Unmanned aerial vehicle assisted porcelain bottle defect off-line inspection method and system

Abstract

The invention relates to the field of electric power inspection, in particular to an unmanned aerial vehicle assisted porcelain bottle defect off-line inspection method and system. The method comprises the steps that the unmanned aerial vehicle collects real-time environment data and carries out imaging and data acquisition on porcelain bottles along the line, and images and three-dimensional model data of the porcelain bottles are obtained; performing primary processing and analysis to obtain a primary analysis result; the preliminary analysis result and the real-time environment data are analyzed, when the suspected defect is displayed, the flight route and the inspection strategy of the unmanned aerial vehicle are adjusted, key review is conducted on the porcelain bottle area with the suspected defect, and a review analysis result is obtained; adjusting the flight route and the inspection strategy of the unmanned aerial vehicle, and continuing to inspect according to the new flight route and the new inspection strategy; when the preliminary analysis result is displayed to be intact, continuing to inspect according to the original flight route; and transmitting all analysis results and the flight path adjustment instruction back to the ground control station in real time until the inspection is completed. According to the invention, the problems of changeable unmanned aerial vehicle inspection environments and task requirements can be effectively solved.

Description

Technical Field

[0001] The present invention relates to the field of electric power inspection, and in particular to a method and system for unmanned aerial vehicle-assisted offline inspection of porcelain bottle defects. Background Art

[0002] In the power system, porcelain bottles, as a type of insulator, play a key insulating role, and the quality of their performance directly affects the stable operation of the power system. Traditional porcelain bottle inspection methods mainly rely on manual climbing or observation using fixed cameras, which have many limitations. First, manual inspections are inefficient and difficult to cover a wide area, especially in areas with complex terrain or difficult to access. Second, manual inspections are greatly affected by subjective factors, and there may be cases of missed inspections or misjudgments. In addition, the viewing angle and range of fixed cameras are limited, and it is impossible to conduct a comprehensive inspection of porcelain bottles. With the development of drone technology, drone-assisted inspections provide a new solution, but they still face the problems of low automation and limited data processing capabilities.

[0003] Existing drone inspection systems mainly rely on preset routes and simple image recognition technology, lack real-time data processing and intelligent decision-making capabilities, and cannot effectively respond to changing inspection environments and task requirements. For example, patent document CN116029974A describes an insulator defect detection method and system based on smart terminals and drones. This method uses a camera carried by a drone to shoot the insulator to obtain image information of the insulator, and performs preprocessing and detection through a smart terminal to obtain the detection results. However, this method mainly focuses on the acquisition and preliminary processing of images, and lacks a description of how to dynamically adjust the inspection strategy and flight route based on the detection results.

[0004] Another patent document CN117437188A proposes an insulator defect detection system for smart grids. The system includes an acquisition module, a detection module, a control module and a communication module. The drone collects environmental images and performs image enhancement. The detection module detects the enhanced image to obtain the detection result. Although the system takes into account the role of the control module and the communication module, its specific implementation details in real-time data processing and intelligent decision-making still need to be improved.

[0005] These limitations lead to low inspection efficiency and accuracy, as well as poor robustness in complex scenarios, which limits the widespread application of drone inspection systems. Therefore, how to improve the automation level, data processing capabilities, and intelligent decision-making capabilities of drone inspections and effectively respond to changing inspection environments and task requirements has become a technical problem that needs to be solved urgently. Summary of the invention

[0006] The purpose of the present invention is to provide a drone-assisted offline inspection method and system for porcelain bottle defects, so as to solve the problem that the existing drone inspection lacks real-time data processing and intelligent decision-making capabilities and cannot effectively cope with the changing inspection environment and task requirements.

[0007] To achieve the above purpose, the following technical solution is adopted.

[0008] A drone-assisted offline inspection method for porcelain bottle defects, comprising:

[0009] The drone uses a multi-sensor system to collect real-time environmental data and image and collect data on the porcelain bottles along the route while executing a flight route covering all preset porcelain bottle locations, obtaining images and three-dimensional model data of the porcelain bottles.

[0010] Based on the image and three-dimensional model data of the porcelain bottle, preliminary processing and analysis are performed to identify the defects of the porcelain bottle and obtain preliminary analysis results; the preliminary analysis results and real-time environmental data are analyzed, and when the preliminary analysis results show that the current porcelain bottle is suspected to have defects, the flight route and inspection strategy of the drone are adjusted, and the suspected defective porcelain bottle area is re-inspected to obtain re-inspection analysis results; according to the re-inspection analysis results and real-time environmental data, the flight route and inspection strategy of the drone are adjusted, and the inspection is continued according to the new flight route and inspection strategy; when the preliminary analysis results show that the current porcelain bottle is intact, the drone continues to inspect according to the original flight route;

[0011] All analysis results and flight route adjustment instructions are transmitted back to the ground control station in real time until the inspection is completed.

[0012] Optionally, the step of setting the flight route covering all preset porcelain bottle positions specifically includes:

[0013] Collect the location data and related geographic information of the porcelain bottles, including but not limited to the latitude and longitude coordinates of the porcelain bottles, topographic data and historical data of previous inspections;

[0014] Analyze the location data, geographic information and historical data collected in the step to assess obstacles, climatic conditions and accessibility of the porcelain vase that may affect the flight of the drone, and determine a safe flight path and necessary flight parameters;

[0015] Using the evaluation results and the flight performance parameters of the drone, including flight speed, endurance, and steering angle limits, an optimal flight route covering all preset porcelain bottle locations is automatically planned;

[0016] Through simulation or actual test flights, verify whether the planned optimal flight route meets the safety and efficiency requirements, and ensure that the route can cover all preset porcelain bottle positions without obstacles;

[0017] According to the verification results, make necessary adjustments to the flight route, optimize the flight path, and form the final flight route;

[0018] The determined final flight route is converted into a set of flight instructions executable by the drone, including the take-off point, flight path, shooting point and expected flight speed, to guide the drone to perform inspection tasks.

[0019] Optionally, based on the obtained image and three-dimensional model data of the porcelain bottle, preliminary processing and analysis are performed to identify defects of the porcelain bottle and obtain preliminary analysis results, specifically including:

[0020] The porcelain bottle images and 3D model data collected by the drone are transmitted to the onboard processing system in real time;

[0021] De-noising and enhancing the transmitted image data to obtain optimized image data;

[0022] Use deep learning algorithms to extract features from optimized images and identify key features and potential defect areas on the surface of the porcelain bottle;

[0023] Combined with the extracted features, the 3D model data is analyzed to identify structural anomalies and morphological deviations of the porcelain bottle;

[0024] Comprehensively compare the analysis results, identify possible defects, and form preliminary analysis results;

[0025] The preliminary analysis results of the identification are recorded in the storage device and associated with the corresponding porcelain bottle location information for subsequent review and analysis.

