Photovoltaic panel angle anomaly monitoring method, device, equipment and storage medium

Abstract

The application discloses a photovoltaic panel angle anomaly monitoring method, device, equipment and storage medium, and belongs to the technical field of photovoltaic panel monitoring. The method comprises the following steps: detecting the pose information of a plurality of micro visual reference objects in an inspection image; performing distortion correction and view angle standardization processing on the inspection image based on the pose information to obtain a standard top view; determining a suspected abnormal area based on the standard top view; obtaining the relative area of each photovoltaic panel in the historical inspection period of the suspected abnormal area; determining the linear regression slope of the relative area of each photovoltaic panel; if the linear regression slope meets a preset determination condition, updating the relative area change threshold, the absolute area reference baseline and the relative area reference baseline; based on the updated data, calculating the absolute area change rate and the relative area change amount, and determining whether the suspected abnormal area has an angle anomaly based on the absolute area change rate and the relative area change amount. The application can improve the accuracy of photovoltaic panel angle anomaly monitoring.

Description

Technical Field

[0001] This application relates to the field of photovoltaic monitoring technology, and more specifically, to a method, device, equipment, and storage medium for monitoring abnormal angles of photovoltaic panels. Background Technology

[0002] With the large-scale application of photovoltaic power generation, monitoring technology for the condition of photovoltaic panels (such as tilt anomalies) based on UAV aerial images has become a research hotspot. Existing technologies typically use the comparison of the projected area changes of the same photovoltaic panel in images taken at different times to make judgments. However, such methods have the following inherent drawbacks: First, they heavily rely on repeated and precise shooting poses by the UAV, which is difficult to achieve in actual operations due to environmental influences. Image perspective distortion caused by pose differences can introduce huge measurement errors and generate a large number of false alarms. Second, they often select a single reference panel, and the identification error or failure of the reference panel itself can cause the entire system to fail, resulting in poor robustness. Finally, such methods usually perform uniform high-precision analysis on all photovoltaic panels, resulting in huge computational and storage costs and poor economic efficiency.

[0003] To overcome the aforementioned shortcomings, fixed visual reference objects (such as large-size QR code boards) deployed at photovoltaic (PV) power plants were introduced to correct image distortion, and a local relative relationship model between PV panels was constructed to improve robustness. While this approach is effective, it also introduces new engineering challenges: 1) The contradiction between deployment and maintenance: To improve correction accuracy and redundancy, multiple high-precision fixed reference objects need to be deployed. However, large reference objects are costly, easily obscured, or damaged, increasing installation and maintenance costs and system failure points; 2) The contradiction between model stability and adaptability: Local relationship models are usually static and cannot adapt to the slow changes in the local layout of PV arrays caused by maintenance, replacement, or foundation settlement, which may lead to false alarms in the long term; 3) The contradiction between accuracy / range and computational load: Performing high-precision image correction and feature comparison on every PV panel in the entire field incurs huge computational and storage costs, limiting the real-time performance or cost-effectiveness of monitoring. Summary of the Invention

[0004] The main objective of this application is to provide a method, device, equipment, and storage medium for monitoring photovoltaic panel angle anomalies, so as to solve the technical problems of false alarms and high monitoring costs of photovoltaic panel angle anomalies in the prior art, and to achieve the technical effects of improving the accuracy of photovoltaic panel angle anomaly monitoring and reducing monitoring costs.

[0005] To achieve the above objectives, the first aspect of this application proposes a method for monitoring photovoltaic panel angle anomalies, comprising: In response to the inspection request of the current inspection cycle, the drone is controlled to fly along a preset route to capture inspection images of the photovoltaic array with a distributed network of micro visual reference objects. Obtain the pose information of multiple miniature visual reference objects in the inspection image; Based on the pose information, the inspection image is subjected to distortion correction and viewpoint standardization processing to obtain a standard top view. Based on the standard top view, identify suspected abnormal areas; Obtain the relative area of ​​each photovoltaic panel in the suspected abnormal area for multiple historical inspection cycles; wherein, the relative area is the ratio of the projected area of ​​each photovoltaic panel to the average projected area of ​​all neighboring panels; The linear regression slope of the relative area of ​​each photovoltaic panel in the suspected abnormal area is determined based on the relative area of ​​the multiple historical inspection cycles. If the linear regression slope of the relative area of ​​at least one photovoltaic panel in the suspected abnormal area meets the preset judgment conditions, it is determined that the suspected abnormal area has experienced a slow trend change in the site structure, and the relative area change threshold, absolute area reference baseline and relative area reference baseline of all photovoltaic panels in the suspected abnormal area are updated. Based on the updated absolute area reference baseline and relative area reference baseline, calculate the absolute area change rate and relative area change amount for the current inspection cycle; If the absolute area change rate is greater than the preset absolute area change threshold, and the relative area change is greater than the updated relative area change threshold, then the suspected abnormal area is determined to have an angular abnormality.

