Photovoltaic power station intelligent inspection method and system based on unmanned aerial vehicle

Abstract

The invention provides a photovoltaic power station intelligent inspection method and system based on an unmanned aerial vehicle, and the method comprises the steps: collecting visible light and thermal imaging data, and synchronously binding the geographic coordinates of equipment; splicing and generating a panoramic base map based on visible light and thermal imaging data, and associating the panoramic base map with geographic coordinates of equipment to construct a spatialized topological database; a panoramic base map and a topology database are integrated on a GIS platform, and equipment hierarchical positioning is carried out; generating an optimized route in response to a positioning instruction of equipment hierarchical positioning, controlling the unmanned aerial vehicle to execute a dual-light collaborative fixed-point inspection task, and obtaining inspection data; the visible light and thermal imaging data and the inspection data are fused, cross-modal fault diagnosis is carried out, and an environment adaptive decision report is generated; the equipment degradation model is constructed based on historical fault feature data output by cross-modal fault diagnosis, the maintenance strategy is generated and the inspection task is fed back and optimized, and compared with a mode of independently collecting image information and manually marking, the inspection efficiency and accuracy of the unmanned aerial vehicle to the photovoltaic power station are improved.

Description

Technical Field

[0001] The present invention relates to the field of photovoltaic power station fire protection technology, and in particular to a photovoltaic power station intelligent inspection method and system based on unmanned aerial vehicle (UAV) technology. Background Art

[0002] With the large-scale development of photovoltaic power plants, traditional manual inspection methods, due to low efficiency and limited coverage, have become unable to meet the needs of refined power plant operation and maintenance. In recent years, drone inspection technology has been gradually adopted. Existing drone aerial photography often collects image information individually and uses manual annotation or simple GPS recording, resulting in a high error rate in fault location. This single data source also results in poor fault identification accuracy.

[0003] Therefore, how to improve the efficiency and accuracy of drone inspections of photovoltaic power stations has become a technical problem that technical personnel in this field urgently need to solve. Summary of the Invention

[0004] The present invention provides a method and system for intelligent inspection of photovoltaic power stations based on drones, which are used to solve the defects of drone inspection in the prior art, such as poor efficiency and accuracy.

[0005] In a first aspect, the present invention provides a method for intelligent inspection of photovoltaic power stations based on drones, comprising: Visible light and thermal imaging data are collected by drones equipped with dual-light cameras, and the geographic coordinates of the devices are simultaneously bound; Generate a panoramic base map based on the visible light and thermal imaging data, and build a spatial topological database by associating the device geographic coordinates; Integrate the panoramic base map and topology database on the GIS platform to perform hierarchical positioning of equipment; In response to the positioning instructions of the hierarchical positioning of the equipment, an optimized route is generated, and the UAV is controlled to perform the dual-light coordinated fixed-point inspection task and obtain the inspection data; fusing the visible light and thermal imaging data with the inspection data to perform cross-modal fault diagnosis and generate an environmentally adaptive decision report; An equipment degradation model is constructed based on the historical fault feature data output by the cross-modal fault diagnosis, and a maintenance strategy is generated and feedback is provided to optimize the inspection task.

[0006] According to a method for intelligent inspection of a photovoltaic power station based on a drone provided by the present invention, associating the geographical coordinates of the device includes: Parsing the geolocation metadata of aerial images as spatial reference; Based on the spatial reference, the device pixel position is mapped to the geographic coordinate system through a coordinate transformation algorithm, and a topological association relationship between the device unique code and the geographic coordinates is established.

[0007] According to the present invention, a method for intelligent inspection of photovoltaic power stations based on drones is provided, wherein the hierarchical positioning of equipment is performed, including: Render the electrical connection topology in the GIS platform according to the three-level structure of array-string-module; In response to the device code input command, the target device and its branch path are highlighted.

[0008] According to a method for intelligent inspection of a photovoltaic power station based on a drone provided by the present invention, generating an optimized route in response to a positioning instruction of the hierarchical positioning of the equipment includes: When receiving manual path drawing instructions, it automatically generates obstacle avoidance flight trajectories based on the 3D obstacle model; In response to a branch coding input instruction, the topology database is called to generate a bionic flight path covering the branch.

[0009] According to a method for intelligent inspection of a photovoltaic power station based on a drone provided by the present invention, cross-modal fault diagnosis includes: Performing spatiotemporal alignment processing on the visible light and thermal imaging data and the inspection data for the same device; When there is spatial overlap between the visible light anomaly features and the thermal imaging temperature anomaly and inspection anomaly data, the cross-modal verification mechanism is triggered to improve the diagnostic confidence.

