Method, system and image processing device for capturing and / or processing electroluminescent images and aerial vehicle

Abstract

Disclosed herein are methods (400) of capturing and processing electroluminescence (EL) images (1910) of a PV array (40). In the described embodiments, the method 400 includes controlling an aerial vehicle (20) to fly along a flight path to capture EL images (1910) of corresponding PV array sub-sections (512b) of the PV array (40), obtaining respective image quality parameters from at least some of the captured EL images, dynamically adjusting a flight speed of the aerial vehicle along the flight path for capturing the EL images (1910) of the PV array sub-sections (512b) based on the respective image quality parameters, extracting a plurality of frames (1500) of the PV array sub-sections (512b) from the EL images (1910); determining a reference frame of the PV array sub-sections (512b) having a highest image quality from the extracted frames (2100); performing image alignment of the extracted frames (2100) with the reference frame to generate image-aligned frames (2130), and processing the image-aligned frames (2130) to produce an enhanced image (2140) of the PV array sub-sections (512b) having a higher resolution than the reference frame. Also disclosed are systems, image processing devices, and aerial vehicles for which the methods are directed.

Description

Technical Field

[0001] This application relates to a method for capturing and processing electroluminescent (EL) images of a photovoltaic (PV) array, an air vehicle for capturing EL images, and an image processing apparatus for processing EL images. Background Technology

[0002] Solar panels have been found to be widely used globally. However, due to the high initial capital investment costs, solar panels installed in the field must operate correctly and efficiently for a period of time to ensure a return on investment. Therefore, maintaining the quality of solar panels installed in the field is important. Due to the large-scale deployment of solar panels in solar farms (or typically photovoltaic (PV) factories), and the remote deployment of solar panels (such as on house rooftops), it is often difficult to monitor the performance of individual solar panels. Various imaging techniques, such as visual, thermal (infrared), ultraviolet (UV) fluorescence, photoluminescence (PL), and electroluminescence (EL) imaging, can be used to detect defects in solar panels. For example, EL inspection is used for quality control during PV manufacturing.

[0003] For EL measurements, the PV modules of these solar panels are connected to a power source and placed under forward bias. The emitted near-infrared light is captured by a camera sensitive to the near-infrared band. EL imaging has also been used on a sampling basis for field inspection. One common method for field EL inspection is the use of a mobile trailer. In this method, a mobile trailer carrying a darkroom is deployed on-site. The PV modules are removed from their installation location for measurement in the darkroom inside the trailer. This method ensures that EL measurements of the PV modules are performed in a controlled environment. However, large-scale inspections using this method are time-consuming and impractical because the PV modules must be disassembled. Additionally, there is a risk of introducing defects during module handling.

[0004] Another method of EL inspection is to perform it at night using a camera with locked current control that is mounted on a tripod during the day. While this method does not require removing the PV module from its support frame, it is also time-consuming and labor-intensive. Furthermore, perspective and intensity distortion in the captured images may be due to limitations imposed by the camera mounted on the tripod.

[0005] Therefore, it is desirable to provide a solution to at least one of the problems mentioned in the prior art, and / or to provide a useful alternative to the public. Summary of the Invention

[0006] According to a first aspect, a method for processing an electroluminescent (EL) image of a PV array is provided. The method includes: (i) extracting a plurality of frames from the EL image of a PV array sub-segment of the PV array, the PV array sub-segment including one or more PV modules of the PV array; (ii) determining a reference frame of the PV array sub-segment having the highest image quality from the extracted frames; (iii) performing image alignment of the extracted frames with the reference frame to generate an image-aligned frame; and (iv) processing the image-aligned frame to produce an enhanced image of the PV array sub-segment having a higher resolution than the reference frame.

[0007] The described embodiments are capable of capturing low-resolution, monochrome images and videos in low-light conditions or in the absence of natural light (such as during nighttime), and can still produce enhanced resolution and noise-reduced images to identify defective PV modules. Generally, if averaging is used, the amount of noise in the processed image can be reduced by approximately the square root of the number of images averaged. As a result, enhanced images of higher image quality are typically achieved.

[0008] Specifically, extracting frames from an image may include determining the corresponding corner points of each PV module in the image, and constructing the corresponding frame for each PV module based on the identified corner points of each PV module.

[0009] In a specific embodiment, determining the corresponding corner point of each PV module in the image may include clustering the corresponding corner points of a specific PV module that repeats in different images, and calculating the corresponding average position for each cluster of the corresponding corner point.

[0010] Preferably, determining the reference frame with the highest image quality may include evaluating the image quality of each frame based on at least one of the frame's sharpness, signal-to-noise ratio, and integrity.

[0011] The method may also include arranging the extracted frames in a stacked arrangement before performing image alignment. Furthermore, the corresponding corner points of the PV modules can be stacked in a stacked arrangement, and performing image alignment may include aligning the corresponding corner point of each PV module in the extracted frames with the corresponding corner point of a PV module in a reference frame.

[0012] The process of processing an image-aligned frame may include grouping the image-aligned frames according to the PV modules in each frame, and performing image averaging on each group of image-aligned frames to obtain the corresponding enhanced frame for each PV module.

[0013] In addition, image averaging can be based on weighted image stacking averaging and / or deep convolutional neural network structures.

[0014] The method may further include: associating each augmented frame with a horizontal and a vertical index according to the position of each PV module in the PV array sub-segment, and arranging the augmented frames according to their horizontal and vertical indices to generate an augmented image of the PV array sub-segment.

[0015] The method may also include scaling the corresponding image intensity for each enhanced frame. Additionally, the method may include mapping the enhanced images of the PV array sub-segments onto a base map of the PV array sub-segments. The base map may include the geographical location of each PV module.

[0016] In a particular embodiment, mapping the enhanced image onto the base map may further include orienting the enhanced image to align PV array sub-segments in the enhanced image with PV array sub-segments in the base map.

[0017] Furthermore, each EL image in the PV array segment may include an image identifier, and orienting the augmented image may include: associating the image identifier of a particular EL image with each augmented frame, wherein the PV module of each augmented frame is characterized in the particular EL image; and determining the orientation of the augmented image based on the image identifier associated with each augmented frame.

[0018] Geographic location can include GPS coordinates.

[0019] Furthermore, these multiple frames can be coherent frames of PV array sub-segments.

[0020] According to a second aspect, an image processing apparatus is provided for processing an EL image of a PV array. The image processing apparatus includes: an image processor configured to extract a plurality of frames of a PV array sub-segment of the PV array from the EL image, the PV array sub-segment including one or more PV modules of the PV array; to determine a reference frame of the PV array sub-segment having the highest image quality from the extracted frames; to perform image alignment of the extracted frames with the reference frame to generate an image-aligned frame; and to process the image-aligned frame to produce an enhanced image of the PV array sub-segment having a higher resolution and a lower noise level than the reference frame.

[0021] The image processor can also be configured to extract frames from an image by determining the corresponding corner points of each PV module in the image and constructing a corresponding frame for each PV module based on the identified corner points of each PV module.

[0022] Preferably, the image processor may also be configured to determine the corresponding corner point of each PV module in the image by clustering the corresponding corner points of specific PV modules that may be repeated in different images, and calculating the corresponding average position for each cluster of the corresponding corner points.

[0023] The image processor can also be configured to determine the reference frame with the highest image quality by evaluating the image quality of each frame based on at least one of frame sharpness, signal-to-noise ratio, and integrity.

[0024] Additionally, the image processor can be configured to arrange the extracted frames in a stacked arrangement before performing image alignment.

[0025] Furthermore, the corner points of the PV modules can be stacked in a stacked arrangement, and the image processor can also be configured to perform image alignment by aligning the corner points of each PV module in the extracted frame with the corresponding corner points of the PV modules in the reference frame.

[0026] In a specific embodiment, the image processor may also be configured to group the image-aligned frames according to the PV modules in each frame, process the image-aligned frames, and perform image averaging on each group of image-aligned frames to obtain a corresponding enhanced frame for each PV module.

[0027] Furthermore, the image processor can be configured to perform image averaging based on weighted image stacking averaging and / or a deep convolutional neural network architecture.

[0028] The image processor can also be configured to associate each augmented frame with a horizontal and a vertical index according to the position of each PV module in the PV array sub-segment, and to arrange the augmented frames according to their horizontal and vertical indices to produce an augmented image of the PV array sub-segment.

[0029] The image processor can be configured to scale the corresponding image intensity for each enhanced frame.

[0030] The image processor can also be configured to map an enhanced image of a PV array sub-segment onto a base map of the PV array sub-segment. The base map may include the geographic location of each PV module. Additionally, the image processor can be configured to map the enhanced image onto the base map by orienting the enhanced image to align the PV array sub-segment in the enhanced image with the PV array sub-segment in the base map.

[0031] Furthermore, each EL image of the PV array sub-segment may include an image identifier, and the image processor may be configured to orient the augmented image by associating the image identifier of a particular EL image with each augmented frame and determining the orientation of the augmented image based on the image identifier associated with each augmented frame, wherein the PV module of each augmented frame is characterized in the particular EL image.

[0032] The geographic location may include GPS coordinates. Additionally, these multiple frames may be a coherent frame of PV array sub-segments.

