Substation unmanned aerial vehicle intelligent inspection data management method and system, and medium

Abstract

The invention belongs to the technical field of intelligent inspection, and particularly relates to a transformer substation unmanned aerial vehicle intelligent inspection data management method and system and a medium. The method comprises the following steps: in an unmanned aerial vehicle inspection stage, acquiring an inspection image shot by an unmanned aerial vehicle according to inspection point position parameters, and binding the inspection image, the corresponding inspection point position parameters, a substation equipment ID and actual shooting information to serve as a structured data packet of the inspection image, so that inspection data is stored in the form of the structured data packet; the inspection point position parameters comprise pose parameters during unmanned aerial vehicle shooting and camera shooting parameters. According to the invention, the technical problem in the prior art that the affiliated equipment is difficult to accurately and quickly determine according to the image data is solved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of intelligent inspection, and particularly relates to a substation unmanned aerial vehicle intelligent inspection data management method and system and a medium. BACKGROUND

[0002] With the in-depth promotion of smart grid construction, the intelligent operation and maintenance level of substations has become the key to ensuring the safe and stable operation of the power system. The traditional manual inspection method has many drawbacks such as low efficiency, high labor intensity, high safety risk, and great influence by subjective factors, and has been difficult to meet the demand of modern power system for lean management of operation and maintenance. In recent years, unmanned aerial vehicle technology has been widely used in substation equipment inspection work due to its high flexibility, wide coverage, and relatively low cost, effectively solving some of the problems of manual inspection. By carrying high-definition visible light, infrared thermal imaging, ultraviolet imaging and other sensors, unmanned aerial vehicles can efficiently collect massive device state image data, greatly improving the efficiency and safety of inspection operations.

[0003] However, the popularization and application of unmanned aerial vehicle inspection technology also derives new technical problems, the core of which is the low efficiency of inspection data management and utilization. First, the image data generated by the current unmanned aerial vehicle inspection is usually named and stored only with a timestamp or a simple serial number, resulting in that massive data cannot be automatically and accurately associated with specific physical devices in the substation (such as "#1 main transformer 110kV side B phase bushing"). The operation and maintenance personnel must spend a lot of time for manual screening, identification and classification, which not only is extremely tedious and prone to errors, but also seriously restricts the generation efficiency of inspection reports, making the "efficiency" of unmanned aerial vehicle inspection stop at the data collection link. Secondly, due to the weak association between data and equipment, when a device defect is found in a picture through manual or AI (Artificial Intelligence) image recognition, it is difficult to quickly and accurately trace and review the historical inspection data of the same device, and it is impossible to effectively compare the trend and conduct in-depth state analysis, the intrinsic value of the data has not been fully tapped, and the development of predictive maintenance is limited. In addition, these unstructured data are stored in the form of loose files, forming a "data island", which is difficult to deeply integrate and intelligently analyze with the existing Production Management System (PMS) and equipment asset management system, hindering the process of digital transformation of substation operation and maintenance.

[0004] At present, in the field of unmanned aerial vehicle intelligent inspection of substations, the common data management scheme mainly relies on the following two technical routes: 1. Manual Labeling and File Directory Management: This is currently the most widely used but also the most primitive method. The specific process is as follows: After the drone completes its inspection mission, maintenance personnel manually copy the massive amounts of image data (photos and videos) collected to a computer. Then, operators need to rely on their experience to examine each image or video clip, visually identifying the equipment content and manually renaming it (e.g., "20231027_Main Transformer_Bushing.jpg") or categorizing it into pre-established folders (e.g., creating folders like "Main Transformer Area" and "Circuit Breaker Area," then manually dragging and dropping to categorize). Finally, combining the inspectors' on-site records or memory, the organized image data is associated with the equipment assets in the production management system. This method heavily relies on manual labor, is extremely inefficient, and is prone to misclassification when equipment appearances are similar and the data volume is huge, leading to chaotic data-equipment correspondences and causing significant difficulties for subsequent data queries and historical comparisons.

[0005] 2. A rough matching method based on GPS coordinates and equipment ledgers: This is a relatively improved, but still significantly flawed, semi-automated method. During drone flight, the system records the GPS coordinates of each photo taken. In the background processing, the system pre-enters a rough GPS location ledger of the substation's main equipment. By calculating the proximity between the photo's GPS and the equipment ledger's GPS, the system automatically matches the photo to the nearest device, achieving a degree of automatic classification. However, this method has a fatal flaw: severely insufficient accuracy. First, civilian GPS devices inherently have errors of several meters or even tens of meters, making it impossible to accurately distinguish adjacent devices (e.g., distinguishing adjacent A-phase, B-phase, and C-phase devices) in densely populated substations. Second, this method cannot distinguish different observation surfaces of the same device (e.g., although the transformer's oil conservator and bushing belong to the same main transformer, their observation points and diagnostic values ​​are completely different). Therefore, this matching result is very coarse, with a high error rate, still requiring significant manual intervention for secondary verification, and cannot achieve true automated traceability.