[0006] In one possible implementation, updating the relative area change threshold, absolute area reference baseline, and relative area reference baseline for all photovoltaic panels within the suspected abnormal area includes: For any of the photovoltaic panels, the relative area change threshold is multiplied by a relaxation coefficient greater than 1 to obtain the updated relative area change threshold. The average projected area of ​​the most recent W inspection cycles is determined as the updated absolute area reference baseline; where W is a positive integer and W≥2; The average relative area of ​​the most recent M inspection cycles is determined as the updated relative area reference baseline; where M is a positive integer and M≥2.

[0007] In one possible implementation, the preset judgment condition is that the linear regression slope of the relative area of ​​at least one photovoltaic panel in the suspected abnormal area is greater than a preset trend threshold for C consecutive inspection cycles; where C is a positive integer and C≥2.

[0008] In one possible implementation, the distributed micro-visual reference network includes multiple micro-visual references disposed inside the outer side of the photovoltaic array frame and facing the inspection direction and in the gaps between adjacent photovoltaic panels.

[0009] In one possible implementation, determining the suspected abnormal region based on the standard top view includes: The standard top view is divided into multiple image blocks of fixed size; Each image block is input into a trained lightweight convolutional neural network, which outputs an anomaly probability score for each image block. The anomaly probability scores of all image blocks are stitched together according to their spatial location to generate an anomaly probability score map; The physical regions corresponding to image blocks in the anomaly probability score map that have an anomaly probability score greater than a preset probability threshold are marked as the suspected anomaly regions.

[0010] In one possible implementation, determining the suspected abnormal region based on the standard top view includes: Obtain health standard images; Spatial registration is performed on the standard top view and the health standard image to align the two images at the pixel level; Gaussian filtering is applied to the two spatially registered images to suppress noise, resulting in the filtered image. Calculate the difference map between the filtered images; The difference map is segmented by a threshold, and regions with a difference exceeding a preset threshold are identified as suspected abnormal regions.

[0011] In one possible implementation, the distortion correction and viewpoint normalization processing of the inspection image based on the pose information to obtain a standard top view includes: Extract at least four image pixel coordinates of all micro visual references in the inspection image; Based on at least four sets of corresponding image pixel coordinates and physical coordinates, the rotation matrix and translation vector of the UAV camera relative to the field station coordinate system are calculated by solving the PnP problem to obtain the pose information. Based on the pose information and camera calibration parameters, the inspection image is subjected to distortion correction to eliminate radial and tangential distortion of the lens, resulting in a distortion-free image. Based on the pose information and the geometric parameters of the plane where the photovoltaic array is located, a mapping relationship from image pixel coordinates to standard top-view physical coordinates is constructed, and the mapping relationship is a homography transformation matrix; Based on the mapping relationship, the distortion-reduced image is subjected to perspective transformation and resampling to convert the tilted shooting perspective into a vertical top view, thus obtaining the standard top view.

[0012] A second aspect of this application proposes a photovoltaic panel angle anomaly monitoring device, comprising: The image capture control module is used to respond to the inspection request of the current inspection cycle, control the drone to fly along the preset route, and capture inspection images of the photovoltaic array with a distributed network of micro visual reference objects. The pose acquisition module is used to acquire the pose information of multiple miniature visual reference objects in the inspection image; The image correction module is used to perform distortion correction and viewpoint standardization processing on the inspection image based on the pose information to obtain a standard top view. The suspected area determination module is used to determine suspected abnormal areas based on the standard top view. The relative area acquisition module is used to acquire the relative area of ​​each photovoltaic panel in the suspected abnormal area for multiple historical inspection cycles; wherein, the relative area is the ratio of the projected area of ​​each photovoltaic panel to the average projected area of ​​all neighboring panels. The slope determination module is used to determine the linear regression slope of the relative area of ​​each photovoltaic panel in the suspected abnormal area based on the relative area of ​​the multiple historical inspection cycles. The update module is used to determine that a slow trend change in the site structure has occurred in the suspected abnormal area if the linear regression slope of the relative area of ​​at least one photovoltaic panel in the suspected abnormal area meets the preset judgment conditions, and to update the relative area change threshold, absolute area reference baseline and relative area reference baseline of all photovoltaic panels in the suspected abnormal area. The calculation module is used to calculate the absolute area change rate and relative area change amount of the current inspection cycle based on the updated absolute area reference baseline and relative area reference baseline. An angle anomaly determination module is used to determine that the suspected abnormal area has an angle anomaly if the absolute area change rate is greater than a preset absolute area change threshold and the relative area change is greater than the updated relative area change threshold.