[0033] According to a third aspect, a method for controlling the movement of an air vehicle having a camera for capturing EL images of a PV array is provided. The method includes controlling the air vehicle to fly along a flight path to capture EL images of corresponding PV array sub-segments, deriving corresponding image quality parameters from at least some of the captured EL images, and dynamically adjusting the air vehicle's flight speed along the flight path based on the corresponding image quality parameters to capture EL images of the PV array sub-segments.

[0034] Advantageously, by dynamically adjusting the flight speed of an air vehicle based on image quality parameters obtained from at least some of the captured EL images, the air vehicle is able to adjust its flight speed, thereby enabling it to capture EL images with higher image quality, such as better signal-to-noise ratio or image sharpness.

[0035] Image quality parameters may include the SNR scan factor and the motion blur scan factor. In a specific example, the SNR scan factor may depend on the target SNR, the measured SNR, and the estimated number of EL images captured, including specific PV modules comprising PV array sub-segments.

[0036] On the other hand, the motion blur scanning factor can be the ratio of the measured object deflection to the predefined maximum object deflection.

[0037] Preferably, dynamically adjusting the flight speed of an air vehicle along its flight path based on corresponding image quality parameters may include deriving a target flight speed based on the minimum of the SNR scan factor and the motion blur scan factor, and dynamically adjusting the current flight speed of the air vehicle to match the target flight speed.

[0038] Additionally, deriving the target flight speed may include applying the minimum of the SNR scan factor and the motion blur scan factor to the current flight speed of the air vehicle to derive the target scan speed, and if the target scan speed can be lower than the maximum flight speed of the air vehicle, then the target scan speed is selected as the target flight speed.

[0039] The method may further include: if the target scanning speed exceeds the maximum flight speed of the air vehicle, then selecting the maximum flight speed as the target flight speed.

[0040] The method may also include adjusting the target's flight speed based on user input factors.

[0041] Additionally, the method may include detecting EL signals emitted by one or more PV modules of the PV array before controlling the air vehicle to fly along the flight path to capture EL images of the corresponding PV array sub-segment of the PV array.

[0042] The method may further include: maneuvering the air vehicle to an initial position before detecting the EL signal, wherein the yaw axis of the air vehicle and the optical axis of the camera may be perpendicular to the ground.

[0043] In addition, the method may include navigating air vehicles to the location of the EL signal.

[0044] In a particular embodiment, detecting an EL signal emitted by one or more PV modules of the PV array may include rotating the air vehicle about the vehicle's yaw axis while increasing the optical axis angle of the camera until the EL signal is detectable.

[0045] The camera's optical axis angle can be increased from 0° to 70°. It is conceivable that the camera's optical axis angle can then be increased at a decreasing pitch rate. Preferably, the air vehicle can also rotate at a decreasing yaw rate.

[0046] The method may also include maneuvering the air vehicle to a predefined altitude before rotating it.

[0047] The method may also include aligning the camera's field of view (FOV) with a corresponding PV array sub-segment by: determining a corresponding key point of a reference PV module in the corresponding PV array sub-segment; deriving a target alignment point to be aligned with the corresponding PV array sub-segment from the corresponding key point of the camera's FOV; performing a perspective transformation to align the corresponding key point with the target alignment point; and manipulating the air vehicle based on the perspective transformation relative to the corresponding PV array sub-segment.

[0048] Furthermore, aligning the camera's FOV with the corresponding PV array sub-segment may also include maneuvering the air vehicle to an appropriate altitude, wherein the corresponding PV array sub-segment is located at a predefined size ratio within the camera's FOV at the appropriate altitude.

[0049] The corresponding PV array sub-segments can occupy 80% to 90% of the camera's FOV at a predefined size ratio.

[0050] Additionally, the method may include dynamically adjusting the camera's focus based on the measured image sharpness.

[0051] The air vehicle may also include a light source aligned with the optical axis of the camera, and the method may also include supplying power to the light source in addition to capturing EL images of the PV array.

[0052] According to a fourth aspect, an air vehicle is provided, including a camera for capturing EL images of a PV array, a propulsion device for actuating the movement of the air vehicle, and a controller communicatively coupled to the camera and the propulsion device. The controller is configured to control the air vehicle to fly along a flight path to capture EL images of corresponding PV array sub-segments, obtain corresponding image quality parameters from at least some of the captured EL images, and dynamically adjust the flight speed of the air vehicle along the flight path based on the corresponding image quality parameters for capturing EL images of the PV array sub-segments.

[0053] Image quality parameters may subsequently include an SNR scan factor and a motion blur scan factor. Preferably, the SNR scan factor may depend on the target SNR, the measured SNR, and the estimated number of captured EL images of a specific PV module, including PV array sub-segments.

[0054] On the other hand, the motion blur scanning factor can be the ratio of the measured object deflection to the predefined maximum object deflection.

[0055] Preferably, the controller can also be configured to dynamically adjust the air vehicle's flight speed along the flight path by deriving the target flight speed based on the minimum of the SNR scan factor and the motion blur scan factor, according to the corresponding image quality parameters, and dynamically adjusting the current flight speed of the air vehicle to match the target flight speed.

[0056] The controller can also be configured to derive the target scan velocity by applying the minimum of the SNR scan factor and the motion blur scan factor to the current flight speed of the air vehicle, and to select the target scan velocity as the target flight velocity if the target scan velocity can be lower than the maximum flight speed of the air vehicle.

[0057] Additionally, the controller can be configured to select the maximum flight speed as the target flight speed if the target scanning speed exceeds the maximum flight speed of the air vehicle.

[0058] The controller can also be configured to adjust the target's flight speed based on user input factors.

[0059] In addition, the controller can be configured to detect EL signals emitted by one or more PV modules from a PV array having an array axis and the one or more PV modules being aligned with the array axis and may include a flat surface; determine the array axis of the PV array; and control a camera to capture EL images of the PV array along the array axis while dynamically adjusting a propulsion device to align the camera's optical axis to a flat surface perpendicular to the one or more PV modules.

[0060] In addition, the controller can be configured to set the air vehicle to its initial position by dynamically adjusting the propulsion equipment to set the yaw axis of the air vehicle to be perpendicular to the ground and dynamically adjusting the optical axis of the camera to be perpendicular to the ground.

[0061] The controller can also be configured to dynamically adjust the propulsion equipment to navigate the air vehicle to the location of the EL signal.

[0062] The controller can also be configured to locate the EL signal by dynamically adjusting the propulsion device to rotate the air vehicle about the yaw axis of the air vehicle, while increasing the optical axis angle of the camera until the EL signal emitted by one or more PV modules of the PV array can be located.

[0063] The camera's optical axis angle can be increased from 0° to 70°. Specifically, the camera's optical axis angle can then be increased at a decreasing pitch rate. Preferably, the air vehicle can also rotate at a decreasing yaw rate.

[0064] The controller can also be configured to dynamically adjust the propulsion system to maneuver the air vehicle to a predefined altitude before rotating it.

[0065] Additionally, the controller can be configured to align the camera's field of view (FOV) with the corresponding PV array sub-segment by: determining the corresponding key point of a reference PV module in the corresponding PV array sub-segment; deriving the target alignment point to be aligned with the corresponding PV array sub-segment from the corresponding key point of the camera's FOV; performing a perspective transformation to align the corresponding key point with the target alignment point; and dynamically adjusting the propulsion equipment based on the perspective transformation to maneuver the air vehicle relative to the corresponding PV array sub-segment.

[0066] The controller can also be configured to dynamically adjust the propulsion equipment to maneuver the air vehicle to an appropriate altitude and align the camera's field of view (FOV) with the corresponding PV array sub-segment, wherein the corresponding PV array sub-segment can be located at a predefined size ratio within the camera's FOV at an appropriate altitude.

[0067] The corresponding PV array sub-segments can occupy 80% to 90% of the camera's FOV at a predefined size ratio.

[0068] The controller can also be configured to dynamically adjust the camera's focus based on the measured image sharpness.

[0069] In addition, the air vehicle may include a light source aligned with the optical axis of the camera, and the controller may be configured to supply power to the light source in addition to capturing EL images of the PV array.

[0070] According to a fifth aspect, a method is provided for obtaining an enhanced image of a PV array sub-segment from an EL image of a PV array sub-segment captured by an air vehicle equipped with a camera. The method includes: (i) controlling the air vehicle to fly along a flight path to capture an EL image of a corresponding PV array sub-segment of the PV array; deriving corresponding image quality parameters from at least some of the captured EL images; dynamically adjusting the flight speed of the air vehicle along the flight path based on the corresponding image quality parameters for capturing the EL image of the PV array sub-segment; (iv) extracting multiple frames of the PV array sub-segment from the EL images; (v) determining a reference frame of the PV array sub-segment with the highest image quality from the extracted frames; (vi) performing image alignment of the extracted frames with the reference frame to generate an image-aligned frame; and (vii) processing the image-aligned frame to produce an enhanced image of the PV array sub-segment with a higher resolution than the reference frame.