[0010] According to a drone-based intelligent inspection method for photovoltaic power stations provided by the present invention, generating an environmental adaptive decision report includes: Dynamically calculate the alarm threshold based on the real-time ambient temperature and the historical temperature rise pattern of the equipment; Based on the alarm threshold, the associated equipment maintenance record library generates a differentiated maintenance plan as an environmental adaptive decision report.

[0011] According to the present invention, a method for intelligent inspection of a photovoltaic power station based on a drone is provided, further comprising: Determining the hot spot properties at the corresponding positions of the spatial overlap; If the hot spot attribute indicates that the hot spot area is circular and located at the edge of the battery cell, it is determined that the diode has failed; If the hot spot attribute indicates that the hot spot is accompanied by shadow features in the visible light image, it is determined to be blocked by vegetation or dust accumulation.

[0012] According to a drone-based intelligent inspection method for photovoltaic power stations provided by the present invention, the construction of an equipment degradation model includes: Extract the historical temperature series of the equipment to train the time series prediction model; When the predicted value of the prediction model exceeds a safety threshold, a preventive maintenance work order is automatically generated and linked to a spare parts inventory system.

[0013] According to the present invention, a method for intelligent inspection of a photovoltaic power station based on a drone is provided, further comprising: Real-time analysis of video streams during flight to detect intruder targets; When an intrusion is detected, the sound and light alarm device is activated and the trajectory is recorded. In a second aspect, the present invention further provides a photovoltaic power station intelligent inspection system based on a drone, comprising: The acquisition module is used to collect visible light and thermal imaging data through a drone equipped with a dual-light camera and synchronize the data with the device's geographic coordinates. An association module, configured to generate a panoramic base map based on the visible light and thermal imaging data, and to associate the device geographic coordinates to construct a spatialized topological database; A positioning module is used to integrate the panoramic base map and the topological database on the GIS platform to perform hierarchical positioning of the device; An inspection module is used to generate an optimized route in response to the positioning instructions of the hierarchical positioning of the equipment, control the UAV to perform dual-light coordinated fixed-point inspection tasks, and obtain inspection data; A diagnostic module, configured to fuse the visible light and thermal imaging data with the inspection data to perform cross-modal fault diagnosis and generate an environmentally adaptive decision report; A maintenance module is used to build an equipment degradation model based on the historical fault feature data output by the cross-modal fault diagnosis, generate a maintenance strategy and provide feedback to optimize the inspection task.

[0014] In a third aspect, the present invention also provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the intelligent inspection method for photovoltaic power stations based on drones as described above is implemented.

[0015] In a fourth aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-described drone-based intelligent inspection methods for photovoltaic power stations.

[0016] In a fifth aspect, the present invention further provides a computer program product, comprising a computer program, which, when executed by a processor, implements any of the above-described drone-based intelligent inspection methods for photovoltaic power stations.

[0017] The present invention provides a method and system for intelligent inspection of photovoltaic power stations based on drones. The method collects visible light and thermal imaging data through drones equipped with dual-light cameras, and synchronously binds the geographic coordinates of the equipment; generates a panoramic base map based on the splicing of visible light and thermal imaging data, and builds a spatialized topological database by associating the geographic coordinates of the equipment; integrates the panoramic base map and the topological database on the GIS platform to perform hierarchical positioning of the equipment; generates an optimized route in response to the positioning instructions of the hierarchical positioning of the equipment, controls the drone to perform dual-light coordinated fixed-point inspection tasks, and obtains inspection data; fuses the visible light and thermal imaging data with the inspection data, performs cross-modal fault diagnosis and generates an environmental adaptive decision report; builds an equipment degradation model based on the historical fault feature data output by the cross-modal fault diagnosis, generates a maintenance strategy and provides feedback to optimize the inspection tasks, splices the visible light and thermal imaging data, associates the geographic coordinates of the equipment to build a spatialized topological database, and fuses the inspection data to perform fault diagnosis and analysis. Compared with the existing method of separately collecting image information and manually annotating it, the method effectively improves the efficiency and accuracy of drone inspections of photovoltaic power stations. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction is given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 Schematic diagram of the process of the intelligent inspection method of photovoltaic power stations based on drones provided in this embodiment; Figure 2 Schematic diagram of the structure of the intelligent inspection system for photovoltaic power stations based on drones provided in this embodiment; Figure 3 Schematic diagram of the structure of the electronic device provided in this embodiment. DETAILED DESCRIPTION

[0020] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0021] Figure 1 This is a flow chart of the intelligent inspection method for photovoltaic power stations based on drones provided in this embodiment.

[0022] like Figure 1As shown, the intelligent inspection method for photovoltaic power stations based on drones provided in an embodiment of the present invention mainly includes the following steps: 101. Use a drone equipped with a dual-light camera to collect visible light and thermal imaging data, and synchronize the device's geographic coordinates.