[0006] In summary, existing intelligent inspection methods using drones for substations suffer from technical challenges in accurately and quickly identifying the equipment involved based on image data. Summary of the Invention

[0007] The purpose of this invention is to provide a method, system, and medium for intelligent inspection data management of substations using unmanned aerial vehicles (UAVs), in order to solve the technical problem that existing technologies cannot accurately and quickly determine the equipment to which it belongs based on image data.

[0008] To solve the above-mentioned technical problems, the present invention provides a technical solution for a substation drone intelligent inspection data management method, which includes: a substation drone intelligent inspection data management method, comprising: During the drone inspection phase, inspection images taken by the drone according to the inspection point parameters are acquired. The inspection images and their corresponding inspection point parameters, substation equipment IDs and actual shooting information are bound together as a structured data packet for the inspection image, so that the inspection data is stored in the form of a structured data packet. The inspection point parameters include the pose parameters during drone shooting and the camera shooting parameters.

[0009] The beneficial effects of the above technical solution are as follows: This invention pre-defines one or more UAV inspection points containing complete shooting parameters (spatial coordinates, gimbal angle, camera parameters) for each device, thereby establishing a precise mapping relationship of "device ↔ inspection image". This positioning method goes beyond simple GPS coordinates, achieving precise control over the device's observation angle. This invention solves the technical problem of existing technologies' difficulty in accurately and quickly determining the device based on image data.

[0010] Furthermore, the method also includes: in the image recognition stage, binding the recognition result of the inspection image with the structured data packet of the inspection image obtained in the UAV inspection stage as a new structured data packet.

[0011] Furthermore, the structured data packets during the UAV inspection phase are obtained as follows: when the UAV captures inspection images, the inspection point parameters and actual shooting information are written as metadata into the EXIF ​​information of the inspection images to obtain the structured data packets of the inspection images.

[0012] Furthermore, the actual shooting information includes the shooting time; the method also includes: storing the structured data packet in a database; the database type is a time-series database or a relational database; the database index includes the substation equipment ID and the shooting time.

[0013] Furthermore, the inspection point parameters are obtained in the following manner: (1) Construct a three-dimensional model of the target substation; the three-dimensional model can realistically reflect the spatial location, geometry and appearance texture of each piece of equipment in the target substation; (2) In the three-dimensional model, according to the preset inspection procedure, at least one inspection point and inspection point parameters are set for each substation equipment that needs to be inspected, so that the drone can achieve the best shooting effect when shooting according to the inspection point parameters.

[0014] Furthermore, the three-dimensional model is obtained through UAV oblique photography, laser scanning, or BIM modeling.

[0015] Furthermore, the camera shooting parameters include the camera focal length.

[0016] Furthermore, the substation equipment ID is consistent with the equipment asset code in the substation production management system.

[0017] This invention also provides a technical solution for a substation drone intelligent inspection data management system: a substation drone intelligent inspection data management system, including a processor, the processor being used to execute a computer program to implement the steps of the substation drone intelligent inspection data management method described below: During the drone inspection phase, inspection images taken by the drone according to the inspection point parameters are acquired. The inspection images and their corresponding inspection point parameters, substation equipment IDs and actual shooting information are bound together as a structured data packet for the inspection image, so that the inspection data is stored in the form of a structured data packet. The inspection point parameters include the pose parameters during drone shooting and the camera shooting parameters.

[0018] The beneficial effects of the above technical solution are as follows: The substation UAV intelligent inspection data management system of the present invention pre-sets one or more UAV inspection points containing complete shooting parameters (spatial coordinates, gimbal angle, camera parameters) for each device, thereby establishing a precise mapping relationship of "device ↔ inspection image". This positioning method goes beyond simple GPS coordinates and achieves refined control over the observation angle of the device. The present invention solves the technical problem of existing technologies that make it difficult to accurately and quickly determine the device based on image data.

[0019] Furthermore, the method also includes: in the image recognition stage, binding the recognition result of the inspection image with the structured data packet of the inspection image obtained in the UAV inspection stage as a new structured data packet.

[0020] Furthermore, the structured data packets during the UAV inspection phase are obtained as follows: when the UAV captures inspection images, the inspection point parameters and actual shooting information are written as metadata into the EXIF ​​information of the inspection images to obtain the structured data packets of the inspection images.

[0021] Furthermore, the actual shooting information includes the shooting time; the method also includes: storing the structured data packet in a database; the database type is a time-series database or a relational database; the database index includes the substation equipment ID and the shooting time.

[0022] Furthermore, the inspection point parameters are obtained in the following manner: (1) Construct a three-dimensional model of the target substation; the three-dimensional model can realistically reflect the spatial location, geometry and appearance texture of each piece of equipment in the target substation; (2) In the three-dimensional model, according to the preset inspection procedure, at least one inspection point and inspection point parameters are set for each substation equipment that needs to be inspected, so that the drone can achieve the best shooting effect when shooting according to the inspection point parameters.

[0023] Furthermore, the three-dimensional model is obtained through UAV oblique photography, laser scanning, or BIM modeling.

[0024] Furthermore, the camera shooting parameters include the camera focal length.

[0025] Furthermore, the substation equipment ID is consistent with the equipment asset code in the substation production management system.