[0013] A third aspect of this application provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the photovoltaic panel angle anomaly monitoring method provided in the first aspect of this invention.

[0014] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the photovoltaic panel angle anomaly monitoring method provided in the first aspect of the present invention.

[0015] The technical solutions provided by the embodiments of this application may include the following beneficial effects: This invention deeply integrates three engineering improvements based on the TRIZ principle, and systematically solves the new contradictions arising from them while retaining the core advantages of "fixed reference correction" and "local relationship analysis". By segmenting a single, large visual reference object into multiple low-cost micro-visual reference objects, a redundant and highly available distributed micro-visual reference object network is formed. High-precision pose calculation and image correction can be completed by identifying any three or more non-collinear marker points in the network, greatly improving anti-occlusion capabilities and deployment and maintenance convenience. The local relationship model is transformed from static to dynamic and adjustable, continuously monitoring historical data on the geometric relationships (such as area ratio and relative position) between each photovoltaic panel and its neighboring panels. Through algorithms (such as time series analysis), the slow trend changes (such as foundation settlement) of the site structure are automatically identified and learned, and the fault judgment threshold or model parameters of the area are dynamically adjusted accordingly, thereby effectively distinguishing between "real sudden failures" and "slow environmental evolution". The whole-site monitoring tasks are layered by cycle and precision, and a lightweight convolutional neural network model is introduced as an intermediary. A rapid general survey is performed during the regular inspection cycle, and only the corrected images are used for rapid anomaly area localization. During the fine inspection cycle or for suspected anomaly areas, the original high-precision full-process analysis is performed again. This mechanism significantly reduces the conventional data processing load while ensuring monitoring coverage. Attached Figure Description

[0016] The accompanying drawings, which form part of this application, are used to provide a further understanding of the application and to make other features, objects, and advantages of the application more apparent. The illustrative embodiments and descriptions of this application are used to explain the application and do not constitute an undue limitation of the application. In the drawings: Figure 1 A flowchart of the photovoltaic panel angle anomaly monitoring method provided in this application; Figure 2 This is a schematic diagram of the photovoltaic panel angle anomaly monitoring device provided in this application; Figure 3 A schematic diagram of the electronic device provided in this application. Detailed Implementation

[0017] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0018] This invention provides a method for monitoring abnormal angles of photovoltaic panels. This method can be executed by a cloud or a ground station data center. The cloud refers to a remote server cluster with massive storage and GPU computing power for model training, big data analysis, and global scheduling. The ground station data center refers to the local edge server or industrial control room of the power plant, deployed in the substation / control building, responsible for receiving data transmitted back from UAVs, real-time inference, storing inspection files, and cooperating with the cloud.

[0019] Figure 1 This is a flowchart illustrating a photovoltaic panel angle anomaly monitoring method provided in this embodiment. Figure 1 As shown, the main process of this method is described below (steps S101 to S109): Step S101: In response to the inspection request of the current inspection cycle, control the UAV to fly along the preset route and take inspection images of the photovoltaic array with a distributed micro visual reference object network. Step S102: Obtain the pose information of multiple miniature visual reference objects in the inspection image; Step S103: Based on the pose information, the inspection image is subjected to distortion correction and viewpoint standardization processing to obtain a standard top view. Step S104: Identify suspected abnormal areas based on the standard top view; Step S105: Obtain the relative area of ​​each photovoltaic panel in the suspected abnormal area for multiple historical inspection cycles; wherein, the relative area is the ratio of the projected area of ​​each photovoltaic panel to the average projected area of ​​all neighboring panels. Step S106: Determine the linear regression slope of the relative area of ​​each photovoltaic panel in the suspected abnormal area based on the relative area of ​​multiple historical inspection cycles. Step S107: If the linear regression slope of the relative area of ​​at least one photovoltaic panel in the suspected abnormal area meets the preset judgment conditions, it is determined that the suspected abnormal area has experienced a slow trend change in the site structure, and the relative area change threshold, absolute area reference baseline and relative area reference baseline of all photovoltaic panels in the suspected abnormal area are updated. Step S108: Based on the updated absolute area reference baseline and relative area reference baseline, calculate the absolute area change rate and relative area change amount for the current inspection cycle. Step S109: If the absolute area change rate is greater than the preset absolute area change threshold (0.03) and the relative area change is greater than the updated relative area change threshold, then the suspected abnormal area is determined to have an angular abnormality.

[0020] In this embodiment, the distributed micro visual reference network includes multiple micro visual references set inside the outer side of the photovoltaic array frame and facing the inspection direction and the gaps between adjacent photovoltaic panels. This ensures that the tags are not blocked by the front panel and do not occupy any power generation area. Even if a single tag is partially blocked, the other tags can still maintain positioning continuity.