[0071] According to a sixth aspect, a system is provided for capturing and processing EL images of PV array sub-segments of a PV array. The system includes an air vehicle and an image processing device. The air vehicle includes: a camera for capturing EL images of the PV array, a propulsion device for actuating movement of the air vehicle, and a controller communicatively connected to the camera and the propulsion device. The air vehicle is configured to control its flight along a flight path to capture EL images of corresponding PV array sub-segments of the PV array, derive corresponding image quality parameters from at least some of the captured EL images, and dynamically adjust the flight speed of the air vehicle along the flight path based on the corresponding image quality parameters to capture EL images of the PV array sub-segments. The image processing apparatus includes: an image processor configured to extract multiple frames of a PV array sub-segment of a PV array from an EL image, the PV array sub-segment including one or more PV modules of the PV array; determine a reference frame with the highest image quality for the PV array sub-segment from the extracted frames; perform image alignment of the extracted frames with the reference frame to generate an image-aligned frame; and process the image-aligned frame to generate an enhanced image of the PV array sub-segment having a higher resolution than the reference frame.

[0072] Brief description of the attached figures

[0073] Exemplary embodiments will be described with reference to the accompanying drawings, wherein:

[0074] Figure 1 This is an exemplary setting for enabling a UAV to capture an EL image of a PV array;

[0075] Figure 2 This is a block diagram illustrating the system architecture of a system used to capture and process EL images. The system includes... Figure 1 UAV;

[0076] Figure 3 Is as Figure 2 The optical subsystem is part of the system;

[0077] Figure 4 It is by Figure 2 A flowchart of an exemplary method executed by the system for capturing and processing EL images;

[0078] Figure 5 Is as Figure 4 A schematic diagram illustrating the UAV performing the POINT function as part of an exemplary method;

[0079] Figure 6 Is as Figure 4 A schematic diagram of the first part of an exemplary method for a UAV to perform a FIND function;

[0080] Figure 7 It is a UAV execution Figure 6 A diagram illustrating the second part of the FIND function;

[0081] Figure 8 It is a description Figure 6 and 7 A diagram illustrating the process and results of the FIND function;

[0082] Figure 9 Is it as Figure 4 Before the ALIGN (alignment) function is performed as part of the exemplary method, the UVA hovers over the UAV perspective above the PV string, where the camera's FOV is not aligned;

[0083] Figure 10 Is it as Figure 4 After the ALIGN function is executed as part of the exemplary method, the UVA hovers over the UAV perspective view above the PV string, where the camera's FOV is aligned.

[0084] Figure 11 It comes from according to Figure 9 The FOV of the camera is not aligned with the EL image of the camera's perspective;

[0085] Figure 12 It comes from according to Figure 10 The FOV of the camera is the EL image of the camera's field of view;

[0086] Figure 13 It comes from according to Figure 11 A schematic diagram of the EL image of the camera's FOV being misaligned with the controller's viewpoint;

[0087] Figure 14 It comes from according to Figure 12 A schematic diagram of the EL image of the camera's FOV pointing towards the controller's field of view;

[0088] include Figure 15 A to 15F Figure 15 Is it as Figure 4 A series of six consecutive EL images of a PV string captured by a camera during the SCAN (scan) function, which is part of an exemplary method;

[0089] Figure 16 After performing image alignment on the PV string Figure 15 A schematic diagram of a series of EL images;

[0090] Figure 17 It shows point-to-point deflection. Figure 15 A schematic diagram of two consecutive EL images;

[0091] Figure 18A It is a line graph, which shows as Figure 4 As part of an exemplary method, the current flight speed of the UAV decreases over time when the SCAN function is performed;

[0092] Figure 18B It is a line graph, which shows as Figure 4 As part of an exemplary method, the current flight speed of the UAV increases over time when the SCAN function is performed;

[0093] include Figure 19 A, Figure 19 B Figure 19 C Figure 19 D and Figure 19 E's Figure 19 This is a schematic diagram, which shows as... Figure 4 The time elapses when the UAV performs the AUTO (automatic) function as part of the exemplary method;

[0094] Figure 20 Is Figure 19 A schematic diagram of the file structure for the stored EL image captured during the AUTO function;

[0095] include Figure 21A, 21B, and 21C Figure 21 Is as Figure 4 A schematic diagram of a frame extracted from three consecutive EL images during the FREEEZE (freeze) function, as part of an exemplary method;

[0096] Figure 22 As Figure 21 A schematic diagram of the image enhancement steps performed by the frame extracted as part of the FREEZE function;

[0097] Figure 23 From Figure 22 The pixel intensity histogram of the enhanced frame obtained from the image enhancement steps;

[0098] Figure 24 It is by Figures 21 to 23 A schematic diagram of the enhanced EL image generated by the FREEEZE function;

[0099] Figure 25A Is it as Figure 4 A schematic diagram of the enhanced EL image during the first part of the MAP (map) function, which is part of an exemplary method, wherein the PV modules are arranged in two rows and two columns;

[0100] Figure 25B yes Figure 25A A schematic diagram of an alternative, wherein the enhanced EL image has PV modules arranged in two rows and three columns;

[0101] Figure 25C yes Figure 25A A schematic diagram of an alternative, wherein the enhanced EL image has PV modules arranged in two rows and five columns;

[0102] Figure 26 Is Figure 25A A schematic diagram of the enhanced EL image mapped onto the background during the second part of the -C MAP function;

[0103] Figure 27 Is Figure 26 A schematic diagram of the UAV during the SCAN function, which is used to approximate the center position of the camera's FOV along the PV string, in the third part of the MAP function.

[0104] Figure 28A yes Figure 27 The fourth part of the MAP function is a schematic diagram of mapping an exemplary enhanced EL image onto an exemplary undermap using a string alignment method; and

[0105] Figure 28B It uses the module alignment method as Figure 28AA schematic diagram of an exemplary enhanced EL image mapped onto an exemplary base map, which is an alternative to the fourth part of the diagram.

[0106] Detailed description

[0107] The following description includes specific examples for illustrative purposes. Those skilled in the art will understand that variations and modifications of the specific examples are possible and within the scope of this disclosure. The accompanying drawings and the following description should not depart from the generality of the foregoing overview.

[0108] Figure 1 An exemplary EL inspection apparatus or setup 100 for capturing EL images of a PV array 10 mounted on a building roof is described. In this embodiment, the PV array 10 includes three PV strings 12. Each PV string 12 includes two rows of five PV modules 14. The PV strings 12 are arranged along the longitudinal axis of each row such that the PV array 10 has an array axis 10a extending along the longitudinal axis of the PV strings 12. The PV strings 12 are connected to a combiner box 16 that combines the electrical output of the PV strings. The combiner box 16 is connected to an inverter box (not shown), which is then connected to the power grid. The inverter box converts the combined electrical output from DC to AC before feeding it to the power grid. In this way, the power generated by the PV modules 14 is fed into the power grid. During EL inspection, the PV array 10 is disconnected from the power grid.

[0109] Setup 100 also includes a switcher box 32, which includes three channels 34. Each PV string 12 of the PV array 10 is connected to a corresponding channel 34 of the switcher box 32. Setup 100 also includes a power supply 36 connected to the switcher box 32. This power supply is configured to supply up to 1000 volts of power to each PV string 12 and a minimum current equal to 10% of the short-circuit current of the PV module 14. By selectively activating the channel 34, a field operator 30 selectively supplies current from the power supply 36 to the PV string 12, which places the PV string 12 under forward bias conditions. When under forward bias conditions, one or more PV modules 14 in the PV string 12 emit light, also known as electroluminescence (EL), and thus generate an EL signal that is detected by the optical subsystem of an aerial vehicle (e.g., an unmanned aerial vehicle (UAV) 20).

[0110] exist Figure 1During the process, worker 30 noticed that the normally functioning PV array 10 was generating less power than expected. Neither visual nor infrared inspection indicated the cause. After disconnecting the PV array 10 from the power grid and electrically connecting it to power supply 36 via switch box 32, worker 30 instructed assistant 22 to deploy UAV 20 to capture EL images of the PV array 12 for EL inspection. UAV 20 includes a main body 210, a propulsion system 230 attached to the main body 210 to allow flight of the UAV 20, and an optical subsystem 220 mounted to the main body 210 for capturing EL images.

[0111] Although an assistant 22 is provided in this embodiment, it should be clear that the worker 30 can deploy the UAV 20 without the assistance of the assistant 22. Additionally, it should be noted that multiple PV strings 12 can be connected to a single channel 34. For example, all three PV strings 12 of the PV array 10 can be connected to a single channel 34. In this case, all three PV strings 12 are simultaneously under forward bias and capture the EL image of the entire PV array 10. It is worth noting that in this case, the amount of current supplied by the power supply 36 is lower compared to when each channel 34 is connected to a corresponding PV string 12, but this does not affect the forward bias condition of the PV strings 12.

[0112] Furthermore, a larger PV array may include multiple combiner boxes 16, which are then connected to an inverter box (not shown). Alternatively, the PV array 10 may not include combiner boxes 16, but instead connect the PV strings 12 directly to the inverter box.

[0113] Preferably, 100% of the short-circuit current of the PV module 14 is supplied to each PV string 12. However, this is not necessary. For example, a current equal to 60% of the short-circuit current of the PV module 14 can be supplied to each PV string 12. Measurements of the same PV array sub-segment under multiple injection currents can be used to estimate the electrical characteristics of the PV module 14 and identify current-related defects.