[0026] The present invention also provides a technical solution for a computer-readable storage medium: a computer-readable storage medium having a computer program stored internally, the computer program being executed by a processor to implement the steps of the substation unmanned aerial vehicle (UAV) intelligent inspection data management method described below: During the drone inspection phase, inspection images taken by the drone according to the inspection point parameters are acquired. The inspection images and their corresponding inspection point parameters, substation equipment IDs and actual shooting information are bound together as a structured data packet for the inspection image, so that the inspection data is stored in the form of a structured data packet. The inspection point parameters include the pose parameters during drone shooting and the camera shooting parameters.

[0027] The beneficial effects of the above technical solution are as follows: The computer program-implemented intelligent inspection data management method for substations using unmanned aerial vehicles (UAVs) of this invention presets one or more UAV inspection points for each device, containing complete shooting parameters (spatial coordinates, gimbal angle, camera parameters), thereby establishing a precise mapping relationship of "device ↔ inspection image". This positioning method goes beyond simple GPS coordinates, achieving precise control over the observation angle of the device. This invention solves the technical problem of existing technologies' difficulty in accurately and quickly determining the device based on image data.

[0028] Furthermore, the method also includes: in the image recognition stage, binding the recognition result of the inspection image with the structured data packet of the inspection image obtained in the UAV inspection stage as a new structured data packet.

[0029] Furthermore, the structured data packets during the UAV inspection phase are obtained as follows: when the UAV captures inspection images, the inspection point parameters and actual shooting information are written as metadata into the EXIF ​​information of the inspection images to obtain the structured data packets of the inspection images.

[0030] Furthermore, the actual shooting information includes the shooting time; the method also includes: storing the structured data packet in a database; the database type is a time-series database or a relational database; the database index includes the substation equipment ID and the shooting time.

[0031] Furthermore, the inspection point parameters are obtained in the following manner: (1) Construct a three-dimensional model of the target substation; the three-dimensional model can realistically reflect the spatial location, geometry and appearance texture of each piece of equipment in the target substation; (2) In the three-dimensional model, according to the preset inspection procedure, at least one inspection point and inspection point parameters are set for each substation equipment that needs to be inspected, so that the drone can achieve the best shooting effect when shooting according to the inspection point parameters.

[0032] Furthermore, the three-dimensional model is obtained through UAV oblique photography, laser scanning, or BIM modeling.

[0033] Furthermore, the camera shooting parameters include the camera focal length.

[0034] Furthermore, the substation equipment ID is consistent with the equipment asset code in the substation production management system. Attached Figure Description

[0035] Figure 1 This is an implementation method of the substation drone intelligent inspection data management system of the present invention. Detailed Implementation

[0036] This invention pre-defines one or more drone inspection points for each device, each containing complete shooting parameters (spatial coordinates, gimbal angle, camera parameters), thereby establishing a precise mapping relationship of "device ↔ inspection image". This positioning method goes beyond simple GPS coordinates, enabling precise control over the device's observation angle. This invention solves the technical problem of existing technologies' difficulty in accurately and quickly determining the device's location based on image data.

[0037] Implementation method of intelligent inspection data management for substations using drones: According to the background description, the core problem with existing technical solutions lies in their failure to establish a high-precision, automated mechanism to uniquely and accurately bind inspection data to specific equipment assets within the substation at the source of data generation. They either rely entirely on inefficient and error-prone manual processing or on insufficiently accurate spatial matching, leading to the following unavoidable defects: Low level of automation: requires a lot of human intervention, resulting in low work efficiency and high costs.

[0038] Poor data association accuracy: prone to incorrect matching, resulting in low data reliability.

[0039] Weak traceability: It is difficult to quickly and accurately query the full life cycle inspection data of specific equipment, and historical comparison and trend analysis are difficult.

[0040] Unable to support refined operation and maintenance: Due to the inability to distinguish specific components of equipment and observation angles, the precision and availability of data are greatly reduced, making it difficult to meet the deep needs of intelligent operation and maintenance and status assessment.

[0041] These shortcomings severely restrict the full realization of the effectiveness of UAV inspection technology. Therefore, this invention aims to specifically solve the following three levels of technical problems: The accuracy and automation of data-equipment association: Current technology cannot automatically and accurately bind each inspection image to a specific physical device (or even a specific component of the device) in the substation at the source of data collection. Relying on manual screening or crude GPS matching leads to low efficiency, high error rate, and an inability to handle the massive amounts of data generated by large-scale inspections. This is the primary bottleneck restricting the improvement of inspection efficiency.

[0042] Issues with historical data tracing and trend analysis of inspection data: Due to inaccurate data correlation, when equipment defects or abnormal conditions are discovered, maintenance personnel find it difficult to quickly and accurately access complete historical inspection images and data records for the equipment. The lack of efficient historical data tracing methods makes time-series-based equipment status trend analysis and defect evolution tracking extremely difficult, severely hindering the implementation of predictive maintenance models.