[0021] These miniature visual references are distributed in specific geometric arrangements (such as equilateral triangles, grids, etc.) to form a reference network covering the entire site. Compared to large fixed references, the resistance to local occlusion and contamination is improved by more than 80%; the multi-point redundancy design ensures that the system correction function remains available even if some marker points fail, thus improving the system's availability from a physical perspective.

[0022] After acquiring the inspection images, the image quality is evaluated, specifically by the number of valid marker point groups identified. and image gradient-based sharpness scoring If the image quality is lower than a preset quality threshold (e.g., ... or The system can provide real-time feedback via data link, prompting the flight control system to perform a go-around or adjust attitude, ensuring the quality of the raw data. Inspection images and identified marker information are compressed and then transmitted back to the data processing center.

[0023] In some optional embodiments, at least four image pixel coordinates of all micro visual references in the inspection image can be extracted using recognition methods such as feature point detection and matching based on OpenCV and feature detection and descriptor matching algorithms (e.g., ORB) in the OpenCV library. Based on at least four sets of corresponding image pixel coordinates and physical coordinates, the rotation matrix and translation vector of the UAV camera relative to the field coordinate system are calculated by solving the Perspective-n-Point (PnP) problem to obtain pose information. Based on the pose information and camera calibration parameters, distortion correction is performed on the inspection image to eliminate radial and tangential distortion of the lens, resulting in a distorted image. Based on the pose information and the geometric parameters of the plane where the photovoltaic array is located, a mapping relationship is constructed from the image pixel coordinates to the standard top-view physical coordinates (a two-dimensional world coordinate system established with the plane where the photovoltaic array is located as the reference under an ideal vertical top-view perspective), and the mapping relationship is a homography transformation matrix. Based on the mapping relationship, the distorted image is subjected to perspective transformation and resampling to convert the tilted shooting perspective into a vertical top-view perspective, resulting in a standard top view.

[0024] It should be noted that the above identification and correction methods can be directly executed by the airborne edge computing unit. However, when the computing power of the airborne edge computing unit is limited, the UAV only takes inspection images and transmits them back. All identification and correction calculations are completed in the cloud or ground station data center.

[0025] For step S104, a lightweight convolutional neural network can be used to identify suspected abnormal regions in the standard top view. Specifically, the standard top view is first divided into multiple fixed-size (e.g., 256×256 pixels) image blocks; then, each image block is input into the trained lightweight convolutional neural network, which outputs an abnormality probability score (0-1) for each image block; then, the abnormality probability scores of all image blocks are concatenated according to their spatial location to generate an abnormality probability score map; finally, the physical regions corresponding to the image blocks whose abnormality probability scores in the abnormality probability score map are greater than a preset probability threshold (e.g., 0.7) are marked as suspected abnormal regions.

[0026] In this embodiment, the lightweight convolutional neural network is a simplified version of the MobileNetV2 architecture.

[0027] It should be noted that when the width or height of the standard top view is not an integer multiple of the preset block size, zero-padding or mirror expansion of the edge areas is required to ensure that the standard top view can be completely segmented; and the pixel coordinates of each image block in the standard top view are recorded for subsequent spatial stitching of the score image.

[0028] In addition, connectivity analysis is performed on the marked suspected anomalous regions to merge spatially adjacent suspected anomalous blocks into continuous suspected anomalous regions.

[0029] For step S104, image difference combined with Gaussian filtering can also be used to identify suspected abnormal regions in the standard top view. Specifically, a healthy standard image is acquired; spatial registration is performed on the standard top view and the healthy standard image to align the two images at the pixel level; Gaussian filtering (used to suppress random noise during image acquisition and spurious differences caused by minor fluctuations in illumination) is applied to the spatially registered two images to suppress noise, resulting in a filtered image; a difference map between the filtered images is calculated; threshold segmentation is performed on the difference map, and regions with a difference exceeding a preset threshold are identified as suspected abnormal regions.

[0030] In this embodiment, the health standard image is obtained by controlling the UAV to perform high-precision calibration inspection during the system initialization phase, collecting health status images of each photovoltaic panel in the photovoltaic array, performing distortion correction and view standardization processing on the health status images, and then storing the health standard image in the standard image database in association with the identification information of the corresponding photovoltaic panel.

[0031] For spatial registration, based on the pose information provided by the distributed micro visual reference network, the standard top view and the health standard image of the current inspection cycle are mapped to a unified world coordinate system; the two mapped images are resampled so that corresponding pixels represent the same physical position; the registration error is calculated, and when the registration error exceeds the preset tolerance, re-registration is triggered or the registration failure area is marked.