[0026] Optionally, the preliminary analysis results and real-time environmental data are analyzed. When the preliminary analysis results show that the current porcelain bottle is suspected to have defects, the flight route and inspection strategy of the drone are adjusted to focus on re-inspecting the porcelain bottle area suspected of defects, and the steps of obtaining the re-inspection analysis results include:

[0027] Based on the preliminary analysis results, use the preset defect recognition standards to determine whether the porcelain bottle has suspected defects, record the location and type of the suspected defects, and obtain the suspected defect identification signal;

[0028] Real-time synchronization of the current flight environment conditions of the drone, including but not limited to wind speed, wind direction, temperature, humidity, and the drone's own battery status and remaining power, to obtain environmental status data;

[0029] Based on suspected defect identification signals and environmental status data, the flight route of the drone is dynamically adjusted to ensure that the drone can safely and efficiently approach and review suspected defect areas, and generate an optimized flight route;

[0030] According to the location, type and environmental status data of suspected defects, the inspection strategy of the drone is intelligently set, including shooting angle, shooting mode, and image resolution parameters, to form a customized inspection strategy for suspected defect areas;

[0031] According to the optimized flight route and customized inspection strategy, the drone performs a focused review of suspected defect areas, collects high-resolution images and 3D model data, and obtains a focused review data set;

[0032] Conduct in-depth analysis on the obtained key review data sets, use deep learning models to accurately classify and evaluate suspected defects, determine the nature, size and possible impact of the defects, and output defect review analysis reports;

[0033] Organize the defect review and analysis report, associate it with the detailed information and location data of the porcelain bottle, generate a complete inspection report, including the confirmation, location, type, severity and repair suggestions of the defect, and transmit the report to the ground control station.

[0034] Optionally, based on the review analysis results and real-time environmental data, adjust the flight path and inspection strategy of the drone, including:

[0035] According to the review and analysis results, the defective porcelain bottles are identified, and the specific locations and defect characteristics of the defective porcelain bottles are recorded to obtain the defective porcelain bottle information;

[0036] Based on the defective porcelain bottle information, the relative position relationship and correlation between the defective porcelain bottle and other porcelain bottles are analyzed, and other related or similar porcelain bottles that need to be checked are determined to obtain a list of related porcelain bottles;

[0037] Collect detailed data of the drone's current environment in real time, including but not limited to wind speed, wind direction, temperature, and humidity, to obtain a real-time environmental data set;

[0038] According to the relevant porcelain bottle list and real-time environmental data set, the flight route of the drone is dynamically adjusted to optimize the flight path, giving priority to covering all relevant porcelain bottles to obtain the optimized flight route;

[0039] According to the list of relevant porcelain bottles, the inspection strategy is intelligently adjusted, including shooting parameters, review frequency and key focus areas, to form a customized inspection strategy for relevant porcelain bottles;

[0040] The drone performs inspections according to the optimized flight route and customized inspection strategies for the relevant porcelain bottles, giving priority to re-inspecting the relevant porcelain bottles and obtaining adjusted inspection results;

[0041] The adjusted inspection results are recorded in a storage device and evaluated to determine whether related porcelain bottles have similar defects, and an inspection result report is obtained and fed back to the ground control station in real time.

[0042] Optionally, the following steps are also included:

[0043] Based on the analysis results and flight path adjustment instructions of the drone, evaluate the drone's inspection path and strategy to determine whether manual adjustments are needed, including adjusting the shooting angle, replanning the flight path, or performing a more detailed scan of the porcelain bottle area suspected of defects;

[0044] When manual intervention is determined to be necessary, remote commands are sent to the drone through the ground control station, including but not limited to changing the flight altitude, turning, hovering, or performing specific actions to optimize the detection process;

[0045] Feedback data after the drone executes remote commands is transmitted back to the ground control station in real time, and the operator evaluates the intervention effect and confirms whether the drone has correctly adjusted its flight route and inspection strategy according to the new instructions;

[0046] If the drone encounters an emergency or the ground control station observes an abnormal situation, the ground control station will immediately activate the emergency plan to ensure the drone's safe return or take necessary risk avoidance measures.

[0047] Optionally, the following steps are also included: comparing and analyzing the review analysis results with historical inspection data, monitoring the changing trend of the porcelain bottle status through a pre-trained deep learning model, predicting the development of potential defects, obtaining defect prediction results, and using the defect prediction results to formulate the next flight route and inspection strategy.

[0048] Optionally, the obtained key review data set is analyzed in depth, and the suspected defects are accurately classified and evaluated using a deep learning model to determine the nature, size, and possible impact of the defects. The steps of outputting a defect review analysis report include:

[0049] Applying an adaptive image enhancement algorithm to the collected images to improve the contrast, and performing gridding processing on the 3D model data to obtain an enhanced image data set and gridded 3D model data;

[0050] A deep convolutional autoencoder is used to learn and extract feature vectors representing defects from the enhanced image dataset, and the abnormal areas in the model are identified through a feature matching algorithm of the 3D model data to obtain defect feature vectors and abnormal area mapping.

[0051] Combining the defect feature vector of the image data set and the abnormal area mapping of the 3D model data, a multi-dimensional feature fusion technology is used to form a panoramic view of the defect features of the porcelain bottle, and a fused panoramic view of the defect features is obtained;

[0052] Based on the fused panoramic view of defect features, the pre-trained support vector machine classifier is used to accurately classify suspected defects, distinguish different types of defects, including cracks, stains, and material defects, and obtain defect classification results;

[0053] Quantitatively analyze the defect classification results, use geometric measurement algorithms to accurately measure the size of the defects, including length, width and depth, and use image processing technology to evaluate the area and distribution of the defects to obtain quantitative defect parameters;

[0054] Finite element analysis is used to simulate the impact of each quantitative defect on the mechanical and electrical properties of the porcelain bottle under operating conditions. The simulation results include stress distribution, electric field strength and potential failure modes, and a simulation analysis report is obtained;

[0055] Intelligent recommendation of maintenance priority: Based on the defect classification results and simulation analysis reports, the decision tree algorithm in machine learning is applied to assign maintenance priority to each defect and make maintenance suggestions, including maintenance time window and required materials, to obtain maintenance priority recommendations and maintenance suggestions;

[0056] Generate a detailed report containing images, 3D models, defect features, classification results, quantitative parameters, simulation analysis reports and repair suggestions based on defect analysis results and repair priority recommendations to obtain an interactive defect analysis report;

[0057] The interactive defect analysis report is reviewed by the expert system. The expert system provides review opinions and report optimization suggestions based on preset rules and historical data. The report is automatically optimized according to the review opinions to obtain the optimized defect analysis report.

[0058] The optimized defect analysis report is associated with the detailed information and location data of the porcelain bottle, and the report is updated in real time to reflect the latest status and maintenance history of the porcelain bottle, and obtain a real-time updated associated report;

[0059] Based on the maintenance priority recommendations and maintenance suggestions, a maintenance work scheduling plan is generated to optimize resource allocation and predict the potential impact of maintenance activities on power grid operation. The scheduling plan is optimized using a genetic algorithm to obtain an optimized maintenance scheduling plan.

[0060] Optionally, according to the relevant porcelain bottle list and the real-time environmental data set, the flight route of the UAV is dynamically adjusted to optimize the flight path, and all relevant porcelain bottles are preferentially covered to obtain the optimized flight route, specifically including:

[0061] Use geographic information system to perform spatial analysis on the obtained list of related porcelain bottles, determine the relative positions and distances between the porcelain bottles, and the relationship between the related porcelain bottles and the current position of the drone, and obtain the spatial analysis results;

[0062] Through integrated sensors and external data sources, detailed data about the drone’s environment is collected in real time, including wind speed, wind direction, temperature, humidity, as well as the drone’s battery status and remaining power, to obtain a real-time environmental data set.