[0043] The problems of data management structuring and system integration: Existing solutions generate mostly unstructured, scattered files, lacking a unified data structure and indexing mechanism that computers can efficiently recognize. This leads to inspection data being isolated from production management systems, equipment asset management systems, etc., forming "data silos," making it difficult to carry out deep data fusion, mining, and intelligent applications, thus limiting the overall improvement of the digital operation and maintenance level of substations.

[0044] The substation drone intelligent inspection data management method of this embodiment is as follows: Figure 1 As shown, a closed-loop management system was constructed that spans the entire process of data generation, association, storage, and application, including the following steps: S1. Constructing a digital twin mapping relationship (equivalent to...) Figure 1 Phase 1: Digital Mapping Construction.

[0045] This step is fundamental to all subsequent operations, aiming to establish a computable mapping in the digital space that fully corresponds to the physical substation. Specifically, it includes: Model Acquisition: First, a high-precision 3D real-world model of the target substation is acquired using existing technologies such as UAV oblique photography, LiDAR scanning, or existing BIM models. This model must accurately reflect the spatial location, geometry, and appearance texture of the equipment within the substation.

[0046] Equipment Digital Definition: In the aforementioned high-precision 3D reality model, each equipment unit requiring inspection (such as circuit breakers, disconnectors, transformer bushings, surge arresters, etc.) is digitally defined manually or with the assistance of image recognition algorithms. A globally unique equipment identification code (i.e., equipment ID) is created for each equipment unit. It is recommended that this ID be consistent with the equipment asset code in the Production Management System (PMS) to achieve information interoperability.

[0047] Pre-set inspection points: For each equipment unit, one or more UAV inspection points are preset in three-dimensional space according to its inspection procedures (such as the angle and location to be observed). Each inspection point is not a simple coordinate, but a set of parameters including spatial coordinates (X, Y, Z), gimbal pitch angle, gimbal yaw angle, and camera zoom. This set of points constitutes the "digital script" for automated UAV operations.

[0048] S2, Generate inspection tasks and bind them to data collection (equivalent to...) Figure 1 Phase Two: Task Execution and Data Acquisition.

[0049] This step enables the automated and standardized collection of data from digital instructions to the physical world.

[0050] Task Planning: Users initiate inspection tasks by selecting an area in the 3D model or by directly selecting the device ID. The task planning system automatically calculates and generates the optimal flight path and shooting sequence for the UAV based on the preset inspection point parameters in S1, ensuring safe and efficient flight paths.

[0051] Autonomous Flight and Data Acquisition: The UAV is equipped with a flight control system. After receiving mission commands, it takes off autonomously and flies to each preset inspection point. Upon arrival, the flight control system automatically controls the gimbal and camera, precisely adjusting the preset parameters for that point (e.g., rotating the gimbal to -30° to photograph the transformer oil level gauge), and then takes pictures.

[0052] Key: Real-time binding of data and device ID: At the moment the camera shutter is triggered, the flight control system automatically writes key information such as the device identification code (ID) associated with the current inspection point, high-precision timestamp (UTC time in this implementation, indicating the image capture time), BeiDou / GPS coordinates, and UAV attitude angle as metadata directly into the EXIF ​​information of the collected raw image data, or binds the metadata with the image data stream via a 5G / 4G data transmission radio and transmits it back to the ground station together. This step ensures from the source that every piece of data "comes with its own ID".

[0053] S3, Intelligent Data Processing and Structured Encapsulation (equivalent to...) Figure 1 Phase 3: Data Processing and Storage.

[0054] This step performs value-added processing on the collected raw data and transforms it into structured data that is easy for computers to manage and analyze.

[0055] AI-powered intelligent recognition: After completing its mission, the drone transmits image data, bound with metadata, back to the data center server. The server then uses a pre-trained AI image recognition algorithm (such as a convolutional neural network (CNN) based on deep learning) to automatically analyze the image data, identifying equipment status, instrument readings, temperature values ​​(infrared thermal imaging analysis), and potential defects (such as rust, damage, and overheating).

[0056] Generate a structured data packet: After AI recognition, the recognition result (e.g., {"Device ID": "PT-001", "Recognition Type": "Oil Temperature Reading", "Result": "67.5℃", "Status": "Normal"}) is used as new high-level metadata and is then bound again to the original data (image + basic metadata) generated in S2 to form a complete structured data packet containing multi-level information.

[0057] Indexed storage: The structured data packets described above are stored in a time-series database or relational database, rather than a traditional file system. Within the database, a composite index is created using the device identification code (ID) and timestamp as the core key fields, while secondary indexes can be created for fields such as device type, defect level, and status. This storage method significantly improves the query speed for massive amounts of data.

[0058] S4, Two-way Visual Traceability and Intelligent Applications (equivalent to...) Figure 1 Phase Four: Data Source Tracing and Application.

[0059] This step demonstrates the application value of the method ultimately achieved by the present invention, providing users with powerful data interaction and insight capabilities.

[0060] Forward tracing (from data to device): When a user finds a "damaged insulator" record in the list of defects identified by AI, they only need to click on the record, and the system can immediately read the device ID it is bound to (such as "IL-202"). Then, it automatically highlights and focuses on the device in the 3D visualization interface, clearly showing its spatial position and panoramic shape, realizing second-level positioning from data to physical location.