[0023] Specifically, a drone equipped with a high-precision navigation module is selected, and the visible light camera and thermal imaging camera are integrated into the drone gimbal so that their optical axes are parallel and maintain a relatively fixed position, while ensuring that the drone navigation module can obtain high-precision geographic coordinate information in real time.

[0024] The drone flies along a pre-set inspection route. During flight, its visible light camera and thermal imaging camera simultaneously capture image data of the photovoltaic power plant. The drone's navigation module acquires the device's own geographic coordinates in real time. The data transmission module timestamps these coordinates with the collected visible light and thermal imaging data and transmits them to the ground control terminal. The ground control terminal uses data processing algorithms to accurately bind the geographic coordinates to the image data at the corresponding time based on the timestamp, generating an image dataset with geographic coordinate information.

[0025] Visible light images can clearly show problems such as damage and dust coverage on the surface of photovoltaic modules. Thermal imaging data can quickly detect abnormal module temperatures. Combined with geographic coordinates, the location of faulty modules can be accurately located, greatly improving the accuracy and efficiency of photovoltaic power station fault detection, reducing the cost and time of manual inspections, and realizing intelligent and efficient inspections of photovoltaic power stations.

[0026] 102. Generate a panoramic base map based on the stitching of visible light and thermal imaging data, and build a spatial topological database by associating the device's geographic coordinates.

[0027] Specifically, the process of stitching together visible light and thermal imaging data to generate a panoramic basemap involves preprocessing the visible light and thermal imaging data, such as denoising and contrast enhancement, to improve image quality. Feature points from the visible light and thermal images are extracted using algorithms such as SIFT and SURF. The similarity between these feature points is calculated using feature descriptors. Methods such as nearest neighbor matching based on Euclidean distance are then used to accurately match feature points between the different images and establish correspondence between them. Based on the matched feature points, a homography matrix is ​​used to calculate the transformation relationship between the images. The images are then geometrically transformed and projected to align overlapping areas. Overlapping areas are then fused using algorithms such as weighted averaging and multi-resolution fusion to eliminate stitching gaps and brightness differences, ultimately generating a complete panoramic basemap.

[0028] The resulting panoramic basemap integrates the advantages of visible light and thermal imaging data, visually displaying both the appearance details of PV power plant components and the temperature distribution, enabling personnel to fully understand the plant's operating status. This panoramic display reduces the time spent manually reviewing individual images, improving inspection efficiency and enabling more accurate detection of minor faults, reducing the risk of missed detections. This facilitates efficient operation and maintenance of PV power plants, provides fault warnings, and reduces operational costs.

[0029] Among them, the spatial topological database is constructed by associating the device's geographic coordinates, including: parsing the geographic location metadata of the aerial image as a spatial reference; based on the spatial reference, mapping the device pixel position to the geographic coordinate system through a coordinate transformation algorithm, and establishing a topological association between the device's unique code and the geographic coordinates.

[0030] Specifically, the visible light and thermal imaging data and geographic coordinates corresponding to each PV module are extracted from the generated panoramic basemap and its associated geographic coordinate information. Image processing algorithms are used to identify key PV plant equipment (such as combiner boxes and inverters), determine their locations within the panoramic basemap, and associate them with their corresponding geographic coordinates. Based on the PV plant's design drawings and actual layout, combined with the extracted geographic coordinates, the topological relationships between the various devices, including electrical connections and physical locations, are determined. For example, the connection paths between PV modules and combiner boxes, and between combiner boxes and inverters, are identified to construct a topological network between the devices. The device data and topological relationship data, along with geographic coordinates, are imported into a spatial topological database using a spatial database storage format (such as ESRI Shapefile or GeoJSON). Indexing and query mechanisms are established to enable efficient database management and rapid retrieval, facilitating subsequent data access and analysis.

[0031] The constructed spatial topological database combines the geographic information and topological relationships of photovoltaic power station equipment to achieve accurate spatial positioning of equipment failures, allowing staff to quickly locate the location of faulty equipment. Through topological relationship analysis in the database, the scope of fault impact can be predicted and more scientific maintenance plans can be formulated. At the same time, it facilitates macro-management and operation and maintenance planning of the entire photovoltaic power station, improves the intelligence and digitalization level of power station operation and maintenance, and reduces the difficulty and cost of operation and maintenance management.

[0032] 103. Integrate panoramic base maps and topological databases into the GIS platform to perform hierarchical positioning of equipment.

[0033] The panoramic base map and topology database can be integrated into the GIS platform, and the electrical connection topology can be rendered in the GIS platform according to the three-level structure of array-string-component; in response to the device code input command, the target device and its branch path are highlighted.