[0063] Based on the spatial analysis results and the real-time environmental data set, a path planning algorithm is used to first cover all relevant porcelain bottles and then inspect the remaining porcelain bottles to obtain a preliminary flight route;

[0064] Conduct risk assessment on the generated preliminary flight route, taking into account factors such as terrain obstacles, weather conditions and no-fly zones, identify potential risk points in the route, and obtain a risk assessment report;

[0065] According to the risk assessment report, the preliminary flight route is dynamically adjusted to avoid risk points and obtain an optimized flight route;

[0066] According to the optimized flight route and environmental data, the flight parameters of the UAV are adjusted to adapt to different flight sections and environmental conditions to obtain adjusted flight parameters;

[0067] The adjusted flight parameters and optimized flight route are transmitted to the control system of the UAV. The UAV executes the confirmed flight route, conducts key inspections on the relevant porcelain bottles, and obtains the execution results.

[0068] A drone-assisted porcelain bottle defect offline inspection system, comprising:

[0069] The flight control module is used to control the UAV to perform the porcelain bottle inspection task according to the preset flight route, and adjust the flight route in real time during the flight to cover all preset porcelain bottle positions;

[0070] The multi-sensor data acquisition module is mounted on the drone and is used to collect real-time environmental data of the porcelain bottles and to image and collect data of the porcelain bottles along the route to obtain images and three-dimensional model data of the porcelain bottles;

[0071] The image processing and analysis module is used to perform preliminary processing and analysis on the collected porcelain bottle images and 3D model data, identify the defects of the porcelain bottle, and obtain preliminary analysis results;

[0072] The flight route and inspection strategy dynamic adjustment module is used to analyze the preliminary analysis results and real-time environmental data. When the preliminary analysis results show that the current porcelain bottle is suspected to have defects, the flight route and inspection strategy of the drone are adjusted to focus on re-inspecting the porcelain bottle area with suspected defects; according to the re-inspection analysis results and real-time environmental data, the flight route and inspection strategy of the drone are adjusted, and the inspection is continued according to the new flight route and inspection strategy; when the preliminary analysis results show that the current porcelain bottle is intact, the drone continues to inspect according to the original flight route;

[0073] The defect detection review module is used to conduct a focused review of the suspected defective porcelain bottle area to obtain the review analysis results when the preliminary analysis results show that the current porcelain bottle is suspected to have defects;

[0074] The data synchronization and transmission module is used to transmit all analysis results and flight route adjustment instructions back to the ground control station in real time until the inspection mission is completed.

[0075] Compared with the prior art, the present invention has the following beneficial effects:

[0076] The present invention collects real-time environmental data and performs imaging and data collection on the porcelain bottles along the route when executing a flight route covering all preset porcelain bottle positions through a multi-sensor system, thereby obtaining images and three-dimensional model data of the porcelain bottles. The method uses a deep learning model to perform preliminary processing and analysis on the collected data, identify defects in the porcelain bottles, and dynamically adjust the flight route and inspection strategy when suspected defects are found, and conduct a focused review of suspected defective areas. In addition, the method also includes the function of transmitting all analysis results and flight route adjustment instructions back to the ground control station in real time.

[0077] The present invention can significantly improve the automation and accuracy of inspections. Through real-time data collection and deep learning analysis, the UAV can automatically identify potential defects in porcelain bottles, and quickly adjust the flight route when suspected defects are detected for focused review. This automated adjustment mechanism not only improves the efficiency of inspections, but also enhances the robustness of the system, enabling it to work stably in a changing environment. In addition, by transmitting analysis results and flight route adjustment instructions in real time, the ground control station can promptly understand the working status of the UAV and intervene as needed, further improving the response speed and flexibility of inspections. The combination of these technical features enables the present invention to have significant improvements in improving inspection efficiency and accuracy, and provides strong technical support for the stable operation of the power system.

[0078] The present invention significantly reduces the reliance on manual inspections through an automated drone inspection method, thereby greatly improving the efficiency of porcelain bottle inspections. Using a multi-sensor system and deep learning algorithm, the present invention can more accurately identify minor defects in porcelain bottles, reducing the risk of missed inspections and misjudgments. The data processing unit carried by the drone can process the collected data in real time, quickly identify and respond to potential defects, and improve the timeliness of inspections.

[0079] When a suspected defect is detected, the present invention can intelligently adjust the flight route and inspection strategy to ensure that the suspected defect area is reviewed in detail. The automated inspection process reduces the demand for human resources, allowing inspection resources to be allocated and utilized more efficiently. UAV inspections reduce the risk of manual climbing of porcelain bottles, thereby improving the safety of operators. The present invention can adapt to changing environmental conditions and can work stably even in complex or severe weather.

[0080] All analysis results and flight route adjustment instructions can be transmitted back to the ground control station in real time, ensuring timely update and sharing of information. By analyzing historical inspection data, the present invention helps to predict potential defects of porcelain bottles, thereby achieving predictive maintenance.

[0081] Operators can perform manual intervention based on real-time feedback data, which improves the system's adaptability and flexibility when facing emergencies.

[0082] The UAV-assisted offline inspection method and system for porcelain bottle defects of the present invention, by combining advanced UAV technology, multi-sensor data acquisition, real-time data processing and intelligent decision support, not only improves the efficiency and accuracy of inspection, but also enhances the safety of operations and the robustness of the system, providing a strong technical guarantee for the stable operation of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] Figure 1 It is a flow chart of an embodiment of a method for offline inspection of porcelain bottle defects assisted by a drone according to the present invention.

[0084] Figure 2 It is a module schematic diagram of an embodiment of a drone-assisted offline inspection system for porcelain bottle defects according to the present invention. DETAILED DESCRIPTION

[0085] The present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that the embodiments and features in the embodiments of the present application can be combined with each other without conflict.

[0086] The following detailed description is an exemplary description, which is intended to provide further detailed description of the present invention. Unless otherwise specified, all technical terms used in the present invention have the same meaning as those generally understood by those skilled in the art to which the present application belongs. The terms used in the present invention are only for describing specific embodiments, and are not intended to limit exemplary embodiments according to the present invention.

[0087] Example 1

[0088] like Figure 1 As shown, a drone-assisted offline inspection method for porcelain bottle defects includes:

[0089] Step 1: Pre-flight preparation

[0090] Before the drone takes off, the location data of the porcelain bottles is first imported through the GIS system. These data include the latitude and longitude coordinates of the porcelain bottles and the preset flight route. The flight control system of the drone integrates a route planning algorithm, which can automatically generate the optimal flight path covering all porcelain bottles based on the imported data.

[0091] Step 2: Calibration and testing of the multi-sensor system

[0092] Before the mission, the multi-sensor system on the drone is calibrated to ensure the accuracy and synchronization of the high-resolution camera, infrared camera and LiDAR when collecting data. Test flights are carried out to verify the accuracy and completeness of the sensor data.

[0093] Step 3: Execute the flight mission

[0094] The drone automatically takes off according to the preset route and starts flying along the planned path. During the flight, the multi-sensor system works together, with the high-resolution camera capturing images of the porcelain bottle from different angles, the infrared camera detecting the temperature distribution of the porcelain bottle, and the LiDAR building a 3D model of the porcelain bottle in real time.