[0032] In this embodiment, a difference metric is calculated pixel by pixel for the filtered standard top view and health standard images; the difference metric uses at least one of absolute difference, squared difference, or normalized cross-correlation; the difference map is a two-dimensional matrix with the same dimensions as the input image, and each element represents the degree of difference at the corresponding position.

[0033] For threshold segmentation, pixels exceeding the threshold in the difference map are marked as suspected abnormal pixels; connected component analysis is performed on suspected abnormal pixels to extract connected suspected regions; morphological filtering is performed on the extracted suspected regions to remove noise regions with an area smaller than the preset minimum area.

[0034] In step S105, for any photovoltaic panel, a circle is drawn with the geometric center of the photovoltaic panel as the center and a preset radius. Other photovoltaic panels whose center points fall within the circle constitute a neighboring panel.

[0035] Preferably, for the calculation of relative area, the average projected area can be the average projected area weighted by Gaussian weights.

[0036] In step S106, the slope of the linear regression for the relative area of ​​any photovoltaic panel can be calculated by analyzing the relative area of ​​the most recent W historical inspection cycles using time series analysis. To adjust the sliding window size, for example .

[0037] In some optional embodiments, the preset judgment condition is that the linear regression slope of the relative area of ​​at least one photovoltaic panel in the suspected abnormal area is greater than a preset trend threshold (e.g., 0.001 per cycle) for C consecutive inspection cycles; where C is a positive integer and C≥2, for example C=10.

[0038] When the preset judgment conditions are met, it can be determined that a slow trend change in the structure of the photovoltaic power station has occurred in the suspected abnormal area. The slow trend change in the structure of the photovoltaic power station refers to the phenomenon that the angle of the photovoltaic panels slowly and uniformly shifts due to the settlement of the photovoltaic power station support foundation, seasonal temperature deformation, long-term wind load creep, etc.

[0039] At this point, it is necessary to update the relative area change threshold, absolute area reference baseline, and relative area reference baseline for all photovoltaic panels within the suspected abnormal area. Optionally, for any photovoltaic panel, the relative area change threshold for that panel can be adaptively relaxed by multiplying the preset relative area change threshold (e.g., 0.015) by a relaxation coefficient greater than 1 (e.g., 1.5) to obtain the updated relative area change threshold.

[0040] The average projected area over the most recent W inspection cycles is determined as the updated absolute area reference baseline; where W is a positive integer and W≥2. The absolute area reference baseline is used as a reference standard value to determine whether the area of ​​the photovoltaic panel is abnormal, i.e., the projected area of ​​the photovoltaic panel under normal conditions.

[0041] The average relative area over the most recent M inspection cycles is determined as the updated relative area reference baseline; where M is a positive integer and M≥2. The relative area reference baseline refers to the reference value of the ratio of the photovoltaic panel area to the average area of ​​its Gaussian weighted neighborhood, used to eliminate systematic biases and detect relative anomalies.

[0042] It should be noted that the values ​​of W and M can be the same or different.

[0043] In this embodiment, the formula for calculating the absolute area change rate is: In the formula, Let be the absolute area change rate of the k-th photovoltaic panel during the current inspection cycle t. Let be the absolute area of ​​the k-th photovoltaic panel during the current inspection cycle t. This serves as the absolute area reference baseline for the k-th photovoltaic panel.

[0044] The formula for calculating the relative area change rate is: In the formula, Let be the relative area change rate of the k-th photovoltaic panel during the current inspection cycle t. Let be the relative area of ​​the k-th photovoltaic panel during the current inspection cycle t. Let be the relative area reference baseline for the k-th photovoltaic panel.

[0045] The adaptive local relational model in this embodiment enables the system to effectively distinguish between sudden faults and slow deformation of the site. Simulation tests show that it can reduce the long-term false alarm rate caused by slow environmental changes by about 60%, while maintaining high sensitivity to sudden faults (detection rate >95%). The regional processing mechanism reduces the image area that the system needs to perform high-precision calculations in routine inspections to 5%-20% of the entire site, shortens the overall data processing time by about 70%, and reduces storage space requirements by about 50%, significantly reducing computing power costs and energy consumption.