[0061] Reverse tracing (from device to data): When a user clicks on any device in the 3D visualization interface (e.g., clicking on "#1 main transformer" in the 3D model), the system automatically retrieves its device ID and initiates a query request to the database. The system immediately returns inspection images, raw data, and AI analysis results reports for all historical cycles of that device, organized in a timeline format. Users can easily compare data from different periods to analyze trends in device status changes.

[0062] Advanced Applications: Based on structured data, the system can automatically generate equipment health status reports and early warning information, and push them to relevant production management systems to drive the automation and intelligence of operation and maintenance decisions.

[0063] By organically combining the above four steps, this invention forms a complete closed loop from digital mapping to physical acquisition, and then to intelligent analysis and visual traceability.

[0064] Implementation method of substation drone intelligent inspection data management system: A substation drone intelligent inspection data management system includes a processor for executing computer programs to implement the steps of the substation drone intelligent inspection data management method described above. Specific implementation methods for the substation drone intelligent inspection data management method can be found in the above-described embodiments.

[0065] Based on the implementation method described in the embodiment of the intelligent inspection data management method for substation drones, this embodiment can use the following software modules to implement the above method: Digital mapping module: Used to load 3D models and provide an interface for users to define device IDs and inspection points.

[0066] Specifically, model acquisition involves first obtaining a high-precision 3D real-world model of the target substation using existing technologies such as UAV oblique photography, LiDAR scanning, or existing BIM models. This model must accurately reflect the spatial location, geometry, and appearance texture of the equipment within the substation.

[0067] Equipment Digital Definition: In the aforementioned high-precision 3D reality model, each equipment unit requiring inspection (such as circuit breakers, disconnectors, transformer bushings, surge arresters, etc.) is digitally defined manually or with the assistance of image recognition algorithms. A globally unique equipment identification code (i.e., equipment ID) is created for each equipment unit. It is recommended that this ID be consistent with the equipment asset code in the Production Management System (PMS) to achieve information interoperability.

[0068] Pre-set inspection points: For each equipment unit, one or more UAV inspection points are preset in three-dimensional space according to its inspection procedures (such as the angle and location to be observed). Each inspection point is not a simple coordinate, but a set of parameters including spatial coordinates (X, Y, Z), gimbal pitch angle, gimbal yaw angle, and camera zoom. This set of points constitutes the "digital script" for automated UAV operations.

[0069] Task Management and Data Binding Module: Used to generate flight missions and control the real-time binding of data and device IDs between the UAV flight control system and the flight data.

[0070] Specifically, task planning: Users initiate inspection tasks by selecting an area in the 3D model or by directly selecting the device ID. The task planning system automatically calculates and generates the optimal flight path and shooting sequence for the UAV based on the preset inspection point parameters in S1, ensuring safe and efficient flight paths.

[0071] Autonomous Flight and Data Acquisition: The UAV is equipped with a flight control system. After receiving mission commands, it takes off autonomously and flies to each preset inspection point. Upon arrival, the flight control system automatically controls the gimbal and camera, precisely adjusting the preset parameters for that point (e.g., rotating the gimbal to -30° to photograph the transformer oil level gauge), and then takes pictures.

[0072] Key: Real-time binding of data and device ID: At the moment the camera shutter is triggered, the flight control system automatically writes key information such as the device identification code (ID) associated with the current inspection point, high-precision timestamp (UTC time in this implementation, indicating the image capture time), BeiDou / GPS coordinates, and UAV attitude angle as metadata directly into the EXIF ​​information of the collected raw image data, or binds the metadata with the image data stream via a 5G / 4G data transmission radio and transmits it back to the ground station together. This step ensures from the source that every piece of data "comes with its own ID".

[0073] AI processing and storage module: used to analyze the returned data and organize it into structured data packets for storage in the index database.

[0074] AI-powered intelligent recognition: After completing its mission, the drone transmits image data, bound with metadata, back to the data center server. The server then uses a pre-trained AI image recognition algorithm (such as a convolutional neural network (CNN) based on deep learning) to automatically analyze the image data, identifying equipment status, instrument readings, temperature values ​​(infrared thermal imaging analysis), and potential defects (such as rust, damage, and overheating).

[0075] Generate a structured data packet: After AI recognition, the recognition result (e.g., {"Device ID": "PT-001", "Recognition Type": "Oil Temperature Reading", "Result": "67.5℃", "Status": "Normal"}) is used as new high-level metadata and is then bound again to the original data (image + basic metadata) generated in S2 to form a complete structured data packet containing multi-level information.

[0076] Indexed Storage: The structured data packets mentioned above are stored in a time-series database or relational database, rather than a traditional file system. In the database, a composite index is created using the device identification code (ID) and timestamp as the core key fields, while auxiliary indexes can be created for fields such as device type, defect level, and status. This storage method significantly improves the query speed for massive amounts of data. AI Intelligent Recognition: After completing its mission, the drone transmits image data bound to metadata back to the data center server. The server uses pre-trained AI image recognition algorithms (such as convolutional neural networks based on deep learning, CNN) to automatically analyze the image data, identifying device status, instrument readings, temperature values ​​(infrared thermal imaging analysis), and potential defects (such as corrosion, damage, and overheating).