[0095] Step 4: Data collection and real-time transmission

[0096] The collected images and 3D model data are transmitted back to the ground control station in real time via the wireless communication module. The data is compressed before transmission to optimize bandwidth usage while ensuring data integrity and accuracy.

[0097] Step 5: Image Preprocessing

[0098] At the ground control station, the received raw image data is first preprocessed, including noise removal, contrast adjustment and image sharpening, to improve the accuracy of subsequent analysis. This step can be achieved through automated image processing software.

[0099] Step 6: Deep Learning Analysis

[0100] The preprocessed image data is fed into a deep learning model for feature extraction and defect identification. The model is pre-trained to identify cracks, stains, and other potential defects on the surface of the porcelain bottle. The analysis results are output in the form of defect probabilities.

[0101] Step 7: Intelligent decision making and strategy adjustment

[0102] When the output of the deep learning model shows that a certain porcelain bottle has a high probability of being defective, the operator at the ground control station will receive an alert and manually confirm whether the drone's flight path needs to be adjusted for a focused re-inspection. If confirmed, the drone will automatically adjust its flight path to scan the suspected defective area in more detail.

[0103] Step 8: Review and confirm defects

[0104] Drones conduct focused re-inspections of suspected defect areas, collecting higher-resolution images and 3D model data, which are used for more in-depth analysis to determine the nature, size and possible impact of the defects.

[0105] Step 9: Generate a test report

[0106] All analysis results and defect details are recorded in the storage device and associated with the location data of the porcelain bottle to generate a complete inspection report. The report includes the defect type, location, size, severity and repair recommendations.

[0107] Step 10: Data synchronization and task completion

[0108] The inspection report is transmitted back to the ground control station in real time through the data synchronization module, and the operator can view the report in real time and make subsequent maintenance decisions. The drone automatically returns to the take-off point after completing the inspection of all preset porcelain bottles.

[0109] The drone first covers all the locations of the porcelain bottles through a preset route and collects data using a multi-sensor system. A high-resolution camera is used to capture detailed images of the porcelain bottles, an infrared camera detects temperature anomalies, and LiDAR generates a three-dimensional model to identify structural anomalies. After the data is transmitted to the onboard processing system, it is denoised and enhanced to improve image quality.

[0110] Deep learning algorithms are applied to the processed image data to automatically extract features and identify potential defects. If a suspected defect is detected, the system will automatically adjust the drone's flight path and inspection strategy to ensure that key areas are reviewed. During the review process, the collected data will be analyzed again to confirm the specific nature and impact of the defect.

[0111] All analysis results and adjustment instructions are transmitted to the ground control station in real time through the data synchronization module. The operator can monitor the status of the drone through the user interface and make manual interventions when necessary. The entire inspection process ensures safety and efficiency while providing detailed inspection reports to support subsequent maintenance decisions.

[0112] Example 2

[0113] The GIS system is used to collect the latitude and longitude coordinates, topographic data, and historical inspection data of the porcelain bottles, and the data is preprocessed to ensure its accuracy and availability.

[0114] Analyze the collected location data, geographic information and historical data, evaluate obstacles, climatic conditions and accessibility of the porcelain vase that may affect the flight of the drone, and determine a safe flight path and necessary flight parameters.

[0115] Using the evaluation results and combined with the UAV's flight performance parameters such as flight speed, endurance and steering angle limit, an optimal flight route covering all preset porcelain bottle positions is automatically planned.

[0116] Through simulation or actual test flights, verify whether the planned optimal flight route meets the safety and efficiency requirements, and ensure that the route can cover all preset porcelain bottle positions without obstacles.

[0117] Based on the verification results, necessary adjustments are made to the flight route, the flight path is optimized, and the final flight route is formed to improve inspection efficiency and safety.

[0118] The determined final flight route is converted into a set of flight instructions executable by the drone, including the take-off point, flight path, shooting point and expected flight speed, to guide the drone to perform inspection tasks.

[0119] First, the precise location data and topographic information of the porcelain bottles are collected through the GIS system and high-resolution maps. Then, this data is analyzed to evaluate possible flight obstacles and climate conditions, such as wind force, wind direction, temperature and humidity, which are important factors affecting the flight of drones. Based on these analyses, professional route planning software is used, combined with the performance parameters of the drone, such as maximum flight speed, endurance time, maximum steering angle, etc., to automatically generate the optimal flight route covering all porcelain bottles.

[0120] Next, the safety and efficiency of the route are verified through simulated flights or test flights in real environments. During the test flight, it is checked whether the drone can fly smoothly over all the preset porcelain bottle positions and does not encounter any obstacles or weather conditions. If any problems are found, such as insufficient flight altitude or excessive turning radius, the route will be adjusted accordingly to ensure that the drone can complete the mission safely and efficiently.

[0121] Finally, the adjusted route is converted into a set of flight instructions that can be executed by the drone, including the precise coordinates of each waypoint, flight speed, and shooting parameters. These instructions will be loaded into the drone's control system to guide it to perform inspection tasks according to the predetermined route. In this way, the drone can automatically and efficiently inspect the porcelain bottles, greatly improving the efficiency and accuracy of the inspection while reducing the risk and cost of manual inspection.

[0122] Example 3

[0123] The multi-sensor system carried by the drone, including high-resolution cameras, infrared cameras and laser radar (LiDAR), collects images and 3D model data of the porcelain bottles in real time while executing a flight route covering all preset porcelain bottle positions. These data are transmitted to the onboard processing system in real time through the wireless communication module on the drone in a specific coding format, such as MPEG-4 or H.264.

[0124] After receiving the raw image data, the onboard processing system first performs denoising to eliminate interference introduced by sensor noise or wireless transmission. The denoising process can use algorithms such as median filtering or Gaussian filtering. Subsequently, image enhancement processing such as histogram equalization or contrast-limited adaptive histogram equalization (CLAHE) is performed to improve the contrast and clarity of the image and prepare high-quality image data for feature extraction.

[0125] A pre-trained deep convolutional neural network (CNN) model was used to extract features from the enhanced images. The model was trained to identify microscopic features and potential defects on the surface of the porcelain bottles, such as cracks, stains, and material defects. The feature extraction process involves segmenting the image into small patches, applying filters to extract texture and shape features, and performing feature learning through a multi-layer network.

[0126] Using the airborne LiDAR data, the 3D model was reconstructed using the PCL (Point Cloud Library). The reconstructed 3D model was analyzed to identify structural anomalies and morphological deviations of the porcelain bottle. This step calculates the position deviation of each point in the point cloud and compares it with the preset model to identify any significant deviations.

[0127] The 2D features extracted by the deep learning algorithm are compared with the 3D model analysis results. The feature vectors are classified by pattern recognition techniques, such as support vector machines (SVM) or random forests, to identify possible defects on the porcelain bottles. The classification process involves training a classifier to distinguish between defective and non-defective samples, and then making predictions on new data.

[0128] The preliminary analysis results of the identification are recorded in an onboard storage device, such as a solid-state drive (SSD), and associated with the corresponding porcelain bottle location information. This is achieved through geographic information system (GIS) tags, where the image and analysis results of each porcelain bottle are associated with its specific location in the power grid.