[0114] Figure 2 The system architecture of the image acquisition and processing system 200 is explained. System 200 includes an unmanned aerial vehicle (UAV) 20 and an image processing device 260. In addition to an optical subsystem 220 and a propulsion device 230, the UAV 20 also includes an onboard processing subsystem 240 and a power supply 242 (e.g., a battery). The power supply 242 is connected to and supplies power to the optical subsystem 220, the propulsion device 230, and the onboard processing subsystem 240. The onboard processing subsystem 240 is communicatively coupled to the optical subsystem 220 and the propulsion device 230 and is configured to control the optical subsystem 220 and the propulsion device 230 to perform various functions.

[0115] First refer to Figure 3 The optical subsystem 220 is described. The optical subsystem 220 includes a camera 222 having an optical axis 222a, and in this embodiment, the camera 222 is operable for capturing monochrome video recording. The camera 222 is sensitive in the near-infrared and / or short-infrared (NIR, SWIR) EL bands and is adapted to capture EL images in such bands. The camera 222 includes a focusing lens 223, which is also adapted for the NIR / SWIR EL bands. The lens 223 (e.g., a motorized focusing lens, a pressure-controlled polymer lens, or a liquid lens) allows the onboard processing subsystem 240 to adjust the focus of the lens 223 depending on the distance of the lens 223 to the PV array 10. The camera 222 also includes a lens filter (not shown) for filtering any unwanted spectra.

[0116] The optical subsystem 220 also includes optical distance measurement devices (such as a light detection and ranging device (LiDAR) 224). The optical axis (224a) of the LiDAR is aligned with the optical axis 222a of the camera 222. The LiDAR 224 is operable to measure the distance between the optical subsystem 220 and the PV array 10.

[0117] The optical subsystem 220 also includes a focused light source (such as a laser 226). The optical axis (226a) of the laser is also aligned with the optical axis 222a of the camera 222. The laser 226 is arranged to emit light in the visible spectrum and has a beam divergence no larger than the field of view (FOV) of the camera, which minimizes optical interference from the laser. Furthermore, the laser 226 allows for low-power operation, emits narrow-band light, and creates a focused shape that is easily identifiable by the worker 30. The focused shape is asymmetrical, which advantageously allows the worker 30 to identify the position pointed to by the camera 222 and also to identify the rotation of the camera's FOV.

[0118] The optical subsystem 220 also includes a single-axis gimbal 228 for attaching the optical subsystem 220 to the main body 210 of the UAV 20. Figure 2 (As shown in the figure). The airborne processing subsystem 240 controls the gimbal to raise / lower the optical axis 222a of the camera 222 by one degree of freedom (i.e., pitch).

[0119] Alternatively, the optical subsystem 220 can be mounted to the main body 210 via a bi- or tri-axis gimbal to allow additional degrees of freedom (yaw, roll) for adjusting the optical axis 222a of the camera 222 and to provide enhanced FOV stability. Furthermore, the focused shape created by the laser 226 can be symmetrical. An LED can also be used instead of the laser 226. The focal point of the lens 223 can be adjustable, whether mechanically or electrically driven.

[0120] See Figure 2The propulsion device 230 is described next. The propulsion device 230 includes four sets of thrusters 232, each driven by a corresponding motor 234, to allow the UAV to fly and perform aerial maneuvers, such as about the yaw axis 210a of the air vehicle (see...). Figure 1 Rotation. The yaw axis 210a is a vertical axis passing through the middle part of the body 210 when the UAV 20 is upright.

[0121] The onboard processing subsystem 240 includes a controller 250 and a memory unit 252. The controller is configured to perform five functions (FOCUS, POINT, FIND, ALIGN, SCAN, AUTO) according to an instruction set stored in the memory unit 252. The controller 250 receives information from the photonic system 220, including the distance from the PV array 10 to the lidar 224 and the visual feed from the camera. Using the information received from the optical subsystem 220, the controller 250 is configured to operate the optical subsystem 220 and the propulsion device 230 to execute the POINT, FIND, FOCUS, ALIGN, SCAN, and AUTO algorithms. Once an EL image is captured, the UAV 20 returns to its base to transfer the EL image to the image processing device 260 for further processing.

[0122] Image processing device 260 is configured to perform the functions FREEZE and MAP. Image processing device 260 includes frame extraction module 270, image enhancement module 280, mapping module 290, and image processor 300. Image processing device 260 takes an EL image as input and outputs an enhanced EL image of PV array 10.

[0123] The operation of each component of an air vehicle is described in more detail in the following sections.

[0124] Figure 4 This is a block diagram of an exemplary method 400 for capturing and processing EL images by system 200. Where applicable, along with the corresponding... Figures 5 to 26 B describes an exemplary method 400 together. In this embodiment, an unmanned aerial vehicle 20 is deployed to perform an EL inspection of the PV array 40, preferably at night or under low natural light conditions. The PV array 40 is similar to the PV array 10, except that the PV array 40 includes more PV modules. Combiner boxes (not shown) from each group of the PV array 10 are connected to a switch box 532 controlled by a field worker 30.

[0125] At step 410 of method 400, controller 250 executes the POINT function. Figure 5This is a schematic diagram 500 of UAV 20 performing the POINT function. During UAV 20 deployment, controller 250 is configured to control power supply 250 to power laser 226. Worker 30 spots the area 530 illuminated by laser 226 and identifies the location pointed to by camera 222. The laser has a light intensity within safe limits (laser: Class 1 or Class 2), ensuring that field worker 30 will not suffer any eye damage even in the event of unintentional direct eye exposure to laser 226. Laser 226 is on throughout most of the operation of UAV 20. This allows worker 30 to quickly identify where camera 222 is pointing, especially when it is unclear which PV string is currently forward biased. Laser 226 is turned off before UAV 20 performs the SCAN function, so that laser 226 does not appear in the EL image captured by camera 222.

[0126] Worker 30 consults a string connection diagram, which informs worker 30 which PV string is forward biased based on the activated / open channel. In this embodiment, the string connection diagram contains an error and informs worker 30 that PV string 512a is forward biased for a specific channel. In reality, another PV string 512b is forward biased, and one or more PV modules 514b of PV string 512b emit EL signals. PV strings 512a and 512b are part of PV array 40 and are also referred to as PV array sub-segments 512a and 512b of PV array 40.

[0127] After activating a specific channel, worker 30 manually guides UAV20 along flight path 520 to PV string 512a. Worker 30 notices that PV string 512a is not emitting an EL signal and deduces an error in the string connection diagram. To determine the location of PV string 512b, which is under forward bias (i.e. emitting an EL signal), worker 30 instructs controller 250 to activate the FIND function.

[0128] It should be noted that worker 30 does not need to manually guide UAV20 to PV string 512a. Worker 30 can initiate the FIND function immediately after UAV20 deployment, thus eliminating the need for the POINT function. Alternatively, controller 250 can also be configured to automatically initiate the FIND function upon UAV20 deployment.

[0129] At step 420 of method 400, controller 250 performs the FIND function. Figure 6 This is a schematic diagram 600 showing the first part of the UAV 20 performing the FIND function. It is worth noting that the worker 30 and the lighting area 530 are... Figure 6Not explained in a (and subsequent figures). The first part of the FIND function involves setting UAV 20 to an initial position. Upon startup, controller 250 is configured to dynamically adjust propulsion device 220 to maneuver UAV 20 to an initial position where the UAV's yaw axis 210a is perpendicular to the ground. Controller 250 is also configured to dynamically adjust gimbal 228 so that the camera's optical axis 222a is also perpendicular to the ground, or in other words, the camera's field of view shows the area directly below UAV 20. In this position, optical axis 222a has an angle of 0°.

[0130] In addition, the controller 250 is also configured to dynamically adjust the propulsion device 220 to maneuver the UAV 20 (along the yaw axis 210a of the UAV) from the ground to a predetermined altitude 610.

[0131] Figure 7 This is a schematic diagram 700 illustrating the second part of the FIND function performed by UAV 20. The second part of the FIND function involves performing a sweep of the PV array 40 in a sweep path that begins in an area directly beneath UAV 20 and spirals outwards. To perform the sweep, controller 250 is configured to dynamically adjust propulsion device 220 to rotate UAV 20 about the UAV's yaw axis 210a (see arrow 710). Controller 250 is also configured to simultaneously increase the optical axis angle of camera 720. This causes the camera's FOV to move outwards from UAV 20. Combined with rotation 720, the camera's sweep path forms a spiral 810. This... Figure 8 Chinese commentary Figure 8 The diagram 800 illustrates the result of the second part of the FIND function.

[0132] For example, in the initial position, the unmanned aerial vehicle hovers at a height of 10m above the ground. If the camera's optical axis angle is moved from 0° to 70°, a radial area of ​​55m is observed (using the sine law). Camera 222 has a 60° field of view. This results in a 10m field of view for a view directly pointing to the ground. Therefore, three rotations are sufficient to cover the PV array 40 with a radius of 25m.

[0133] It is worth noting that the camera's optical axis angle increases with a decreasing pitch rate. From the perspective of gimbal 228, the gimbal's pitch rate decreases as the pitch angle increases. Since the sweep path 810 increases with 2π times the radius, the camera's FOV travels a distance five times greater in its final rotation (2π·5m≈31; 2π·25m≈157m). Therefore, the camera's rotation 710, or yaw rate, is adjusted to be five times lower in the final rotation.