[0077] Generate a structured data packet: After AI recognition, the recognition result (e.g., {"Device ID": "PT-001", "Recognition Type": "Oil Temperature Reading", "Result": "67.5℃", "Status": "Normal"}) is used as new high-level metadata and is then bound again to the original data (image + basic metadata) generated in S2 to form a complete structured data packet containing multi-level information.

[0078] Indexed storage: The structured data packets described above are stored in a time-series database or relational database, rather than a traditional file system. Within the database, a composite index is created using the device identification code (ID) and timestamp as the core key fields, while secondary indexes can be created for fields such as device type, defect level, and status. This storage method significantly improves the query speed for massive amounts of data.

[0079] Visual traceability module: Provides a 3D interactive interface to enable bidirectional traceability.

[0080] Specifically, forward tracing (from data to device): When a user finds a "damaged insulator" record in the list of defects identified by AI, they only need to click on the record, and the system can immediately read the device ID it is bound to (such as "IL-202"). Then, it automatically highlights and focuses on the device in the 3D visualization interface, clearly showing its spatial position and panoramic shape, achieving second-level positioning from data to physical location.

[0081] Reverse tracing (from device to data): When a user clicks on any device in the 3D visualization interface (e.g., clicking on "#1 main transformer" in the 3D model), the system automatically retrieves its device ID and initiates a query request to the database. The system immediately returns inspection images, raw data, and AI analysis results reports for all historical cycles of that device, organized in a timeline format. Users can easily compare data from different periods to analyze trends in device status changes.

[0082] Advanced Applications: Based on structured data, the system can automatically generate equipment health status reports and early warning information, and push them to relevant production management systems to drive the automation and intelligence of operation and maintenance decisions.

[0083] Computer storage media implementation methods: A computer-readable storage medium stores a computer program internally, the computer program being executed by a processor to implement the steps of the substation drone intelligent inspection data management method described above. Specific implementation methods for the substation drone intelligent inspection data management method can be found in the above-described embodiments.

[0084] Step 1: In the path planning stage of drone inspection, the inspection point parameters for each substation equipment used for path planning include the substation equipment ID to be inspected, the pose parameters of the drone during shooting, and the camera shooting parameters; the camera shooting parameters include the camera focal length.

[0085] The inspection point parameters are obtained in the following ways: (1) Construct a three-dimensional model of the target substation; the three-dimensional model can truly reflect the spatial position, geometry and appearance texture of each piece of equipment in the target substation; (2) In the three-dimensional model, according to the preset inspection procedure, set at least one inspection point and inspection point parameters for each piece of substation equipment that needs to be inspected, so that the drone can achieve the best shooting effect when shooting according to the inspection point parameters.

[0086] In other words: Model acquisition: First, a high-precision 3D real-world model of the target substation is acquired using existing technologies such as UAV oblique photography, LiDAR scanning, or existing BIM models. This model must accurately reflect the spatial location, geometry, and appearance texture of the equipment within the substation.

[0087] Equipment Digital Definition: In the aforementioned high-precision 3D reality model, each equipment unit requiring inspection (such as circuit breakers, disconnectors, transformer bushings, surge arresters, etc.) is digitally defined manually or with the assistance of image recognition algorithms. A globally unique equipment identification code (i.e., equipment ID) is created for each equipment unit. It is recommended that this ID be consistent with the equipment asset code in the Production Management System (PMS) to achieve information interoperability.

[0088] Pre-set inspection points: For each equipment unit, one or more UAV inspection points are preset in three-dimensional space according to its inspection procedures (such as the angle and location to be observed). Each inspection point is not a simple coordinate, but a set of parameters including spatial coordinates (X, Y, Z), gimbal pitch angle, gimbal yaw angle, and camera zoom. This set of points constitutes the "digital script" for automated UAV operations.

[0089] Step 2: During the UAV inspection phase, the UAV takes inspection images according to the above inspection point parameters, and binds the inspection images and their corresponding inspection point parameters and actual shooting information as a structured data packet for the inspection images, so that the inspection data is stored in the form of a structured data packet.

[0090] When the drone captures inspection images, the inspection point parameters and actual shooting information are written as metadata into the EXIF ​​information of the inspection image to obtain a structured data packet of the inspection image.

[0091] In short: Autonomous flight and data acquisition: The UAV is equipped with a flight control system. After receiving mission instructions, it takes off autonomously and flies to each preset inspection point. Upon arrival, the flight control system automatically controls the gimbal and camera, precisely adjusting the preset parameters for that point (such as rotating the gimbal to -30° to photograph the transformer oil level gauge), and then takes pictures.

[0092] Key: Real-time binding of data and device ID: At the moment the camera shutter is triggered, the flight control system automatically writes key information such as the device identification code (ID) associated with the current inspection point, high-precision timestamp (UTC time in this implementation, indicating the image capture time), BeiDou / GPS coordinates, and UAV attitude angle as metadata directly into the EXIF ​​information of the collected raw image data, or binds the metadata with the image data stream via a 5G / 4G data transmission radio and transmits it back to the ground station together. This step ensures from the source that every piece of data "comes with its own ID".