[0129] The stored preliminary analysis results and the location information of the porcelain bottle are transmitted back to the ground control station in real time through a wireless data link. Data synchronization uses efficient data compression and transmission protocols such as LTE or 5G to ensure data integrity and real-time transmission.

[0130] Example 4

[0131] The image and 3D model data obtained from the preliminary analysis are evaluated in detail through the onboard processing system. Using a set of predefined defect feature libraries, which contain image and model features of various known defects, the collected data is compared with the feature library through pattern matching algorithms, such as support vector machines (SVM) or neural networks, to determine whether there are suspected defects in the porcelain bottle.

[0132] When the evaluation results show suspected defects, the system will automatically record the location, type and appearance characteristics of these defects and generate a suspected defect identification signal. This information will be used for subsequent review and analysis.

[0133] The sensors onboard the drone monitor and record environmental conditions in real time, including wind speed, wind direction, temperature, humidity, etc., while also monitoring the drone’s battery status and remaining power. These data are crucial for adjusting flight routes and inspection strategies.

[0134] Based on the location of the suspected defect and the environmental condition data, the flight control system of the drone integrates a route planning algorithm, such as the A* algorithm or the Dijkstra algorithm, to dynamically adjust the flight route. The adjusted route will guide the drone to approach the suspected defect area for a more detailed review.

[0135] Intelligently set inspection strategies based on the type and location of suspected defects and current environmental conditions. This includes setting the best shooting angle, selecting the appropriate shooting mode, adjusting image resolution, etc. to optimize data collection quality.

[0136] The drone will conduct a focused review of suspected defect areas according to the adjusted flight route and set inspection strategy, using high-resolution cameras and high-precision LiDAR equipment to collect more detailed images and 3D model data.

[0137] The captured high-resolution images and 3D model data are transmitted back to the ground control station for in-depth analysis using more powerful processing capabilities. Using deep learning models, such as convolutional neural networks (CNN), the data is classified and evaluated to determine the nature, size and possible impact of the defects.

[0138] Based on the analysis results, a defect review analysis report is generated, which records in detail the location, type, severity and repair suggestions of each suspected defect.

[0139] The defect review and analysis report is linked to the detailed information and location data of the porcelain bottle to form a complete inspection report. The report contains the defect confirmation information, location, type, severity and repair suggestions.

[0140] The complete inspection report is transmitted back to the ground control station in real time via a wireless data link. The operator can view the report in real time and make maintenance decisions based on the report content.

[0141] Example 5

[0142] Adjust the flight route and inspection strategy of the drone based on the review analysis results and real-time environmental data:

[0143] First, a defective porcelain bottle information library is established through historical inspection data and maintenance records. The information library contains the location, historical defect records and repair status of each porcelain bottle. When the onboard system analyzes the initially collected images and 3D model data through deep learning algorithms, it will automatically compare with this information library to identify and record the specific location and defect characteristics of the defective porcelain bottles.

[0144] Use spatial analysis tools, such as GIS software, to analyze the relative position of defective bottles with other bottles. Based on the physical layout of the bottles and historical defect data, determine which bottles may have similar defect risks, and generate a list of related bottles that need to be checked.

[0145] The drone's sensor suite collects real-time environmental data, including wind speed, wind direction, temperature and humidity. In addition, the drone's battery management system provides real-time information on battery status and remaining power. This data is transmitted to the onboard processing system to assess flight conditions and adjust flight routes.

[0146] Based on the relevant porcelain bottle list and real-time environmental data, the drone's flight control system integrates dynamic programming algorithms, such as the dynamic A* algorithm, to automatically adjust the flight path. The optimized flight path ensures that the drone can cover all porcelain bottles that need to be reviewed in the most economical way, while taking into account environmental constraints and battery life.

[0147] Based on the relevant porcelain bottle list and environmental data, the drone's onboard processing system intelligently adjusts the inspection strategy. For example, for porcelain bottles located in areas with high wind speeds, the drone may adjust the shooting angle to reduce the impact of wind; for porcelain bottles located in areas with extreme temperatures, the shooting frequency may be increased to capture defect changes that may be caused by thermal expansion and contraction.

[0148] The drone follows an optimized flight route and customized inspection strategy to conduct focused re-inspections of suspected defective areas. This may include lowering the flight altitude to obtain higher-resolution images, or performing multiple scans of the suspected defective porcelain bottle area to ensure data integrity.

[0149] After review, the onboard processing system compares the acquired high-resolution images and 3D model data with the previous analysis results to determine the nature, size and possible impact of the defect. The system then generates a detailed inspection report, including the defect confirmation, location, type, severity and repair recommendations. The report is transmitted back to the ground control station in real time via a wireless communication link for the operator to make subsequent maintenance decisions.

[0150] The present invention not only provides an automated drone inspection strategy, but also ensures the flexibility and adaptability of the inspection process, and can effectively identify and evaluate the defects of porcelain bottles under various environmental conditions. This implementation method provides an efficient and reliable solution for power system maintenance, ensuring the stable operation of the power system.

[0151] Example 6

[0152] Based on the images and 3D model data collected by the drone, as well as flight path adjustment instructions, the operator at the ground control station uses professional analysis software to evaluate the drone's inspection path and strategy. This step involves post-processing the drone's flight data to determine if there are areas that require manual adjustment, for example, by evaluating factors such as image clarity and accuracy of defect detection.

[0153] If the assessment results show that manual adjustments are needed, the operator will intervene and adjust the shooting angle, re-plan the flight path, or perform a more detailed scan of the suspected defective porcelain bottle area through the manual control interface of the ground control station. This may include adjusting the drone's flight altitude to obtain better image resolution, or changing the flight direction to cover previously undetected areas.

[0154] When manual intervention is required, the operator at the ground control station sends remote commands to the drone via a high-bandwidth communication link. These commands include but are not limited to changing flight altitude, turning, hovering, or performing specific actions to optimize the inspection process. The sending of remote commands ensures that the drone can dynamically adjust based on real-time operator input.

[0155] After receiving the remote commands, the drone’s flight control system will execute them and begin to implement the new flight route and inspection strategy. During the execution, the drone’s sensors and cameras will collect feedback data, which will be transmitted back to the ground control station in real time so that the operator can evaluate the intervention effect.

[0156] The operator will evaluate the feedback data after the drone executes the remote command to confirm whether the drone has correctly adjusted the flight path and inspection strategy according to the new command. This includes checking the quality of the image and data to ensure that the drone covers all critical areas and performs defect detection as expected.

[0157] If the drone encounters an emergency, such as low battery, interrupted communication link, or encountering severe weather conditions, the ground control station will immediately initiate an emergency plan. The emergency plan may include automatic return, safe landing, or transfer to a preset safe area. The operator will also take necessary on-site emergency measures according to the situation, such as dispatching a rescue team or technical support.

[0158] All manual adjustment decisions, executed remote commands, drone feedback data, and intervention effect evaluation results will be recorded in the ground control station's log system. These records will be used to generate detailed inspection reports, which will include defect confirmation, location, type, severity, and repair recommendations.

[0159] Example 7

[0160] Using the pre-trained deep learning model, the review analysis results are compared with the historical inspection data. This step involves matching and comparing the currently collected porcelain bottle images and 3D model data with the historical data stored in the database to monitor the long-term trend of the porcelain bottle status.