[0034] Specifically, the generated panoramic base map is converted into a raster data format supported by the GIS platform (such as GeoTIFF) and imported into the GIS platform through a data interface. The vector data (device coordinates, topological relationships, etc.) in the spatialized topological database is connected to the GIS platform using standard spatial data protocols (such as WFS and WMS), integrating the panoramic base map with the topological database. The GIS platform then analyzes the device connection relationship data in the topological database based on the actual electrical design of the PV power plant. The electrical connection topology is rendered layered according to the three-level structure of array, string, and module, using different symbols, colors, and hierarchical relationships. The GIS platform's map rendering engine clearly and intuitively displays each level of structure on the interface. A device code input box is provided on the GIS platform interface. Once the user enters the device code, the platform retrieves the corresponding device record in the topological database based on the code. The coordinates of the target device and the connection relationship data of its associated branch are obtained. The panoramic base map uses visual effects such as highlighting colors and thickening lines to highlight the target device and its associated branch path. Users can also click to view detailed device information.

[0035] The GIS platform integrates a panoramic basemap and a topology database for visualization, enabling PV power plant operators to intuitively and comprehensively understand the plant's equipment layout and electrical connections. Three-level structural rendering facilitates a macro- to micro-level understanding of the plant structure, allowing for quick identification of problem areas. The device code input highlighting function accurately locates faulty equipment and its impact area, significantly improving troubleshooting efficiency. This visual management approach facilitates more effective O&M planning, enhances the scientific and intelligent nature of PV power plant O&M management, and reduces the costs of manual inspections and troubleshooting.

[0036] 104. Respond to the positioning instructions of the equipment's hierarchical positioning to generate optimized routes, control the drone to perform dual-light coordinated fixed-point inspection tasks, and obtain inspection data.

[0037] The system generates optimized flight paths in response to positioning commands from hierarchical device positioning. This includes automatically generating obstacle-avoidance flight trajectories based on a 3D obstacle model when receiving manual path-drawing commands, and generating biomimetic flight paths covering these branches using a topological database when responding to branch-coding input commands. The system then controls the drone based on the biomimetic flight path to perform dual-optical coordinated fixed-point inspections and acquire inspection data.

[0038] Specifically, a manual path drawing module is included in the GIS platform interface, allowing users to draw the desired drone flight path using a mouse or touch. After obtaining the manually drawn path, the system combines it with a pre-built 3D obstacle model of the photovoltaic power plant (including components, brackets, and equipment), using a path planning algorithm to calculate a safe flight trajectory that avoids obstacles. The generated trajectory is then smoothed to prevent sudden changes during flight.

[0039] When the user enters the branch code, the GIS platform retrieves the device connection relationship and coordinate information of the branch from the topological database. Based on bionic principles such as bees searching for honey and birds migrating, combined with the actual layout and equipment distribution of the photovoltaic power station, it generates a bionic flight path that covers all devices in the branch and has the shortest path and the highest flight efficiency. At the same time, the path is optimized to ensure that the drone remains stable during flight.

[0040] The generated bionic flight path or obstacle avoidance flight trajectory is transmitted to the UAV flight control system; the UAV flies according to the preset path. During the flight, the dual-light camera automatically collects visible light and thermal imaging data of the photovoltaic power station according to the set time interval or distance interval; at the same time, the UAV feeds back its own position and status information to the ground control terminal in real time to ensure the smooth completion of the inspection mission.

[0041] The automatic obstacle avoidance function of manually drawn paths gives users the ability to flexibly plan inspection routes. Combined with the obstacle avoidance trajectory generated by the three-dimensional obstacle model, it ensures the flight safety of the drone in complex environments. The bionic flight path generated based on branch coding can perform efficient and comprehensive inspections on specific branches to avoid missing key equipment. The execution of dual-light coordinated fixed-point inspection tasks based on the bionic path can accurately obtain equipment data and improve the integrity and accuracy of the inspection data. Through automated path generation and inspection task execution, the inspection efficiency of photovoltaic power stations is greatly improved, the risk of manual operation and operation and maintenance costs are reduced, and strong support is provided for the intelligent operation and maintenance of power stations.

[0042] 105. Integrate visible light and thermal imaging data with inspection data to perform cross-modal fault diagnosis and generate environmental adaptive decision reports.

[0043] The collected visible light images, thermal images, and inspection data are preprocessed separately, including image denoising, enhancement, and geometric correction, while the inspection data is formatted and outliers are removed to improve data quality. Image recognition algorithms are used to extract PV module appearance features (such as damage and dust coverage) from visible light images, and temperature distribution features (such as hot spots and areas of abnormal temperature) from thermal images. The features of the visible light and thermal imaging data are correlated and integrated by combining geographic coordinates and device information from the inspection data. Using algorithms such as weighted fusion and decision-level fusion, comprehensive data containing module appearance, temperature, and location information is generated and stored in a database.