[0093] During the image recognition stage, the recognition result of the inspection image is bound to the structured data packet of the inspection image as a new structured data packet.

[0094] AI-powered intelligent recognition: After completing its mission, the drone transmits image data, bound with metadata, back to the data center server. The server then uses a pre-trained AI image recognition algorithm (such as a convolutional neural network (CNN) based on deep learning) to automatically analyze the image data, identifying equipment status, instrument readings, temperature values ​​(infrared thermal imaging analysis), and potential defects (such as rust, damage, and overheating).

[0095] Generate a structured data packet: After AI recognition, the recognition result (e.g., {"Device ID": "PT-001", "Recognition Type": "Oil Temperature Reading", "Result": "67.5℃", "Status": "Normal"}) is used as new high-level metadata and is then bound again to the original data (image + basic metadata) generated in S2 to form a complete structured data packet containing multi-level information.

[0096] Step 3: Store the structured data packets in a database; the database type is a time-series database or a relational database; the database index includes the substation equipment ID and the shooting time.

[0097] Instead of a traditional file system, the structured data packets are stored in a time-series database or a relational database. Within the database, a composite index is created using the device identification code (ID) and timestamp as the core key fields, while secondary indexes can be created for fields such as device type, defect level, and status. This storage method significantly improves the query speed for massive amounts of data.

[0098] Specifically, the computer-readable storage medium can be volatile memory or non-volatile memory, or may include both. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be random access memory (RAM), which serves as an external cache. For example, Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced Synchronous DRAM (ESDRAM), SynchLink DRAM (SLDRAM), or Direct Rambus RAM (DRRAM).

[0099] This invention has the following characteristics: A sophisticated inspection point pre-definition and mapping mechanism based on digital twins: This invention pioneers a method that, within a high-precision 3D model of a substation, not only digitally defines equipment but also pre-defines one or more UAV inspection points for each device, each containing complete shooting parameters (spatial coordinates, gimbal angle, camera parameters). This establishes a precise mapping relationship of "equipment ID ←→ position in the 3D model ←→ optimal UAV shooting pose." This positioning method surpasses simple GPS coordinates, enabling precise control over equipment observation angles. It solves the problems of non-standard UAV shooting poses, inability to cover key equipment observation areas, and inconsistent data quality due to reliance on pilot experience in traditional methods.

[0100] A real-time automatic binding method for inspection data and equipment identity at the data collection source: At the moment the drone takes a picture, the flight control system automatically writes the equipment identification code (ID) associated with the current inspection point as core metadata, along with other information (timestamp, coordinates, etc.), directly into the raw data. This is a "pre-processing" or "endogenous" data association method, rather than a later "post-processing" matching. It fundamentally solves the problems of low efficiency and high error rate caused by later manual screening or coarse spatial matching, achieving 100% automation and accuracy in data association, laying a reliable data foundation for all subsequent advanced applications.

[0101] A multi-dimensional indexed storage and bidirectional traceability model based on device identification and timestamps: This invention creatively encapsulates unstructured inspection image data, AI recognition results, device IDs, timestamps, etc., into structured data packets, and adopts a database storage scheme with device IDs and timestamps as the core indexes. Based on this model, seamless bidirectional visual traceability is achieved "from data to device" (clicking on a defect record to locate the device in a 3D model) and "from device to data" (clicking on a 3D device to view its full historical data). This solves the problems of slow querying of massive inspection data, difficulty in historical comparison, and data silos, providing unprecedented data insight capabilities and decision support efficiency.

[0102] Achieve precise and automated association of data sources: By establishing accurate digital mapping relationships before the execution of inspection tasks and automatically binding device identity information at the moment of data collection, the accuracy and automation of the association between inspection data and physical equipment are completely solved from the source, eliminating manual intervention and greatly improving the efficiency and reliability of data processing.

[0103] Build efficient and accurate two-way traceability capabilities: simultaneously support two-way visualized traceability "from data to device" and "from device to data". This enables users to quickly locate the geographical location of problematic devices from massive amounts of data, and also easily obtain all inspection records for a single device throughout its entire lifecycle, providing strong data support for equipment status assessment and intelligent diagnosis.

[0104] Establish a structured data management system: Transform traditional file-based storage into structured storage and management based on database indexes. By defining a unified data packet format and a multi-dimensional index centered on device ID and timestamps, break down data silos, and lay a solid foundation for advanced querying of inspection data, big data analysis, and seamless integration with upper-level management systems. Ultimately, this will drive the substation operation and maintenance towards a comprehensive digital and intelligent transformation.

[0105] 1. Association Precision and Accuracy: Ultra-high precision and extremely high reliability. Direct association via predefined, precise inspection points (including angles) achieves 100% accuracy. It eliminates the need for post-matching with error-prone GPS, fundamentally eliminating mismatches. This achieves unique, accurate, and error-free association between data and devices, providing a reliable data foundation for all subsequent advanced applications and resolving the most fundamental shortcomings of existing technologies.