[0046] Based on the same inventive concept, embodiments of the present invention provide a photovoltaic panel angle anomaly monitoring device. Figure 2 This is a structural block diagram of a photovoltaic panel angle anomaly monitoring device 200 provided in an embodiment of the present invention. Figure 2 As shown, the photovoltaic panel angle anomaly monitoring device 200 mainly includes: The image capture control module 201 is used to respond to the inspection request of the current inspection cycle, control the UAV to fly along the preset route, and capture inspection images of the photovoltaic array with a distributed micro visual reference object network. The pose acquisition module 202 is used to acquire the pose information of multiple miniature visual reference objects in the inspection image; Image correction module 203 is used to perform distortion correction and viewpoint standardization processing on the inspection image based on pose information to obtain a standard top view; Suspected Area Determination Module 204 is used to determine suspected abnormal areas based on a standard top view. The relative area acquisition module 205 is used to acquire the relative area of ​​each photovoltaic panel in the suspected abnormal area for multiple historical inspection cycles; wherein, the relative area is the ratio of the projected area of ​​each photovoltaic panel to the average projected area of ​​all neighboring panels. The slope determination module 206 is used to determine the linear regression slope of the relative area of ​​each photovoltaic panel in the suspected abnormal area based on the relative area of ​​multiple historical inspection cycles. The update module 207 is used to determine that a slow trend change in the structure of the power station has occurred in the suspected abnormal area if the linear regression slope of the relative area of ​​at least one photovoltaic panel in the suspected abnormal area meets the preset judgment conditions, and to update the relative area change threshold, absolute area reference baseline and relative area reference baseline of all photovoltaic panels in the suspected abnormal area. The calculation module 208 is used to calculate the absolute area change rate and relative area change amount for the current inspection cycle based on the updated absolute area reference baseline and relative area reference baseline. The angle anomaly determination module 209 is used to determine that an angle anomaly has occurred in a suspected abnormal area if the absolute area change rate is greater than a preset absolute area change threshold and the relative area change is greater than the updated relative area change threshold.

[0047] As an optional implementation of this embodiment, the update module 207 is specifically used to, for any photovoltaic panel, multiply the relative area change threshold by a relaxation coefficient greater than 1 to obtain an updated relative area change threshold; determine the average projected area of ​​the most recent W inspection cycles as the updated absolute area reference baseline; where W is a positive integer and W≥2; and determine the average relative area of ​​the most recent M inspection cycles as the updated relative area reference baseline; where M is a positive integer and M≥2.

[0048] As an optional implementation of this embodiment, the preset judgment condition is that the linear regression slope of the relative area of ​​at least one photovoltaic panel in the suspected abnormal area is greater than the preset trend threshold for C consecutive inspection cycles; where C is a positive integer and C≥2.

[0049] As an optional implementation of this embodiment, the distributed micro visual reference network includes multiple micro visual references disposed inside the outer side of the photovoltaic array frame and facing the inspection direction and in the gaps between adjacent photovoltaic panels.

[0050] As an optional implementation of this embodiment, the suspected region determination module 204 is specifically used to divide the standard top view into multiple fixed-size image blocks; input each image block into a trained lightweight convolutional neural network, and output an anomaly probability score for each image block; stitch together the anomaly probability scores of all image blocks according to their spatial locations to generate an anomaly probability score map; and mark the physical regions corresponding to the image blocks whose anomaly probability scores in the anomaly probability score map are greater than a preset probability threshold as suspected anomaly regions.

[0051] As an optional implementation of this embodiment, the suspected region determination module 204 is specifically used to acquire a health standard image; perform spatial registration on the standard top view and the health standard image to align the two images at the pixel level; apply Gaussian filtering to suppress noise in the two spatially registered images to obtain a filtered image; calculate the difference map between the filtered images; perform threshold segmentation on the difference map, and determine the regions with a difference exceeding a preset threshold as suspected abnormal regions.

[0052] As an optional implementation of this embodiment, the image correction module 203 is specifically used to extract at least four image pixel coordinates of all micro visual references in the inspection image; based on at least four sets of corresponding image pixel coordinates and physical coordinates, by solving the PnP problem, calculate the rotation matrix and translation vector of the UAV camera relative to the station coordinate system to obtain pose information; perform distortion correction on the inspection image based on the pose information and camera calibration parameters to eliminate radial and tangential distortion of the lens and obtain a distorted image; construct a mapping relationship from image pixel coordinates to standard top-view physical coordinates based on the pose information and the geometric parameters of the plane where the photovoltaic array is located, the mapping relationship being a homography transformation matrix; perform perspective transformation and resampling on the distorted image based on the mapping relationship to convert the tilted shooting view to a vertical top view to obtain a standard top view.

[0053] The functional modules in the embodiments of this invention can be integrated together to form an independent unit, such as integrated into a processing unit, or each module can exist physically separately, or two or more modules can be integrated to form an independent unit. The integrated unit can be implemented in hardware or as a software functional unit. If the function is implemented as a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause an electronic device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory, random access memory, magnetic disks, or optical disks.

[0054] Various variations and specific examples of the methods provided in the embodiments of the present invention are also applicable to the photovoltaic panel angle anomaly monitoring device provided in this embodiment. Through the foregoing detailed description of the photovoltaic panel angle anomaly monitoring method, those skilled in the art can clearly understand the implementation method of the photovoltaic panel angle anomaly monitoring device in this embodiment. For the sake of brevity, it will not be described in detail here.