[0044] By integrating visible light and thermal imaging data with inspection data, a multi-dimensional presentation of the equipment status of photovoltaic power stations can be achieved. Workers can not only visually view the appearance of components, but also understand their temperature anomalies, effectively avoiding missed detections and false detections caused by single data, and improving the accuracy and comprehensiveness of fault diagnosis. By associating data fusion with geographic coordinates, faulty equipment can be quickly located, providing an accurate basis for maintenance decisions, optimizing the operation and maintenance management process, improving the overall operation and maintenance efficiency of photovoltaic power stations, and reducing operation and maintenance costs and safety risks.

[0045] Perform spatiotemporal alignment of visible light, thermal imaging, and inspection data for the same device. When visible light anomaly overlaps with thermal imaging temperature anomalies and inspection anomaly data, a cross-modal verification mechanism is triggered to enhance diagnostic confidence. The hot spot attributes corresponding to the spatial overlap are determined. If the hot spot attributes indicate a circular hot spot located at the edge of the cell, the diode is considered faulty. If the hot spot attributes indicate a hot spot accompanied by shadows in the visible light image, the hot spot is considered obscured by vegetation or dust accumulation.

[0046] Specifically, relevant data of the same device is extracted from the fused visible light, thermal imaging, and inspection data. Based on the geographic coordinate information recorded in the data, different modal data are mapped to a unified geographic space coordinate system through coordinate conversion and projection transformation. According to the timestamp of data acquisition, the data is sorted and integrated in chronological order to achieve precise alignment of the same device data in time and space dimensions. Image recognition and data analysis algorithms are used to detect abnormal features in visible light data (such as component cracks), abnormal temperature areas in thermal imaging data, and abnormal records in inspection data (such as voltage fluctuations). When the three types of abnormalities mentioned above are detected to overlap in spatial position, a cross-modal verification mechanism is triggered. The multimodal data of the overlapping area is jointly analyzed through machine learning algorithms, and the diagnostic confidence of the abnormal situation is calculated by combining historical fault data with the operation rules of the equipment. If the confidence exceeds the set threshold, the equipment is determined to be faulty and a detailed diagnostic report is generated.

[0047] By aligning the spatiotemporal data of multimodal data from the same device, the consistency of different types of data in the spatiotemporal dimensions is ensured, laying the foundation for accurate analysis of device status. The introduction of a cross-modal verification mechanism avoids the limitations of single-modal data diagnosis and utilizes the complementarity of multimodal data to significantly improve the confidence and reliability of fault diagnosis, reducing misjudgments and missed judgments. Accurate fault diagnosis can help operation and maintenance personnel quickly identify problematic equipment, develop targeted maintenance plans, improve the operation and maintenance efficiency of photovoltaic power stations, and reduce operation and maintenance costs and potential safety hazards.

[0048] The environmental adaptive decision report dynamically calculates the alarm threshold based on the real-time ambient temperature and the historical temperature rise pattern of the equipment; based on the alarm threshold, the associated equipment maintenance record library generates differentiated maintenance plans as the environmental adaptive decision report.

[0049] Specifically, the system uses a thermal imaging camera and environmental monitoring equipment mounted on a drone to collect real-time data on the equipment and ambient temperature of the photovoltaic power station. The system then retrieves the temperature rise data of the equipment at different time periods and ambient temperatures from the equipment's historical data database, analyzes the historical temperature rise patterns, and establishes a correspondence model between the equipment temperature rise and ambient temperature. The real-time collected ambient temperature data is substituted into this correspondence model, and combined with the equipment's normal operating temperature range, the alarm threshold for the equipment in the current environment is dynamically calculated. The alarm threshold is weighted and adjusted based on factors such as the equipment's importance and operating age to ensure the rationality of the threshold setting. The calculated alarm threshold is compared with the equipment's current temperature. When the equipment temperature exceeds the alarm threshold, a maintenance plan generation process is triggered. The system then links the equipment's maintenance record database to retrieve information such as the equipment's historical maintenance history, fault type, and handling method. A differentiated maintenance plan is generated for the equipment based on the equipment's current status, historical maintenance history, and operational experience. This information, including alarm thresholds, equipment status, and maintenance plan, is integrated to generate an environmentally adaptive decision report.