[0134] The camera's yaw rate depends on the amount of motion blur acceptable within the frame during the exposure time. For a maximum yaw of 10 pixels during exposure, an exposure time of 7ms, and a horizontal sensor resolution of 640px, a yaw rate of 22m / s is possible. For an approximate travel distance of 282m for the entire spiral (the sum of three circles with radii of 5m, 15m, and 25m), the Find method can take a maximum of 13 seconds if no PV string under forward bias is detected. While the Find function is in progress, the controller 250 examines the image captured by the camera 222 for characteristics of the forward biased PV string 512b. The Find function stops when an EL signal is detected from the PV string 512b. Upon detection of EL, the controller 250 is configured to dynamically adjust the propulsion device 220 to maneuver the UAV 20 toward the EL signal. In this way, the EL signal is used as an optical marker to guide the UAV 20.

[0135] At step 430 of method 400, controller 250 performs the FOCUS function. Controller 250 receives information from lidar 224 regarding the distance from camera 222 to one or more PV modules in PV string 512b. Controller 250 is configured to dynamically adjust the camera focus based on the measured distance so that the distance between camera lens 223 and the focus matches the distance between camera lens 223 and the imaging object, thereby maintaining the focus of the camera lens.

[0136] At step 440 of method 400, controller 250 executes the ALIGN function, then regarding... Figures 9 to 14 Describe the ALIGN functionality. Figure 9 and Figure 10 These are perspective views 900 and 1000 of UAV 20 hovering above PV string 512b, where the camera's FOV is misaligned with and aligned with PV string 512b, respectively. Figure 11 and 12 These are EL images 1100 and 1200 from the viewpoint of camera 222, where the camera's FOV is misaligned and aligned, respectively. Figure 13 and 14 These are schematic diagrams of EL images 1100, 1200, 1300, and 1400 from the perspective of controller 250.

[0137] Reference Figure 9 The camera's FOV 910 is not aligned with PV string 512b. Before method 400 can proceed to the SCAN function, the camera's FOV 910 must be aligned with PV string 512b (e.g., ...). Figure 10 (As depicted in the text).

[0138] Controller 250 receives EL images 1100 from camera 222 (e.g., ... Figure 11(As depicted). The EL image 1100 includes PV string 512b (or a portion of PV string 512b), which appears brighter (higher light intensity) than the background 1110 (i.e., the ground) due to the EL signals emitted by one or more PV modules 514b. The controller 250 applies an algorithm to determine if the camera's FOV 910 is misaligned relative to the PV string 512b. This algorithm utilizes the intensity difference between the bright PV string 512b and the dark background 1110 to detect the position and orientation of the PV string 512b (or reference PV module 514b). Specifically, and referring to... Figure 13 The algorithm detects the edges 1310 of the PV string and derives keypoints 1320 (e.g., module corner points) from the edges 1310. The algorithm then determines a set of alignment points 1330 corresponding to each keypoint 1320, which minimizes misalignment and any angle or angular distortion. The alignment points 1330 are set on the upper horizontal indicator 1340 and the lower horizontal indicator 1340.

[0139] Furthermore, the algorithm determines the appropriate height of the UAV 20 relative to the PV string 512b, which places the PV string 512b within the camera's FOV 910 at a predefined size ratio. This predefined size ratio is set to maintain approximately 5%–10% space between the top and bottom of the PV string 512b and the image boundary 1410 to allow for tolerance of positional oscillations of the UAV 20. In other words, the PV string 512b occupies 80% to 90% of the camera's FOV at the predefined size ratio.

[0140] The algorithm then determines the perspective transformation to align keypoint 1320 with aligned point 1330 (e.g., ...). Figure 14 (As depicted in the image). Then, the controller 250 is configured to make appropriate adjustments to the gimbal 228 and the propulsion device 220 based on perspective transformation. (As shown in...) Figure 14 As can be seen, the camera's field of view (FOV) 910 is aligned with the PV string 512b, and the PV string 512b is at a predefined size ratio within the camera's FOV 910. Once aligned, the camera's optical axis is perpendicular to the flat surface of the PV string. Since the PV string includes one or more PV modules 514b, the flat surface of the PV string is formed by the flat surfaces of one or more PV modules. These flat surfaces of the one or more PV modules are defined as surfaces arranged to receive sunlight.

[0141] It is worth noting that the controller 250 is configured to perform a repeated ALIGN function while the SCAN function is in progress. This ensures that the camera's optical axis is perpendicular to the flat surface of the PV string when capturing EL images during the SCAN function. Advantageously, this minimizes perspective distortion and increases the image resolution of the EL images captured by the camera 222. Furthermore, this allows the camera 222 to capture EL images with more consistent focus on the EL image. Additionally, accurately capturing the EL intensity from each PV module 514b is important for analytical purposes.

[0142] It is worth noting that if the controller 250 does not detect the end 1210 of the PV string 512b in the EL image 1200, the controller 250 is configured to adjust the propulsion device 220 along the longitudinal axis 10a of the PV string (see [link]). Figure 1 Manipulate UAV 20 until the end 1210 of PV string 512b is in EL image 1200.

[0143] At step 450, controller 250 performs the following combination Figures 15 to 17 The SCAN function is described. Figure 15 include Figure 15 Figures A to 15F illustrate a series of six coherent EL frames (1500) of the PV string 512b, captured by camera 222 at different locations along the PV string 512b. Notably, the EL frames overlap, allowing the PV module 1510 to appear more than once. Figure 15 A to 15E.

[0144] When considering large PV facilities and the limited flight time of UAV 20, scan rate becomes a key parameter for determining system efficiency. Longer camera exposure times generally result in better image quality, i.e., a better signal-to-noise ratio (SNR). However, long camera exposure times coupled with fast scan rates lead to motion blur, which degrades image quality. On the other hand, short camera exposure times result in excessive noise in the EL images, especially when the injected current is low, which also degrades image quality.

[0145] Because PV module 1510 appeared Figure 15 A to Figure 15 In F, there are therefore five frames from PV module 1510 available for image averaging (again, in a later stage) (which will be extracted in a later stage). The SNR of image averaging 1520 roughly increases with the number of frames available for image averaging (n). frames The image noise of the averaged image increases by the square root of the averaged image noise. In other words, the image noise of the averaged image decreases as the number of frames used to create the averaged image noise decreases. The number of frames available for image averaging is calculated in real time by the controller 250, as it also depends on the scanning speed.

[0146] Furthermore, even if the ALIGN function is repeated during the SCAN function, the camera's FOV struggles to remain completely stable throughout the entire SCAN function. This is especially true considering the positional oscillations of the UAV 20 caused by external forces acting on it, such as wind. This is in Figure 16 The middle is obvious. Figure 16 Is executing Figure 15 A schematic diagram of Series 1500 after aligning the EL image in A-15F with the image in PV string 512b. Position oscillations of the UAV 20 are evident in the point-to-point offset between the EL images along PV string 512b. While position oscillations are mitigated (and largely corrected) by the ALIGN function, any calculations of the scan rate must take this deflection into account, as well as limiting noise and motion blur.

[0147] Noise Limitation:

[0148] Controller 250 is configured to perform optical flow analysis (e.g., the Lucas-Kanade method) during the SCAN function. For images captured by camera 222... Figure 15 A to Figure 15 For each frame in F, controller 250 calculates the key points in the current frame and compares the key points in the current frame with the key points in the previous frame to determine the length of the bias steering.

[0149] Figure 17 This shows two consecutive EL images of a PV string 512b with point-to-point deflection (i.e., Figure 15 Schematic diagram 1700 for D and 15E). Figure 15 The preceding EL image of E is shown as a dashed line, while the current EL image F is shown as a thick line. According to... Figure 15 The average deflection of keypoints detected in EL images D and 15E is used to calculate the deflection of object 1710 in image center 1720. The length of this deflection is called d. f2f .

[0150] Controller 250 calculates a line 1730 passing through the image center 1720 and at a deflection angle. Line 1730 intersects the image boundary 1740 at intersection points 1750 and 1760. The distance between intersection points 1750 and 1760 represents the distance of the object traveling through the image plane. Controller 250 then calculates the deflection angle d. f2f n is calculated as the ratio of the length to the distance between the intersection points 1750 and 1760. frames .

[0151] The impact of noise on image quality can be quantified using SNR:

[0152]

[0153] In this embodiment, the SNR is calculated as follows. The captured Otsu method is used to obtain the threshold (t) between the dark background 1110 and the PV string 512b (or a portion of the PV string 512b). Otsu This threshold appears brighter due to the EL signals emitted by one or more PV modules 514b. By comparing the threshold t... Otsu The intensity of all brighter pixels is averaged to obtain the 'signal' value. Before EL measurement, the 'noise' value is obtained by averaging the standard deviations of pixels from multiple images taken with similar or equivalent series of imaging parameters (e.g., exposure time, sensor temperature, and gain).