[0055] Figure 3 This is a structural block diagram of an electronic device 300 provided in an embodiment of the present invention. Figure 3 As shown, the electronic device 300 includes a memory 301, a processor 302, and a communication bus 303; the memory 301 and the processor 302 are connected through the communication bus 303.

[0056] The memory 301 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 301 may include a program storage area and a data storage area. The program storage area may store instructions for implementing an operating system, instructions for at least one function, and instructions for implementing the photovoltaic panel angle anomaly monitoring method provided in the above embodiments, etc. The data storage area may store data involved in the photovoltaic panel angle anomaly monitoring method provided in the above embodiments, etc.

[0057] Processor 302 may include one or more processing cores. Processor 302 executes instructions, programs, code sets, or instruction sets stored in memory 301, and calls data stored in memory 301 to perform various functions and process data as described in this application. Processor 302 may be at least one of the following: Application Specific Integrated Circuit (ASIC), Digital Signal Processor (DSP), Digital Signal Processing Device (DSPD), Programmable Logic Device (PLD), Field Programmable Gate Array (FPGA), Central Processing Unit (CPU), controller, microcontroller, and microprocessor. It is understood that for different devices, the electronic devices used to implement the functions of processor 302 may also be other types, and this embodiment of the invention does not specifically limit the specific devices used.

[0058] The communication bus 303 may include a path for transmitting information between the aforementioned components. The communication bus 303 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. The communication bus 303 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 3 The symbol is represented by only one double arrow, but this does not indicate that there is only one bus or one type of bus. Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of the present invention.

[0059] This invention also provides a computer-readable storage medium storing a computer program that can be loaded by a processor and executed as described in the above embodiments for monitoring photovoltaic panel angle anomalies.

[0060] In this embodiment, the computer-readable storage medium can be a tangible device that holds and stores instructions used by an instruction execution device. The computer-readable storage medium can be, but is not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any combination thereof. Specifically, the computer-readable storage medium can be a portable computer disk, a hard disk, a USB flash drive, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), staging random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory stick, floppy disk, optical disk, magnetic disk, mechanical encoding device, or any combination thereof.

[0061] The computer program in this embodiment includes functions for executing... Figure 1 The program code for the method shown may include instructions corresponding to the execution of the method steps provided in the above embodiments. The computer program may be downloaded from a computer-readable storage medium to various computing / processing devices, or downloaded via a network (e.g., the Internet, local area network, wide area network, and / or wireless network) to an external computer or external storage device. The computer program may be executed entirely on the user's computer as a standalone software package.

[0062] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0063] Obviously, those skilled in the art should understand that the various units or steps of this application described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby storing them in a storage device for execution by a computing device, or fabricating them separately as individual integrated circuit modules, or fabricating multiple modules or steps into a single integrated circuit module. Thus, this application is not limited to any particular combination of hardware and software.

[0064] It should be noted that the data used in this way can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, meaning that a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0065] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for monitoring abnormal angles of photovoltaic panels, characterized in that, include: In response to the inspection request of the current inspection cycle, the drone is controlled to fly along a preset route to capture inspection images of the photovoltaic array with a distributed network of micro visual reference objects. Obtain the pose information of multiple miniature visual reference objects in the inspection image; Based on the pose information, the inspection image is subjected to distortion correction and viewpoint standardization processing to obtain a standard top view. Based on the standard top view, identify suspected abnormal areas; Obtain the relative area of ​​each photovoltaic panel in the suspected abnormal area for multiple historical inspection cycles; wherein, the relative area is the ratio of the projected area of ​​each photovoltaic panel to the average projected area of ​​all neighboring panels; The linear regression slope of the relative area of ​​each photovoltaic panel in the suspected abnormal area is determined based on the relative area of ​​the multiple historical inspection cycles. If the linear regression slope of the relative area of ​​at least one photovoltaic panel in the suspected abnormal area meets the preset judgment conditions, it is determined that the suspected abnormal area has experienced a slow trend change in the site structure, and the relative area change threshold, absolute area reference baseline and relative area reference baseline of all photovoltaic panels in the suspected abnormal area are updated. Based on the updated absolute area reference baseline and relative area reference baseline, calculate the absolute area change rate and relative area change amount for the current inspection cycle; If the absolute area change rate is greater than the preset absolute area change threshold, and the relative area change is greater than the updated relative area change threshold, then the suspected abnormal area is determined to have an angular abnormality.