[0106] 2. Automation and Efficiency: Fully automated and extremely efficient. From task generation and data collection and binding to AI processing and storage, the entire process requires no manual intervention. It completely frees maintenance personnel from tedious and inefficient data processing work, improving efficiency by over 90%. It truly achieves full-process automation of inspection data management, significantly reducing labor and time costs, and making large-scale, high-frequency inspections possible.

[0107] 3. Data Traceability Capabilities: Powerful two-way visual traceability. Simultaneously supports: 1) Forward traceability (from data to device): Precise positioning. 2) Reverse traceability (from device to data): One-click access to the entire lifecycle archive. Provides unprecedented data insight capabilities. Makes massive historical data "come alive," providing powerful and convenient tools for equipment status evaluation, defect evolution tracking, and predictive maintenance, resulting in a qualitative leap in decision support capabilities.

[0108] 4. Data Value and Application Depth: The data is structured and has high value density. It forms multi-dimensional structured data packages containing raw data, asset information, and diagnostic results, facilitating integration with upper-level management systems and enabling big data analysis and intelligent decision-making. It breaks down "data silos," deeply integrating inspection data into production business processes, driving the upgrade of the operation and maintenance model from "passive maintenance" to "proactive early warning" and "predictive maintenance."

[0109] 5. Standardization and Replicability: Standardized processes and replicable results. Inspection points and shooting parameters are all predefined "digital standards," ensuring consistent and comparable data quality for each inspection, without discrepancies due to different personnel. This standardization and digitization of inspection operations facilitates large-scale promotion and application, improving the standardization and consistency of maintenance across the entire industry.

[0110] In summary, compared with the closest existing technologies, this invention fundamentally solves the three major pain points of existing technologies—inaccurate correlation, low automation, and difficulty in traceability—through two core innovations: "predefined digital mapping" and "data source binding." This invention is not merely an improvement on a single technology, but rather the construction of a complete, closed-loop intelligent data management system. It achieves a leap from "drones replacing humans" to "data-driven intelligent operation and maintenance," significantly improving the digitalization, automation, and intelligence levels of substation operation and maintenance, and possesses outstanding substantive characteristics and significant progress.

[0111] Finally, it should be noted that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still make modifications to the technical solutions described in the foregoing embodiments without creative effort, or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for managing intelligent inspection data of substations using unmanned aerial vehicles (UAVs), characterized in that, The method includes: During the drone inspection phase, inspection images taken by the drone according to the inspection point parameters are acquired. The inspection images and their corresponding inspection point parameters, substation equipment IDs and actual shooting information are bound together as a structured data packet for the inspection image, so that the inspection data is stored in the form of a structured data packet. The inspection point parameters include the pose parameters during drone shooting and the camera shooting parameters.

2. The method for intelligent inspection data management of substations using unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The method further includes: in the image recognition stage, binding the recognition result of the inspection image with the structured data packet of the inspection image obtained in the UAV inspection stage as a new structured data packet.

3. The substation unmanned aerial vehicle (UAV) intelligent inspection data management method according to claim 1 or 2, characterized in that, The structured data packets during the UAV inspection phase are obtained as follows: when the UAV captures inspection images, the inspection point parameters and actual shooting information are written as metadata into the EXIF ​​information of the inspection images to obtain the structured data packets of the inspection images.

4. The substation UAV intelligent inspection data management method according to claim 1 or 2, characterized in that, The actual shooting information includes the shooting time; the method further includes: storing the structured data packet in a database; the database type is a time-series database or a relational database; the database index includes the substation equipment ID and the shooting time.

5. The substation unmanned aerial vehicle (UAV) intelligent inspection data management method according to claim 1, characterized in that, The inspection point parameters are obtained in the following way: (1) Construct a three-dimensional model of the target substation; the three-dimensional model can realistically reflect the spatial location, geometry and appearance texture of each piece of equipment in the target substation; (2) In the three-dimensional model, according to the preset inspection procedure, at least one inspection point and inspection point parameters are set for each substation equipment that needs to be inspected, so that the drone can achieve the best shooting effect when shooting according to the inspection point parameters.

6. The substation unmanned aerial vehicle (UAV) intelligent inspection data management method according to claim 5, characterized in that, The 3D model is obtained through UAV oblique photography, laser scanning, or BIM modeling.

7. The substation unmanned aerial vehicle (UAV) intelligent inspection data management method according to claim 1, characterized in that, The camera shooting parameters include the camera focal length.

8. The method for managing intelligent inspection data of substations using unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The substation equipment ID is consistent with the equipment asset code in the substation production management system.

9. A data management system for intelligent inspection of substations using unmanned aerial vehicles (UAVs), comprising a processor, characterized in that, The processor is used to execute a computer program to implement the steps of the substation unmanned aerial vehicle (UAV) intelligent inspection data management method as described in any one of claims 1 to 8.

10. A computer-readable storage medium, wherein a computer program is stored internally, characterized in that, The computer program is executed by a processor to implement the steps of the substation unmanned aerial vehicle (UAV) intelligent inspection data management method as described in any one of claims 1 to 8.

Images