[0154] SNR-related scan rate factor (or simply SNR scan factor) f SNR Applied to the current scan speed to ensure an average image SNR of 1520 (SNR) frame Matching the target SNR using equation (2) target )

[0155]

[0156] For example, for laboratory measurements, SNR target Set to 45. Adjust the camera exposure time during the SCAN function to reduce the SNR. frame Maintain at 5 (minimum requirement for outdoor measurements). Controller 250 then estimates 25 EL images available for image averaging (n... frame =25). Based on equation (1), In other words, the current scan speed should be reduced to 56% of its current value. Essentially, reducing the scan speed increases the number of frames (n) available to create an average of 1520 images. frames ).

[0157] Limiting motion blur:

[0158] To avoid the effects of motion blur, the exposure of the frame (d) exp The object deflection during this period should also be lower than the predetermined maximum value (d). exp—max A value of 0.75 pixels per exposure time is recommended. According to equation (3), two frames (t) can be used. f2f ) and exposure time (t) exp The time difference between the frames will cause the frame to deflect (d) f2f ) Scaled to exposure time deflection (d exp ).

[0159]

[0160] As shown in equation (4), the motion blur correlation scanning factor (f blur ) equals the maximum object deflection d exp_max Deflection d from the current object exp The ratio.

[0161]

[0162] Scan rate factor (f) scan From two factors (f) SNR f blur The smallest of the following is obtained, as shown in equation (5):

[0163] f scan =min(f SNR f blur (5)

[0164] To ensure high EL image quality, the maximum set scan speed v is obtained using equation (6). quality :

[0165] v quality =min(v max f scan ·v cur (6)

[0166] Maximum set scan speed v quality The maximum scan speed can still be achieved with high EL image quality.

[0167] According to equation (6), by using the scanning speed factor f scan With current flight speed v cur The target scan velocity is calculated by multiplying the values. If the target scan velocity is lower than the UAV 20's maximum flight speed v... max Then the target scanning speed is selected as the maximum set scanning speed v. quality In other words, even if UAV-20 could move to its maximum flight speed v more quickly... max Because this reduces the image quality of the EL image, the maximum set scan speed v is... quality Set at maximum flight speed v max the following.

[0168] If the maximum set scan speed exceeds the UAV 20's maximum flight speed v max Then select the maximum flight speed v max As the maximum set scan speed v quality .

[0169] Target flight speed v target It is based on equation (7) from the maximum set scanning speed vquality and user input factor f user Obtained.

[0170] v target =v quality ·f user (7)

[0171] User input factor f user It is obtained from the deflection of a joystick remotely controlled by a worker (30), and ranges from 0% to 100%. At 100%, the target flight speed v target Simply put, it's the maximum set scan speed v quality .

[0172] Flight speed v of the target target Smoothing techniques are applied to minimize the bumpy movement of the UAV 20. In this embodiment, an exponential moving average is used to obtain the set speed v according to equation (8). set α is a smoothing factor in the range of 0 to 100%.

[0173] v set =(α·v cur )+((1-α)·v target (8)

[0174] Next, refer to Figure 18A and 18B Two exemplary embodiments describing the SCAN function, Figure 18A and 18B This shows the current flight speed v of UAV 20 according to the SCAN function. cur Line graphs 1800a and 1800b show the decrease and increase over time, respectively. For these two embodiments, the user input factor f... user Taken as 100%, and the maximum flight speed v of UAV 20 max The value is taken as 9 m / s.

[0175] See Figure 18A At time = 1s, and when UAV20 is flying at its current speed v = 6m / s... cur During movement, the controller 250 obtains a 50% scan rate factor f from equation (5). scan1 If the scan rate factor is below 100%, it indicates that UAV 20 is moving faster than it should be. The controller 250 then sets the target scan rate to 3 m / s. This is because the target scan rate is lower than the UAV 20's maximum flight speed. max According to equation (6), the target scanning speed is selected as the maximum set scanning speed v. quality .

[0176] It is worth noting that, due to the user input factor f user It is 100%, so according to equation (7), the maximum set scanning speed v quality Also the target flight speed v target .

[0177] Then, the controller 250 dynamically reduces the current flight speed of the air vehicle until the target flight speed v is reached. target The smoothing technique according to equation (8) is applied to minimize the turbulent movement of UAV 20, and this can be achieved at the current flight speed v of UAV 20. cur This can be seen in the smooth transition from 6 m / s (at time = 1 s) to 3 m / s (at time = 2 s).

[0178] At time = 2s, controller 250 obtains 100% of the scan rate factor f from equation (5). scan2 At this moment, the UAV20's current flight speed v cur With the maximum set scan speed v quality match.

[0179] See Figure 18B At time = 1s, and when UAV20 is flying at a current speed of 3m / s v cur During movement, the controller 250 obtains a 150% scan rate factor f from equation (5). scan1 If the scan rate factor is above 100%, this indicates that the UAV 20 can move 50% faster while still matching the desired image quality. The controller 250 then sets the target scan rate to 4.5 m / s.

[0180] Because the target scanning speed is lower than the UAV 20's maximum flight speed v max Therefore, according to equation (6), the target scanning speed is selected as the maximum set scanning speed v. quality Instead of choosing the maximum flight speed v max As the maximum set scan speed v quality .

[0181] Similarly, due to the user input factor f user It is 100%, so according to equation (7), the maximum set scanning speed v quality Also the target flight speed v target The smoothing technique based on equation (8) was also applied to minimize the turbulent movement of the UAV 20.

[0182] Then, the controller 250 dynamically increases the current flight speed v of the air vehicle. cur until the target flight speed v is reached targetAt time = 2s, controller 250 obtains 100% of the scan rate factor f from equation (5). scan2 At this moment, the current flight speed v of UAV 20 is... cur With the maximum set scan speed v quality match.

[0183] It is worth noting that the controller 250 continuously executes the SCAN function until the opposite end of the PV string 512b is detected. Once the opposite end of the PV string 512b is detected, the controller 250 terminates the SCAN function and the EL image is stored in the memory unit 252.

[0184] In step 460, controller 250 is configured to perform the AUTO function. Figure 19 (include Figure 19 (A to 19E) shows EL video images captured by UAV20 on PV array 40 (using... Figure 5 A schematic diagram of the passage of time (using PV array 40 as an example). PV array 40 is illustrated as comprising two rows 1810, 1820 of PV strings. Controller 250 is communicatively coupled to and capable of controlling power supply 36 and switch box 532 (via wireless connection), which in turn is electrically connected to PV array 40.

[0185] At t1, such as Figure 19 As explained in section A, UAV 20 begins at the end 1810a of the first row 1810 and captures an EL image of the first PV string 1812a in the scan direction 1830. Upon detecting the end 1814a of the PV string 1812a, controller 250 is configured to instruct switcher box 532 to close the current channel and open the next channel for the next PV string 1812b. Notably, the current injected into PV string 1812b is maintained at the same level as that of PV string 1812a. This process continues until controller 250 detects that it has reached the last PV string 1812f at the opposite end 1810b of the first row 1810 of the PV strings at t2, as... Figure 19 As shown in B.

[0186] In t3, such as Figure 19 As explained in section C, controller 250 is configured to instruct power supply 36 to reduce the current injected into PV string 1812f. The purpose of reducing the injected current is to capture a low-current EL image for comparison with a higher-current EL image. Then, controller 250 dynamically adjusts propulsion device 220 to move UAV 20 in scan direction 1840, which is opposite to scan direction 1830. UAV 20 moves along scan direction 1840 and captures EL images of the PV string in the first row 1810 at the lower injected current.

[0187] Figure 19 As explained in section D, at point t4, controller 250 detects that UAV 40 has reached the end 1810a of the first row 1810 of the PV string. At this time, controller 250 is configured to instruct switcher block 532 to close the current channel and open the next channel to place the first PV string 1820a of the second row 1820 under a positive bias. Controller 250 is also configured to perform a FIND function to locate the EL signal emitted by PV string 1820a. It is worth noting that PV string 1820a is within the camera's FOV during the FIND function, and controller 250 is able to locate PV string 1820a.

[0188] exist Figure 19 At t5 in E, controller 250 is configured to navigate UAV 20 to PV string 1820a. Once UAV 20 has reached PV string 1820a, controller 250 is configured to execute the SCAN function again. The EL image of the PV string in the second row 1820 is captured using a similar process to that used to capture the EL image of the PV string in the first row 1810 (as detailed in t1 to t4). Notably, controller 250 is configured to execute the ALIGN function sequentially throughout the SCAN function. The captured EL images are stored in memory unit 252.

[0189] Figure 20 This is a schematic diagram 1900 of the file structure of the stored EL image. The file structure includes the stored EL image 1910 and an additional block including additional metadata 1920. The metadata 1920 is divided into a header 1930 and a body 1940. The header 1922 includes information about the image correction methods applied to the EL image 1910 before saving it. The image correction methods include dark current subtraction, flat field correction, defective pixel replacement, and lens distortion removal. The body 1924 stores image-related data, including camera exposure time and gain, UAV geographic location, camera orientation (yaw, pitch, roll), injected current, voltage, and channel information.

[0190] Camera 222 captures / digitizes EL images 1910 with a bit depth greater than 8 bits (e.g., 14 or 16 bits). This allows for more precise resolution of the image intensity range within 255 luminance steps than a monochrome 8-bit sensor. To reduce file size, an 8-bit image-based image encoder is used. The upper and lower intensity ranges of each EL image 1910 are stored in metadata 1920. The upper and lower intensity ranges are obtained from the effective dynamic range of each EL image 1910 captured by camera 222. This range can be used to scale each 8-bit EL image to the corresponding lower and higher intensity ranges of the original higher-depth camera image.