2. The method according to claim 1, characterized in that, The updating of the relative area change threshold, absolute area reference baseline, and relative area reference baseline for all photovoltaic panels within the suspected abnormal area includes: For any of the photovoltaic panels, the relative area change threshold is multiplied by a relaxation coefficient greater than 1 to obtain the updated relative area change threshold. The average projected area of ​​the most recent W inspection cycles is determined as the updated absolute area reference baseline; where W is a positive integer and W≥2; The average relative area of ​​the most recent M inspection cycles is determined as the updated relative area reference baseline; where M is a positive integer and M≥2.

3. The method according to claim 1 or 2, characterized in that, The preset judgment condition is that the linear regression slope of the relative area of ​​at least one photovoltaic panel in the suspected abnormal area is greater than the preset trend threshold for C consecutive inspection cycles; where C is a positive integer and C≥2.

4. The method according to claim 1 or 2, characterized in that, The distributed micro-visual reference network includes multiple micro-visual references located inside the outer side of the photovoltaic array frame, facing the inspection direction, and in the gaps between adjacent photovoltaic panels.

5. The method according to claim 1 or 2, characterized in that, The process of identifying suspected abnormal areas based on the standard top view includes: The standard top view is divided into multiple image blocks of fixed size; Each image block is input into a trained lightweight convolutional neural network, which outputs an anomaly probability score for each image block. The anomaly probability scores of all image blocks are stitched together according to their spatial location to generate an anomaly probability score map; The physical regions corresponding to image blocks in the anomaly probability score map that have an anomaly probability score greater than a preset probability threshold are marked as the suspected anomaly regions.

6. The method according to claim 1 or 2, characterized in that, The process of identifying suspected abnormal areas based on the standard top view includes: Obtain health standard images; Spatial registration is performed on the standard top view and the health standard image to align the two images at the pixel level; Gaussian filtering is applied to the two spatially registered images to suppress noise, resulting in the filtered image. Calculate the difference map between the filtered images; The difference map is segmented by a threshold, and regions with a difference exceeding a preset threshold are identified as suspected abnormal regions.

7. The method according to claim 1 or 2, characterized in that, The process of performing distortion correction and viewpoint normalization on the inspection image based on the pose information to obtain a standard top view includes: Extract at least four image pixel coordinates of all micro visual references in the inspection image; Based on at least four sets of corresponding image pixel coordinates and physical coordinates, the rotation matrix and translation vector of the UAV camera relative to the field station coordinate system are calculated by solving the PnP problem to obtain the pose information. Based on the pose information and camera calibration parameters, the inspection image is subjected to distortion correction to eliminate radial and tangential distortion of the lens, resulting in a distortion-free image. Based on the pose information and the geometric parameters of the plane where the photovoltaic array is located, a mapping relationship from image pixel coordinates to standard top-view physical coordinates is constructed, and the mapping relationship is a homography transformation matrix; Based on the mapping relationship, the distortion-reduced image is subjected to perspective transformation and resampling to convert the tilted shooting perspective into a vertical top view, thus obtaining the standard top view.

8. A photovoltaic panel angle anomaly monitoring device, characterized in that, include: The image capture control module is used to respond to the inspection request of the current inspection cycle, control the drone to fly along the preset route, and capture inspection images of the photovoltaic array with a distributed network of micro visual reference objects. The pose acquisition module is used to acquire the pose information of multiple miniature visual reference objects in the inspection image; The image correction module is used to perform distortion correction and viewpoint standardization processing on the inspection image based on the pose information to obtain a standard top view. The suspected area determination module is used to determine suspected abnormal areas based on the standard top view. The relative area acquisition module is used to acquire the relative area of ​​each photovoltaic panel in the suspected abnormal area for multiple historical inspection cycles; wherein, the relative area is the ratio of the projected area of ​​each photovoltaic panel to the average projected area of ​​all neighboring panels. The slope determination module is used to determine the linear regression slope of the relative area of ​​each photovoltaic panel in the suspected abnormal area based on the relative area of ​​the multiple historical inspection cycles. The update module is used to determine that a slow trend change in the site structure has occurred in the suspected abnormal area if the linear regression slope of the relative area of ​​at least one photovoltaic panel in the suspected abnormal area meets the preset judgment conditions, and to update the relative area change threshold, absolute area reference baseline and relative area reference baseline of all photovoltaic panels in the suspected abnormal area. The calculation module is used to calculate the absolute area change rate and relative area change amount of the current inspection cycle based on the updated absolute area reference baseline and relative area reference baseline. An angle anomaly determination module is used to determine that the suspected abnormal area has an angle anomaly if the absolute area change rate is greater than a preset absolute area change threshold and the relative area change is greater than the updated relative area change threshold.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the photovoltaic panel angle anomaly monitoring method as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the photovoltaic panel angle anomaly monitoring method according to any one of claims 1 to 7.

Images