemporal synchronization calibration on the multi-view image data to obtain inspection result data, wherein the inspection result data includes the spatial location of defects and defect image data of the target inspection area. The path optimization unit (4) is used to construct a three-dimensional defect coordinate system based on the spatial location of the defect and the geographic coordinate data, and to correct the first inspection path based on the Bezier curve fitting algorithm and the three-dimensional defect coordinate system to obtain the second inspection path. The report generation unit (5) is used to input the second inspection path into the UAV dynamics model to obtain the second flight control parameter set, perform the return operation based on the second flight control parameter set, and upload the second inspection path and inspection result data to the cloud analysis platform to obtain an inspection analysis report containing defect development trend prediction.

Images