[0191] Once the UAV 20 returns to its base, the stored EL images are then transmitted to the image processing device 260 for further processing.

[0192] At step 470, the image processing device 260 performs the following (see below) Figures 21 to 24 The FREEZE function is described below. For brevity, the FREEZE function is described with reference to processing an EL image that only includes a PV string of 512b. It should be understood that the FREEZE function can process each EL image in a similar manner.

[0193] The FREEZE function includes: (i) a frame extraction step performed by the frame extraction module 270; and (ii) an image enhancement step performed by the image enhancement module 280.

[0194] Figure 21 include Figure 21 Images A through 21C respectively illustrate three consecutive EL frame images 2000a, 2000b, and 2000c of the PV string 512b processed as part of the frame extraction step. The image processor 300 instructs the frame extraction module 270 to determine the corresponding corner point 2010 of each PV module 514b in the PV string 512b from all three EL frame images 2000a, 2000b, and 2000c. In each EL frame image 2000a, 2000b, and 2000c, the detected corner point of each PV module 514b is shown as a black dot 2012. It is worth noting that in this embodiment, the frame extraction module 270 failed to... Figure 21 A specific corner point 2014 of a specific PV module 2016 was detected in the first EL frame image 2000a in A.

[0195] exist Figure 21 All corner points detected in the first EL frame image 2000a of A in 2010 Figure 21 In the second EL frame image 2000b, B is visualized as an empty point 2020. It is worth noting that, compared to... Figure 21 Compared to the corresponding black dots 2012 in the first EL frame image 2000a of A, the empty dots 2030 are slightly shifted to the left. This is due to point-to-point offset correction. Once the local corner point density exceeds a certain threshold (in this embodiment, two detected corner points 2010 are sufficient), the image processor 300 controls the frame extraction module 270 to generate cluster points 2022. The location of the cluster points is obtained by averaging all the corner points 2010 used to generate the cluster points 2022. This advantageously reduces the spatial detection error of individual corner points.

[0196] See Figure 21C. Once cluster point 2022 moves outside the boundary of the EL frame image, cluster point 2022 is stored, as illustrated by cluster point 2030 in EL image 2000c. Once too many EL frame images no longer add corner points 2010 to cluster points, cluster points 2022 within the boundary of the EL frame image are discarded. As long as the ratio of EL frame images with added corner points 2010 to EL frame images without added corner points 2010 is higher than a certain threshold, cluster points 2022 within the boundary of the EL frame image are retained, such as cluster point 2032 in EL image 2000c. Then, the frame extraction module 270 meshes all cluster points 2022 to generate / construct rectangular quadrilaterals or frames. Each frame consists of cluster points 2022 contained in the corresponding PV module within each frame.

[0197] Figure 22 The frame 2100 extracted from the frame extraction step being processed in the image enhancement step is explained. The image processor 300 instructs the image enhancement module 280 to assign horizontal and vertical module indices 2110 to each frame 2100 according to its position in the PV string 512b. For example, the first frame associated with a particular PV module 2514 is located in the first row and first column of the PV string 512b and is assigned horizontal and vertical module indices [1,1].

[0198] The image processor 300 also controls the image enhancement module 280 to group the frames 2100 according to the PV modules 514b contained in each frame (or similarly, according to their module index 2110). Each frame 2100 includes four cluster points 2022 (labeled 'A' to 'D' respectively). The frames 2100 in each group are then arranged in a stacked arrangement (referred to as a stack) such that cluster points 2022 labeled with the same letters ('A'-'D') are stacked on top of each other. Figure 22 An exemplary stack 2120 with module index [1,1] is shown in the figure.

[0199] Image enhancement module 280 is also configured to discard regions 2124 that are not part of frame 2100.

[0200] Using the exemplary stack 2120 as an example, the image enhancement module 280 is also configured to determine a reference frame with the highest image quality from the frame 2100 within the exemplary stack 2120. Image quality is evaluated based on the sharpness, SNR, and integrity of the PV module 514b within the frame.

[0201] Image enhancement module 280 is also configured to perform image alignment of frame 2100. This is done using an image alignment algorithm, such as "parametric image alignment using enhanced correlation coefficients". Image enhancement module 280 aligns the remaining frames 2100 in the exemplary stack 2120 with the reference frame. Image alignment is performed by aligning the cluster point 2022 marked 'A' in the remaining frames with the corresponding cluster point 2022 marked 'A' in the reference frame to obtain an image-aligned frame 2130.

[0202] Image enhancement module 280 is also configured to perform image averaging on the image-aligned frame 2130 to obtain an enhanced frame 2140 for a specific PV module 2514. Image averaging is performed using super-resolution routines such as weighted image stacking averaging 2032 and / or a dedicated deep convolutional network structure 2034. The enhanced frame 2140 has a higher SNR (i.e., lower noise) and higher resolution (resolution improvement factor up to 3) than the reference frame.

[0203] The same process is repeated for the remaining stacks to obtain the corresponding augmented frames for the remaining PV modules 514b in the PV string 512b. The image processor 300 also controls the image augmentation module 280 to determine the corresponding corner point 2010 for each augmented frame and remove any remaining perspective distortion in the augmented frame.

[0204] Image processor 300 also controls image enhancement module 280 to arrange the enhanced frames according to the module index 2110 of the enhanced frames to produce enhanced EL images of PV strings. If the distances between PV modules 514b in PV string 512b are similar, a single enhanced EL image is produced. If the distance varies due to a large gap 2112 between two PV modules (indicating that one of the PV modules belongs to a separate PV string 512c), separate enhanced EL images are produced for the separate PV string 512c.

[0205] Image enhancement module 280 is also configured to scale the image intensity of each enhanced frame to reflect the intensity spectrum of the PV modules 514b from darkest to brightest in the PV string 512b. Since intensity scaling reduces the depth resolution of the PV module intensity range, the enhanced frames of each PV module 514b are stored together with the enhanced EL image of the PV string.

[0206] During image processing, image intensity is represented as a real number or floating-point value. When the resulting image is displayed visually, a brightness value between the darkest and brightest displayable values ​​must be assigned to the image intensity. To reduce the impact of pixels with extreme image (or pixel) intensity values, the brightness range is defined by the lowest pixel intensity bin (dashed line 2201a) containing the minimum number of pixels (formerly known as the effective dynamic range) and the highest pixel intensity bin (dashed line 2201b).

[0207] Figure 23 The pixel intensity histogram after enhancement frame 2140 is shown. Two peaks 2200a and 2200b can be seen, corresponding to the pixel intensity of the dark background and the bright EL signal, respectively. A brightness range between the lowest pixel intensity bin (dashed line 2201a) and the highest pixel intensity bin (dashed line 2201b) is defined to exclude pixel intensities not present in many pixels. Figure 23 In this process, the pixel intensity below the lowest pixel intensity bin (to the left of dashed line 2201a) is set to the minimum value (0), and the pixel intensity above the highest pixel intensity bin (to the right of dashed line 2201b) is set to the maximum value that can be saved according to the selected precision (2 for an 8-bit image). 8 -1 = 255). Image intensity between these two values ​​will be scaled between the minimum (0) and maximum (255) values ​​that define the brightness range.

[0208] Figure 24 The enhanced EL image 2200 of PV string 512b after the image enhancement step has been explained, and it is noted that the resolution of the enhanced EL image 2200 is better than that of the reference frame. It is worth noting that, independent of the scan direction 2210—that is, from left to right, from right to left, from top to bottom, or any other combination—the enhanced EL image 2200 is always aligned with the camera's yaw (i.e., optical axis 222a) because EL images 2000a, 2000b, and 2000c were captured using the camera's FOV 910, which is aligned with PV string 512b. For ease of reference, a black triangle 2220 is used to indicate the lower left corner of the enhanced EL image 2200, where the enhanced EL image 2200 is aligned with the camera's FOV 910.

[0209] Existing known methods can then be used to process the enhanced EL image 2200 to identify any defective PV modules in the PV string 512b based on EL imaging. Knowing the geographical location of the defective PV components can be helpful in identifying defective PV modules. For this purpose, a mapping module 290 can be used.

[0210] Image processor 300 controls mapping module 290 to perform the MAP function as illustrated in Figures 25-28. Mapping module 290 is configured to map the enhanced EL image 2200 of PV string 512b onto the base map of PV string 2410. To identify the location of PV module 514b (and therefore any defective PV modules) in PV string 512b from the enhanced EL image 2200, the following information is captured, processed, and stored during the SCAN function: frame-related timing geolocation (such as time, latitude, longitude, altitude) and camera orientation (e.g., yaw, pitch, roll). This information is stored in metadata 1920 for each captured EL image / video (see reference). Figure 20 ).