[0161] By analyzing the differences between historical data and current data, the deep learning model is able to predict the potential development of porcelain bottle defects. This step includes using time series analysis or trend prediction algorithms to predict the development trend of defects, thereby providing data support for maintenance decisions.

[0162] Based on the defect prediction results, the next flight route and inspection strategy are optimized. This step involves adjusting the flight path so that the drone can preferentially cover those areas of the porcelain bottles with a higher risk of defects. At the same time, the inspection strategy is adjusted based on the prediction results, such as increasing the inspection frequency of specific porcelain bottles or changing the shooting angle to obtain a clearer image.

[0163] Provide intelligent decision support to help operators develop maintenance plans based on prediction results and optimized inspection strategies. This step includes generating maintenance reports that include defect prediction results, recommended maintenance actions, and expected maintenance schedules.

[0164] The prediction results and optimized inspection strategies are synchronized to the ground control station and a detailed inspection report is generated. This step ensures real-time updates of all relevant data and provides operators with an intuitive interface to monitor inspection progress and results.

[0165] Example 8

[0166] An adaptive image enhancement algorithm, such as adaptive histogram equalization (ACE), is applied to the collected images to improve image contrast, and gridding is performed on the 3D model data to obtain an enhanced image data set and gridded 3D model data.

[0167] A deep convolutional autoencoder is used to learn and extract feature vectors representing defects from the enhanced image dataset. The abnormal areas in the model are identified through the feature matching algorithm of the 3D model data, and the defect feature vector and abnormal area mapping are obtained.

[0168] Combining the defect feature vector of the image data set and the abnormal area mapping of the 3D model data, a multi-dimensional feature fusion technology is used to form a panoramic view of the defect features of the porcelain bottle, and a fused panoramic view of the defect features is obtained.

[0169] Based on the fused panoramic view of defect features, the pre-trained support vector machine (SVM) classifier is used to accurately classify suspected defects and distinguish different types of defects, such as cracks, stains, and material defects, to obtain defect classification results.

[0170] The defect classification results are quantitatively analyzed, and the size of the defects, including length, width and depth, is accurately measured using a geometric measurement algorithm. The area and distribution of the defects are evaluated using image processing technology to obtain quantitative defect parameters.

[0171] Finite element analysis (FEA) is used to simulate the impact of each quantitative defect on the mechanical and electrical properties of the porcelain bottle under operating conditions. The simulation results include stress distribution, electric field strength and potential failure modes, and a simulation analysis report is obtained.

[0172] According to the defect classification results and simulation analysis report, the decision tree algorithm in machine learning is applied to assign a maintenance priority to each defect and make maintenance suggestions, including the maintenance time window and required materials, to obtain maintenance priority recommendations and maintenance suggestions.

[0173] Based on the defect analysis results and repair priority recommendations, a detailed report containing images, 3D models, defect characteristics, classification results, quantitative parameters, simulation analysis reports and repair suggestions is generated to obtain an interactive defect analysis report.

[0174] The interactive defect analysis report is reviewed by an expert system, which provides review opinions and report optimization suggestions based on preset rules and historical data. The report is automatically optimized according to the review opinions to obtain an optimized defect analysis report.

[0175] The optimized defect analysis report is associated with the detailed information and location data of the porcelain bottle, and the report is updated in real time to reflect the latest status and maintenance history of the porcelain bottle. Based on the maintenance priority recommendation and maintenance suggestions, a maintenance work scheduling plan is generated to optimize resource allocation and predict the potential impact of maintenance activities on power grid operation. The scheduling plan is optimized using a genetic algorithm to obtain an optimized maintenance scheduling plan.

[0176] Example 9

[0177] A detailed spatial analysis of the relevant porcelain bottle list obtained from the historical inspection data and porcelain bottle location information was performed using a geographic information system (GIS). The relative positions and distances between the porcelain bottles, as well as their relationship to the current position of the drone, were determined using GIS software to obtain the spatial analysis results. This step ensured that the drone could accurately locate the location of each porcelain bottle and provide basic data for subsequent flight route planning.

[0178] Through integrated sensors and external data sources, detailed data about the drone’s environment is collected in real time, including wind speed, wind direction, temperature, humidity, etc. At the same time, the drone’s battery status and remaining power are monitored to obtain a real-time environmental data set. This data is critical for assessing flight conditions and ensuring drone safety.

[0179] Based on the spatial analysis results and the real-time environmental data set, a path planning algorithm, such as the A* algorithm or the Dijkstra algorithm, is used to generate a preliminary flight route. This route prioritizes covering all relevant porcelain bottles to ensure that the drone can perform inspection tasks efficiently.

[0180] Conduct risk assessment on the generated preliminary flight route, taking into account factors such as terrain obstacles, weather conditions and no-fly zones. Use the risk assessment model to identify potential risk points in the route and obtain a risk assessment report. Based on the risk assessment results, dynamically adjust the preliminary flight route to avoid risk points and obtain an optimized flight route.

[0181] According to the optimized flight route and real-time environmental data, the flight parameters of the drone, such as flight speed, flight altitude and shooting angle, are adjusted to adapt to different flight sections and environmental conditions. These adjustments ensure that the drone can perform inspection tasks safely and efficiently.

[0182] The adjusted flight parameters and optimized flight route are transmitted to the control system of the drone. The drone executes the confirmed flight route, conducts key inspections on the relevant porcelain bottles, collects high-resolution images and 3D model data, and obtains the execution results.

[0183] The feedback data after the drone executes the remote command is transmitted back to the ground control station in real time, and the operator evaluates the intervention effect and confirms whether the drone has correctly adjusted the flight route and inspection strategy according to the new command. This step ensures the accuracy of the inspection mission and the safety of the drone.

[0184] If the drone encounters an emergency or the ground control station observes an abnormal situation, the ground control station will immediately activate the emergency plan. The emergency plan may include automatic return, safe landing or transfer to a preset safe area to ensure the drone's safe return or take necessary risk avoidance measures.

[0185] Example 10

[0186] like Figure 2 As shown, a drone-assisted porcelain bottle defect offline inspection system includes:

[0187] Flight control module, which is implemented through the flight control system of the drone, uses advanced control algorithms such as PID control or adaptive control to ensure that the drone can accurately fly along the preset route. During the flight, this module is also responsible for real-time monitoring of the drone's status, such as position, speed and battery power, and adjusting the flight route as needed to ensure that all preset porcelain bottle positions are covered.

[0188] The multi-sensor data acquisition module, including high-resolution cameras, infrared cameras, and LiDAR sensors, is integrated on the drone to collect real-time environmental data and imaging of the porcelain bottle. The data from these sensors is transmitted back to the ground control station in real time through the wireless communication module, providing raw data for subsequent processing and analysis.

[0189] The image processing and analysis module uses deep learning algorithms to perform preliminary processing and analysis on the collected images and 3D model data. Through a well-trained convolutional neural network (CNN) model, the module can automatically identify defects in porcelain bottles, such as cracks, stains, and material defects, and generate preliminary analysis results.

[0190] The module for dynamically adjusting flight routes and inspection strategies dynamically adjusts the flight routes and inspection strategies of drones based on preliminary analysis results and real-time environmental data. When a suspected defect is detected, the module will re-plan the flight route and guide the drone to conduct a focused review of the suspected defect area. In addition, the module will further optimize the flight route and inspection strategy based on the review results and environmental changes.