[0050] Dynamically calculating alarm thresholds based on real-time ambient temperature and the historical temperature rise patterns of the equipment fully accounts for the impact of environmental factors on equipment temperature, avoiding false or missed alarms caused by fixed thresholds and achieving more accurate fault warnings. By generating differentiated maintenance plans based on the alarm threshold-associated equipment maintenance record library, personalized maintenance strategies can be formulated for the specific circumstances of different equipment, improving the pertinence and effectiveness of maintenance plans. Environmentally adaptive decision-making reports provide comprehensive and accurate decision-making basis for operation and maintenance personnel, helping them to quickly formulate scientific and reasonable operation and maintenance plans, thereby improving the intelligent level of photovoltaic power station operation and maintenance, reducing operation and maintenance costs, and ensuring stable and efficient operation of the power station.

[0051] 106. Based on the historical fault feature data output by cross-modal fault diagnosis, an equipment degradation model is constructed to generate maintenance strategies and provide feedback to optimize inspection tasks.

[0052] Specifically, from the results of cross-modal fault diagnosis output, multimodal fault features such as visible light anomaly features, thermal imaging temperature anomaly data, and inspection anomaly records of historical faulty equipment are screened and extracted; at the same time, related information such as the ambient temperature when the fault occurred and the equipment operating time are obtained to construct a complete historical fault feature dataset.

[0053] Use machine learning algorithms (such as LSTM and random forest) to analyze historical fault feature data sets and explore how fault features change over time. Combined with equipment design parameters and operating conditions, establish a mathematical model that reflects the degradation of equipment performance over time and simulate the entire degradation process from normal operation to failure.

[0054] Based on the constructed equipment degradation model, the future degradation trend and failure probability of the equipment are predicted. Based on the prediction results, combined with the equipment's importance level and maintenance cost, differentiated maintenance strategies such as preventive maintenance and condition-based maintenance are formulated. The maintenance time, maintenance content and required resources are clearly defined.

[0055] The generated maintenance strategy is compared and analyzed with the current inspection tasks. For high-risk and easily deteriorated equipment, the inspection frequency, inspection path and data collection parameters are adjusted; for example, the inspection frequency is increased and the flight path is optimized to cover key equipment to ensure that the inspection tasks are more targeted. The optimized inspection tasks are then fed back to the drone inspection system.

[0056] An equipment degradation model is constructed based on cross-modal fault diagnosis data, fully leveraging the advantages of multimodal data to accurately characterize the equipment degradation process and provide a reliable basis for maintenance strategy formulation. The generated differentiated maintenance strategy achieves a shift from passive maintenance to active prevention, reducing the probability of sudden failures and the risk of equipment damage. The optimized inspection tasks focus on potentially faulty equipment, improving the efficiency of inspection resource utilization and avoiding waste of inspection resources. This improves the overall scientificity and accuracy of photovoltaic power station operation and maintenance, effectively reducing operation and maintenance costs and ensuring the long-term stable operation of the power station.

[0057] Furthermore, based on the above embodiment, this embodiment also includes: real-time analysis of video streams during flight to detect intrusion targets; when intrusion behavior is identified, an audible and visual alarm device is linked and the trajectory is recorded.

[0058] Specifically, during a PV power plant inspection, the drone uses its onboard high-definition camera to capture real-time video stream data. This video stream is transmitted to an onboard processor or ground control terminal for pre-processing, such as noise reduction and contrast enhancement, to improve image quality. Deep learning-based object detection algorithms (such as YOLO and SSD) analyze the pre-processed video stream frame by frame to identify potential intruders, such as people and vehicles. By setting filtering conditions such as target size and speed, interference factors such as birds and fluttering debris are eliminated to accurately locate the intruder. Detected targets are continuously tracked, analyzing their movement trajectory, dwell time, and other behavioral characteristics. Based on pre-set intrusion rules (such as entering a restricted area or loitering for an extended period), the drone determines whether the target is engaging in intrusion. Upon detecting an intrusion, the drone uses its wireless communication module to send commands to the PV power plant's audio and visual alarm system, triggering the system to emit sirens and warning lights to deter the intruder. The drone also records the target's real-time location, movement trajectory, and related video footage, stores the data locally, and simultaneously transmits it to the power plant's monitoring center for easy tracing.

[0059] Real-time detection of intrusion targets during drone flight can achieve all-round dynamic monitoring of photovoltaic power stations and promptly identify potential security threats; after identifying intrusion behavior, the sound and light alarm device is activated to effectively deter intruders or vehicles and reduce the probability of security incidents; the trajectory of intrusion targets and video images are recorded to provide strong evidence for security incident investigations and facilitate the clarification of responsibilities; this technology improves the safety protection level of photovoltaic power stations, reduces the cost of manual security inspections, and ensures the safety of power station equipment and property.

[0060] Based on the same general inventive concept, the present invention also protects a photovoltaic power station intelligent inspection system based on drones. The photovoltaic power station intelligent inspection system based on drones described below and the photovoltaic power station intelligent inspection method based on drones described above can refer to each other.

[0061] Figure 2 Schematic diagram of the structure of the intelligent inspection system for photovoltaic power stations based on drones provided in this embodiment.

[0062] like Figure 2 As shown, this embodiment provides a photovoltaic power station intelligent inspection system based on drones, including: The acquisition module 201 is used to collect visible light and thermal imaging data through a drone equipped with a dual-light camera and synchronize the data with the device's geographic coordinates; An association module 202 is configured to generate a panoramic base map based on the visible light and thermal imaging data, and associate the device's geographic coordinates to build a spatialized topological database; Positioning module 203, used to integrate panoramic base map and topology database in GIS platform to perform hierarchical positioning of equipment; Inspection module 204, used to generate optimized routes in response to positioning instructions of equipment hierarchical positioning, control the UAV to perform dual-light coordinated fixed-point inspection tasks, and obtain inspection data; Diagnostic module 205, for fusing visible light and thermal imaging data with inspection data to perform cross-modal fault diagnosis and generate an environmentally adaptive decision report; The maintenance module 206 is used to build an equipment degradation model based on the historical fault feature data output by the cross-modal fault diagnosis, generate a maintenance strategy and provide feedback to optimize inspection tasks.

[0063] Figure 3 Schematic diagram of the structure of the electronic device provided in this embodiment.

[0064] like Figure 3 As shown, the electronic device may include: a processor (processor) 310, a communication interface (Communications Interface) 320, a memory (memory) 330 and a communication bus 340, wherein the processor 310, the communication interface 320, and the memory 330 communicate with each other through the communication bus 340. The processor 310 can call the logic instructions in the memory 330 to execute a drone-based intelligent inspection method for photovoltaic power stations, which includes: collecting visible light and thermal imaging data through a drone equipped with a dual-light camera, and synchronously binding the geographic coordinates of the equipment; generating a panoramic base map based on the splicing of the visible light and thermal imaging data, and building a spatial topological database by associating the geographic coordinates of the equipment; integrating the panoramic base map and the topological database on the GIS platform to perform hierarchical positioning of the equipment; generating an optimized route in response to the positioning instruction of the hierarchical positioning of the equipment, controlling the drone to perform dual-light coordinated fixed-point inspection tasks, and obtaining inspection data; fusing the visible light and thermal imaging data with the inspection data to perform cross-modal fault diagnosis and generate an environmental adaptive decision report; building an equipment degradation model based on the historical fault feature data output by the cross-modal fault diagnosis, generating a maintenance strategy and providing feedback to optimize the inspection task.

[0065] Furthermore, the logic instructions in the aforementioned memory 330 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product, stored in a storage medium, includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0066] On the other hand, the present invention also provides a computer program product, which includes a computer program, which can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the drone-based intelligent inspection method for photovoltaic power stations provided by the above methods, the method including: collecting visible light and thermal imaging data through a drone equipped with a dual-light camera, and synchronously binding the geographic coordinates of the equipment; generating a panoramic base map based on the splicing of the visible light and thermal imaging data, and constructing a spatialized topological database by associating the geographic coordinates of the equipment; integrating the panoramic base map and the topological database on the GIS platform to perform hierarchical positioning of the equipment; generating an optimized route in response to the positioning instruction of the hierarchical positioning of the equipment, controlling the drone to perform dual-light collaborative fixed-point inspection tasks, and obtaining inspection data; fusing the visible light and thermal imaging data with the inspection data to perform cross-modal fault diagnosis and generate an environmental adaptive decision report; constructing an equipment degradation model based on the historical fault feature data output by the cross-modal fault diagnosis, generating a maintenance strategy and providing feedback to optimize the inspection task.

[0067] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the drone-based intelligent inspection method for photovoltaic power stations provided by the above-mentioned methods, the method comprising: collecting visible light and thermal imaging data by a drone equipped with a dual-light camera, and synchronously binding the geographic coordinates of the equipment; generating a panoramic base map based on the splicing of the visible light and thermal imaging data, and constructing a spatialized topological database by associating the geographic coordinates of the equipment; integrating the panoramic base map and the topological database on a GIS platform to perform hierarchical positioning of the equipment; generating an optimized route in response to the positioning instruction of the hierarchical positioning of the equipment, controlling the drone to perform dual-light collaborative fixed-point inspection tasks, and obtaining inspection data; fusing the visible light and thermal imaging data with the inspection data, performing cross-modal fault diagnosis and generating an environmentally adaptive decision report; constructing an equipment degradation model based on the historical fault feature data output by the cross-modal fault diagnosis, generating a maintenance strategy and providing feedback to optimize the inspection task.