[0191] The defect detection review module is activated when the preliminary analysis results show that the porcelain bottle is suspected of having defects, and conducts a more detailed review of the suspected defective area. Using high-resolution imaging and advanced image processing technology, this module can accurately identify the nature, size and possible impact of the defect and generate a review analysis result.

[0192] The data synchronization and transmission module is responsible for transmitting all analysis results and flight path adjustment instructions back to the ground control station in real time. Through high-speed wireless data links such as 4G or 5G networks, this module ensures real-time updating and accurate transmission of data, providing operators with timely information for decision-making and intervention.

[0193] The drone-assisted porcelain bottle defect offline inspection system of the present invention can perform inspection tasks efficiently and accurately, while ensuring the flexibility and adaptability of the system and meeting the needs of practical applications.

[0194] It is known from common technical knowledge that the present invention can be implemented by other embodiments that do not deviate from its spirit or essential features. Therefore, the above disclosed embodiments are only illustrative in all respects and are not exclusive. All changes within the scope of the present invention or within the scope equivalent to the present invention are included in the present invention.

Claims

1. A drone-assisted offline inspection method for porcelain bottle defects, characterized in that: include, The drone uses a multi-sensor system to collect real-time environmental data and image and collect data on the porcelain bottles along the route while executing a flight route covering all preset porcelain bottle locations, obtaining images and three-dimensional model data of the porcelain bottles. Based on the image and 3D model data of the porcelain bottle, preliminary processing and analysis are performed to identify the defects of the porcelain bottle and obtain preliminary analysis results; Analyze the preliminary analysis results and real-time environmental data. When the preliminary analysis results show that the current porcelain bottle is suspected to have defects, adjust the flight route and inspection strategy of the drone, conduct a focused review of the porcelain bottle area suspected of defects, and obtain the review analysis results; adjust the flight route and inspection strategy of the drone based on the review analysis results and real-time environmental data, and continue the inspection according to the new flight route and inspection strategy; When the preliminary analysis results show that the current porcelain bottle is intact, the drone continues the inspection along the original flight route; All analysis results and flight route adjustment instructions are transmitted back to the ground control station in real time until the inspection is completed.

2. The method for offline inspection of porcelain bottles with the aid of drones according to claim 1 is characterized in that: The step of setting the flight route covering all preset porcelain bottle positions specifically includes: Collect the location data and related geographic information of the porcelain bottles, including but not limited to the latitude and longitude coordinates of the porcelain bottles, topographic data and historical data of previous inspections; Analyze the location data, geographic information and historical data collected in the step to assess obstacles, climatic conditions and accessibility of the porcelain vase that may affect the flight of the drone, and determine a safe flight path and necessary flight parameters; Using the evaluation results and the flight performance parameters of the drone, including flight speed, endurance, and steering angle limits, an optimal flight route covering all preset porcelain bottle locations is automatically planned; Through simulation or actual test flights, verify whether the planned optimal flight route meets the safety and efficiency requirements, and ensure that the route can cover all preset porcelain bottle positions without obstacles; According to the verification results, make necessary adjustments to the flight route, optimize the flight path, and form the final flight route; The determined final flight route is converted into a set of flight instructions executable by the drone, including the take-off point, flight path, shooting point and expected flight speed, to guide the drone to perform inspection tasks.

3. The method for offline inspection of porcelain bottles with the aid of drones according to claim 1 is characterized in that: Based on the obtained image and three-dimensional model data of the porcelain bottle, preliminary processing and analysis are performed to identify the defects of the porcelain bottle and obtain the preliminary analysis results, specifically including: The porcelain bottle images and 3D model data collected by the drone are transmitted to the onboard processing system in real time; De-noising and enhancing the transmitted image data to obtain optimized image data; Use deep learning algorithms to extract features from optimized images and identify key features and potential defect areas on the surface of the porcelain bottle; Combined with the extracted features, the 3D model data is analyzed to identify structural anomalies and morphological deviations of the porcelain bottle; Comprehensively compare the analysis results, identify possible defects, and form preliminary analysis results; The preliminary analysis results of the identification are recorded in the storage device and associated with the corresponding porcelain bottle location information for subsequent review and analysis.

4. The method for offline inspection of porcelain bottles with the aid of drones according to claim 1 is characterized in that: Analyze the preliminary analysis results and real-time environmental data. When the preliminary analysis results show that the current porcelain bottle is suspected to have defects, adjust the flight route and inspection strategy of the drone, and conduct a focused review of the porcelain bottle area suspected of defects. The steps to obtain the review analysis results include: Based on the preliminary analysis results, use the preset defect recognition standards to determine whether the porcelain bottle has suspected defects, record the location and type of the suspected defects, and obtain the suspected defect identification signal; Real-time synchronization of the current flight environment conditions of the drone, including but not limited to wind speed, wind direction, temperature, humidity, and the drone's own battery status and remaining power, to obtain environmental status data; Based on suspected defect identification signals and environmental status data, the flight route of the drone is dynamically adjusted to ensure that the drone can safely and efficiently approach and review suspected defect areas, and generate an optimized flight route; According to the location, type and environmental status data of suspected defects, the inspection strategy of the drone is intelligently set, including shooting angle, shooting mode, and image resolution parameters, to form a customized inspection strategy for suspected defect areas; According to the optimized flight route and customized inspection strategy, the drone performs a focused review of suspected defect areas, collects high-resolution images and 3D model data, and obtains a focused review data set; Conduct in-depth analysis on the key review data sets obtained, use deep learning models to accurately classify and evaluate suspected defects, determine the nature, size and possible impact of the defects, and output defect review analysis reports; Organize the defect review and analysis report, associate it with the detailed information and location data of the porcelain bottle, generate a complete inspection report, including the confirmation, location, type, severity and repair suggestions of the defect, and transmit the report to the ground control station.

5. The method for offline inspection of porcelain bottles with the aid of drones according to claim 1 is characterized in that: According to the review and analysis results and real-time environmental data, the steps to adjust the flight route and inspection strategy of the drone include: According to the review and analysis results, the defective porcelain bottles are identified, and the specific locations and defect characteristics of the defective porcelain bottles are recorded to obtain the defective porcelain bottle information; Based on the defective porcelain bottle information, the relative position relationship and correlation between the defective porcelain bottle and other porcelain bottles are analyzed, and other related or similar porcelain bottles that need to be checked are determined to obtain a list of related porcelain bottles; Collect detailed data of the drone's current environment in real time, including but not limited to wind speed, wind direction, temperature, and humidity, to obtain a real-time environmental data set; According to the relevant porcelain bottle list and real-time environmental data set, the flight route of the drone is dynamically adjusted to optimize the flight path, giving priority to covering all relevant porcelain bottles to obtain the optimized flight route; According to the list of relevant porcelain bottles, the inspection strategy is intelligently adjusted, including shooting parameters, review frequency and key focus areas, to form a customized inspection strategy for relevant porcelain bottles; The drone performs inspections according to the optimized flight route and customized inspection strategies for the relevant porcelain bottles, giving priority to re-inspecting the relevant porcelain bottles and obtaining adjusted inspection results; The adjusted inspection results are recorded in a storage device and evaluated to determine whether related porcelain bottles have similar defects, and an inspection result report is obtained and fed back to the ground control station in real time.