[0068] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0069] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0070] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for intelligent inspection of photovoltaic power stations based on drones, characterized in that: include: Visible light and thermal imaging data are collected by drones equipped with dual-light cameras, and the geographic coordinates of the devices are simultaneously bound; Generate a panoramic base map based on the visible light and thermal imaging data, and build a spatial topological database by associating the device geographic coordinates; Integrate the panoramic base map and topology database on the GIS platform to perform hierarchical positioning of equipment; In response to the positioning instructions of the hierarchical positioning of the equipment, an optimized route is generated, and the UAV is controlled to perform the dual-light coordinated fixed-point inspection task and obtain the inspection data; fusing the visible light and thermal imaging data with the inspection data to perform cross-modal fault diagnosis and generate an environmentally adaptive decision report; An equipment degradation model is constructed based on the historical fault feature data output by the cross-modal fault diagnosis, and a maintenance strategy is generated and feedback is provided to optimize the inspection task.

2. The method for intelligent inspection of photovoltaic power stations based on drones according to claim 1, characterized in that: The associating the device geographic coordinates includes: Parsing the geolocation metadata of aerial images as spatial reference; Based on the spatial reference, the device pixel position is mapped to the geographic coordinate system through a coordinate transformation algorithm, and a topological association relationship between the device unique code and the geographic coordinates is established.

3. The intelligent inspection method for photovoltaic power stations based on drones according to claim 1 is characterized in that: The device hierarchical positioning includes: Render the electrical connection topology in the GIS platform according to the three-level structure of array-string-module; In response to the device code input command, the target device and its branch path are highlighted.

4. The method for intelligent inspection of photovoltaic power stations based on drones according to claim 1, characterized in that: The generating of the optimized route in response to the positioning instruction of the hierarchical positioning of the device includes: When receiving manual path drawing instructions, it automatically generates obstacle avoidance flight trajectories based on the 3D obstacle model; In response to a branch coding input instruction, the topology database is called to generate a bionic flight path covering the branch.

5. The method for intelligent inspection of photovoltaic power stations based on drones according to claim 1, characterized in that: The cross-modal fault diagnosis includes: Performing spatiotemporal alignment processing on the visible light and thermal imaging data and the inspection data for the same device; When there is spatial overlap between visible light anomaly features and thermal imaging temperature anomaly and inspection anomaly data, the cross-modal verification mechanism is triggered to improve diagnostic confidence.

6. The intelligent inspection method for photovoltaic power stations based on drones according to claim 5 is characterized in that: The generating of the environment adaptive decision report includes: Dynamically calculate the alarm threshold based on the real-time ambient temperature and the historical temperature rise pattern of the equipment; Based on the alarm threshold, the associated equipment maintenance record library generates a differentiated maintenance plan as an environmental adaptive decision report.

7. The method for intelligent inspection of photovoltaic power stations based on drones according to claim 5, characterized in that: Also includes: Determining the hot spot properties at the corresponding positions of the spatial overlap; If the hot spot attribute indicates that the hot spot area is circular and located at the edge of the battery cell, it is determined that the diode has failed; If the hot spot attribute indicates that the hot spot is accompanied by shadow features in the visible light image, it is determined to be blocked by vegetation or dust accumulation.

8. The intelligent inspection method for photovoltaic power stations based on drones according to claim 1, characterized in that: The constructing of the equipment degradation model includes: Extract the historical temperature series of the equipment to train the time series prediction model; When the predicted value of the prediction model exceeds a safety threshold, a preventive maintenance work order is automatically generated and linked to a spare parts inventory system.

9. The intelligent inspection method for photovoltaic power stations based on drones according to claim 1, characterized in that: Also includes: Real-time analysis of video streams during flight to detect intruder targets; When an intrusion is detected, the sound and light alarm device will be activated and the trajectory will be recorded.

10. A photovoltaic power station intelligent inspection system based on drones, characterized in that: include: The acquisition module is used to collect visible light and thermal imaging data through a drone equipped with a dual-light camera and synchronize the data with the device's geographic coordinates. An association module, configured to generate a panoramic base map based on the visible light and thermal imaging data, and to associate the device geographic coordinates to construct a spatialized topological database; A positioning module is used to integrate the panoramic base map and the topological database on the GIS platform to perform hierarchical positioning of the device; An inspection module is used to generate an optimized route in response to the positioning instructions of the hierarchical positioning of the equipment, control the UAV to perform dual-light coordinated fixed-point inspection tasks, and obtain inspection data; A diagnostic module, configured to fuse the visible light and thermal imaging data with the inspection data to perform cross-modal fault diagnosis and generate an environmentally adaptive decision report; A maintenance module is used to build an equipment degradation model based on the historical fault feature data output by the cross-modal fault diagnosis, generate a maintenance strategy and provide feedback to optimize the inspection task.

Images