6. The method for offline inspection of porcelain bottles with the aid of drones according to claim 1 is characterized in that: The following steps are also included: Based on the analysis results and flight path adjustment instructions of the drone, evaluate the drone's inspection path and strategy to determine whether manual adjustments are needed, including adjusting the shooting angle, replanning the flight path, or performing a more detailed scan of the porcelain bottle area suspected of defects; When manual intervention is determined to be necessary, remote commands are sent to the drone through the ground control station, including but not limited to changing the flight altitude, turning, hovering, or performing specific actions to optimize the detection process; Feedback data after the drone executes remote commands is transmitted back to the ground control station in real time, and the operator evaluates the intervention effect and confirms whether the drone has correctly adjusted its flight route and inspection strategy according to the new instructions; If the drone encounters an emergency or the ground control station observes an abnormal situation, the ground control station will immediately activate the emergency plan to ensure the drone's safe return or take necessary risk avoidance measures.

7. The method for offline inspection of porcelain bottles with the aid of drones according to claim 1 is characterized in that: The process also includes the following steps: comparing the review and analysis results with the historical inspection data, monitoring the changing trend of the porcelain bottle status through a pre-trained deep learning model, predicting the development of potential defects, and obtaining defect prediction results. Based on the defect prediction results, the next flight route and inspection strategy are formulated.

8. The method for offline inspection of porcelain bottles with the aid of drones according to claim 4 is characterized in that: The obtained key review data set is analyzed in depth, and the suspected defects are accurately classified and evaluated using a deep learning model to determine the nature, size, and possible impact of the defects. The steps to output a defect review analysis report include: Applying an adaptive image enhancement algorithm to the collected images to improve the contrast, and performing gridding processing on the 3D model data to obtain an enhanced image data set and gridded 3D model data; A deep convolutional autoencoder is used to learn and extract feature vectors representing defects from the enhanced image dataset, and the abnormal areas in the model are identified through a feature matching algorithm of the 3D model data to obtain defect feature vectors and abnormal area mapping. Combining the defect feature vector of the image data set and the abnormal area mapping of the 3D model data, a multi-dimensional feature fusion technology is used to form a panoramic view of the defect features of the porcelain bottle, and a fused panoramic view of the defect features is obtained; Based on the fused panoramic view of defect features, the pre-trained support vector machine classifier is used to accurately classify suspected defects, distinguish different types of defects, including cracks, stains, and material defects, and obtain defect classification results; Quantitatively analyze the defect classification results, use geometric measurement algorithms to accurately measure the size of the defects, including length, width and depth, and use image processing technology to evaluate the area and distribution of the defects to obtain quantitative defect parameters; Finite element analysis is used to simulate the impact of each quantitative defect on the mechanical and electrical properties of the porcelain bottle under operating conditions. The simulation results include stress distribution, electric field strength and potential failure modes, and a simulation analysis report is obtained; Intelligent recommendation of maintenance priority: Based on the defect classification results and simulation analysis reports, the decision tree algorithm in machine learning is applied to assign maintenance priority to each defect and make maintenance suggestions, including maintenance time window and required materials, to obtain maintenance priority recommendations and maintenance suggestions; Generate a detailed report containing images, 3D models, defect features, classification results, quantitative parameters, simulation analysis reports and repair suggestions based on defect analysis results and repair priority recommendations to obtain an interactive defect analysis report; The interactive defect analysis report is reviewed by the expert system. The expert system provides review opinions and report optimization suggestions based on preset rules and historical data. The report is automatically optimized according to the review opinions to obtain the optimized defect analysis report. The optimized defect analysis report is associated with the detailed information and location data of the porcelain bottle, and the report is updated in real time to reflect the latest status and maintenance history of the porcelain bottle, and obtain a real-time updated associated report; Based on the maintenance priority recommendations and maintenance suggestions, a maintenance work scheduling plan is generated to optimize resource allocation and predict the potential impact of maintenance activities on power grid operation. The scheduling plan is optimized using a genetic algorithm to obtain an optimized maintenance scheduling plan.

9. The method for offline inspection of porcelain bottles with the aid of drones according to claim 5 is characterized in that: According to the relevant porcelain bottle list and real-time environmental data set, the flight route of the drone is dynamically adjusted to optimize the flight path, giving priority to covering all relevant porcelain bottles, and the steps of obtaining the optimized flight route include: Use geographic information system to perform spatial analysis on the obtained list of related porcelain bottles, determine the relative positions and distances between the porcelain bottles, and the relationship between the related porcelain bottles and the current position of the drone, and obtain the spatial analysis results; Through integrated sensors and external data sources, detailed data about the drone’s environment is collected in real time, including wind speed, wind direction, temperature, humidity, as well as the drone’s battery status and remaining power, to obtain a real-time environmental data set. Based on the spatial analysis results and the real-time environmental data set, a path planning algorithm is used to first cover all relevant porcelain bottles and then inspect the remaining porcelain bottles to obtain a preliminary flight route; Conduct risk assessment on the generated preliminary flight route, taking into account factors such as terrain obstacles, weather conditions and no-fly zones, identify potential risk points in the route, and obtain a risk assessment report; According to the risk assessment report, the preliminary flight route is dynamically adjusted to avoid risk points and obtain an optimized flight route; According to the optimized flight route and environmental data, the flight parameters of the UAV are adjusted to adapt to different flight sections and environmental conditions to obtain adjusted flight parameters; The adjusted flight parameters and optimized flight route are transmitted to the control system of the UAV. The UAV executes the confirmed flight route, conducts key inspections on the relevant porcelain bottles, and obtains the execution results.

10. A drone-assisted porcelain bottle defect offline inspection system, based on a drone-assisted porcelain bottle defect offline inspection method according to any one of claims 1 to 9, characterized in that: include, The flight control module is used to control the UAV to perform the porcelain bottle inspection task according to the preset flight route, and adjust the flight route in real time during the flight to cover all preset porcelain bottle positions; The multi-sensor data acquisition module is mounted on the drone and is used to collect real-time environmental data of the porcelain bottles and to image and collect data of the porcelain bottles along the route to obtain images and three-dimensional model data of the porcelain bottles; The image processing and analysis module is used to perform preliminary processing and analysis on the collected porcelain bottle images and 3D model data, identify the defects of the porcelain bottle, and obtain preliminary analysis results; The flight route and inspection strategy dynamic adjustment module is used to analyze the preliminary analysis results and real-time environmental data. When the preliminary analysis results show that the current porcelain bottle is suspected to have defects, the flight route and inspection strategy of the drone are adjusted to conduct a focused review of the porcelain bottle area with suspected defects; according to the review analysis results and real-time environmental data, the flight route and inspection strategy of the drone are adjusted, and the inspection is continued according to the new flight route and inspection strategy; When the preliminary analysis results show that the current porcelain bottle is intact, the drone continues the inspection along the original flight route; The defect detection review module is used to conduct a focused review of the suspected defective porcelain bottle area to obtain the review analysis results when the preliminary analysis results show that the current porcelain bottle is suspected to have defects; The data synchronization and transmission module is used to transmit all analysis results and flight route adjustment instructions back to the ground control station in real time until the inspection mission is completed.

Images