[0211] Image processor 300 controls mapping module 290 to map the enhanced image onto the base map by orienting the enhanced image to align a PV array sub-segment in the enhanced image with a PV array sub-segment in the base map. If the geographic location of the PV string 512b in the enhanced EL image 2200 is known, but its orientation (i.e., camera yaw) is unknown, there are four possible orientations (rotation 0°, 90°, 180°, 270°) relative to the black triangle 2220 to align the image with the PV string in the base layer. Commercial PV modules typically have a 4×8, 6×10, or 6×12 cell grid and are usually rectangular rather than square. In such cases, only two of the four orientations are reasonable: rotations of 0° and 180° relative to the black triangle 2220. This will refer to... Figure 25A , 25B A more detailed explanation of 25C is provided below. Figure 25A , 25B The 25C describes three enhanced EL images 2310, 2320, and 2330, where the PV modules are arranged in two rows, two columns, three columns, and five columns, respectively.

[0212] Figure 25A The enhanced EL image 2310 has the same number of rows and columns. Due to the shape of the PV module 2314, when the enhanced EL image 2310 is mapped onto the undermap of its PV string, two of the four orientations (rotated 90° and 270° relative to the black triangle 2220) will cause the enhanced EL image 2310 to be incorrectly tilted.

[0213] although Figure 25BThe enhanced EL image 2320 in the image has a square shape, but the enhanced EL image 2320 has a different number of PV modules 2324 in its rows and columns. In two of the four orientations (rotated 90° and 270° relative to the black triangle 2220), the number of PV modules 2324 in a row or column will not match the number of PV modules 2324 in the same row or column of its PV string in the base image.

[0214] Figure 25C The enhanced EL image 2330 in the image is depicted with two orientations (rotated 0° and 180° relative to the black triangle 2220). As can be seen, only these two orientations will result in a uniform, unbiased image that matches its basemap.

[0215] Figure 26 The enhanced EL image 2200, mapped onto the base map 2410 of the PV string 512b, is explained. Because the PV string 512b has a vertical axis (i.e....) Figure 1 The array axis 10a is used, and the camera 222 is arranged to capture EL images of the PV string 512b at a predefined size ratio, so that no single EL image captures the entire PV string 512b. Specifically, in this embodiment, 11 EL images of the PV string 512b are captured. Each EL image is associated with a unique image identifier 2420. In this embodiment, each EL image is numbered in ascending order from 1 to 11.

[0216] When a specific EL image is used for image averaging to generate an enhanced frame, the image processor 300 further instructs the mapping module 290 to associate the image identifier 2420 of the specific EL image with the enhanced frame. For example, the enhanced frame 2430 is associated with the unique numeric identifiers "2", "3", "4", "5", and "6". In other words, the corresponding frames extracted from the EL images "2", "3", "4", "5", and "6" are used for image averaging to generate the enhanced frame 2430.

[0217] The mapping module 290 is then able to determine the orientation of the enhanced image 2200 (from two available orientations: 0° and 180° rotation relative to the black triangle 2220) based on the image identifier 2420 associated with each enhanced frame. The enhanced frame associated with image identifier "11" represents the frame captured at the end of the SCAN function, while the enhanced frame associated with image identifier "1" represents the frame captured at the start of the SCAN function.

[0218] Other indicators for the position and orientation of the augmented EL image 2200 within the base map 2410 are discussed. The approximate center position 2440 of the camera's FOV 910 along the PV string 512b can be calculated based on the UAV's flight path 2450 (including the flight start point 2452 and flight end point 2454), flight altitude, and camera orientation.

[0219] Figure 27 This is a schematic diagram 2500 showing a side view of UAV 20 during a scan function at the center position 2440 of the FOV 10 used for approximating the camera along PV string 512b. The height 2510 of UAV 20 above the ground is d. z Camera 222 is aligned with the ground at a pitch angle α2520. The pitch angle 2520 is aligned with the tilt angle 2530 relative to the ground and the PV string 512b. The distance 2540 between camera 222 and PV string 512b is d. L The distance 2540 is easily available from lidar readings or from lidar 224 during FOCUS functionality. The horizontal distance 2550 between UAV 20 and PV string 512b is d xy Then, based on d using equation (3) L To calculate the horizontal distance 2550:

[0220] d xy，1 =d L ·cos(α) (9)

[0221] Or based on using equation (4) d z The height of PV string 512b above the ground is 2560d. PV To calculate the horizontal distance 2550:

[0222] d xy，2 =(d z -d PV )·tan(α) (10)

[0223] Equation (9) is better than equation (10) because d xy，2 The height (d) of the object to be imaged needs to be known or estimated. PV In addition, UAV20 estimates d based on air pressure. z This might be wrong for longer flight times and constantly changing weather.

[0224] refer to Figure 27The center position 2440 of the camera's FOV 910 is located within the PV string 512b, while the UAV's flight path 2450 is slightly lower. Utilizing high-quality position data and simple PV string interconnects (as is the case with PV string 512b), the mapping module 290 can accurately map the enhanced EL image 2200 onto the base map 2410 without additional input. If more complex PV string interconnects and / or low-quality position data are present, additional PV string input can be obtained by comparing measurement numbers, switch boxes, and information about which inverter box or combiner box 16 is connected to which switch box and when.

[0225] Figure 28A An exemplary enhanced EL image 2610 is described using a string alignment method mapped onto an exemplary base map 2620. Figure 28B An exemplary enhanced EL image 2630 is described using a module alignment method mapped onto an exemplary base map 2640.

[0226] Reference Figure 28A For the string alignment method, the corner points 2622 of the base image 2620 are known, and the enhanced EL image 2610 is roughly aligned to the top of the base image 2620. If the base image contains multiple PV strings below the enhanced EL image 2610, the PV array 2620 that shares the most overlapping area with the enhanced EL image 2610 is selected. Then, the four corner points 2612 of the enhanced EL image 2610 are affinely or perspectively aligned 2624 by the mapping module 290 with the corner points 2622 of the PV strings in the base image 2620 according to the most similar direction.

[0227] refer to Figure 28B In this embodiment, the number of PV modules in the enhanced EL image 2630 does not match the number of PV modules in the base image 2640. This occurs when the enhanced EL image is generated from only a portion of the scanned PV string or when the interconnection of PV modules within the PV string does not follow a regular pattern. In this case, the number of modules in each row and column of the PV string must be known, except for the corner point 2642 of the base image 2640. Furthermore, the PV modules in the enhanced EL image 2630 must be approximately aligned with the PV modules in the base image 2640. The image processor 300 then instructs the mapping module 290 to obtain an affine or perspective image transformation from the deflection of the corner point 2632 in the enhanced EL image 2630 to all corresponding corner points 2642 of the PV modules in the base image 2640. The image processor 300 also instructs the mapping module 290 to map the enhanced EL image 2630 onto the base image 2640 based on the maximum overlap between the corner points 2632 and 2642, and according to the most similar orientation.

[0228] Once the enhanced EL images 2200, 2610, and 2630 are mapped onto the base maps 2410, 2620, and 2640, the geographical location (such as GPS coordinates) information of the PV module defects identified in the enhanced EL images 2200, 2610, and 2630 can be easily identified from the base maps 2410, 2620, and 2640 for repair work and / or maintenance.

[0229] Advantageously, according to the described embodiment, the UAV 20 can capture low-resolution monochrome video in low-light conditions and generate images with enhanced resolution and improved quality to identify defective PV modules and estimate PV module power loss. Specifically, the onboard processing subsystem 240 is capable of autonomously navigating the UAV 20 and executing the exemplary method 400.

[0230] Furthermore, by processing information such as frame-related timing geolocation (e.g., time, latitude, longitude, altitude) and camera orientation (e.g., yaw, pitch, roll), the location of the PV string in a specific EL image can be reliably and accurately reproduced.

[0231] It should be noted that the various embodiments described herein should not be construed as limiting. For example, UAV 20 may also be equipped with ultrasonic equipment for additional distance measurements. Furthermore, camera 222 can capture still EL images of the PV string under forward bias or record video of the PV string. Additionally, a color sensor may be used instead of a monochrome sensor. Although the described embodiment uses "PV string" as an example, any other PV electrical connection can be used, and broadly speaking, this embodiment can be used with any PV array.

[0232] Other types of air transport, such as drones, can be used, not just UAVs.

[0233] While exemplary method 400 is described as including all eight functions: FOCUS, POINT, FIND, ALIGN, SCAN, AUTO, FREEZE, and MAP, it will be understood that system 200 can perform any number of functions in any reasonable order. For example, in an alternative embodiment, the onboard processing subsystem 240 may not perform the AUTO function because worker 30 may want better control over the PV string to be inspected. In this case, the worker manually controls switch box 32 and power supply 36 after the SCAN function completes, and accordingly initiates the FIND or SCAN function. Image processing device 260 can also perform the FREEZE function without the MAP function.

[0234] Furthermore, the FOCUS function can be performed throughout method 400, especially during the SCAN function, to ensure high-quality sharpness of the captured EL image. Alternatively, if the distance between UAV 20 and PV array 10 can be kept within a narrow range, the FOCUS function is not necessary at all. In such an embodiment, a fixed-focus lens without controlled focus adjustment can be used instead of focusing lens 223.

[0235] In an alternative embodiment, UAV 20 can remotely transmit the captured EL image to image processing device 260 without first returning to the base.

[0236] In addition, the predetermined maximum value (d) can be set. exp_max It is set to 1.5.

[0237] In another example, if the lens filter is arranged to filter out any optical interference from the laser, it may not be necessary to turn off the laser.

[0238] In another example, during the SCAN function, the predefined size ratio can also be set to maintain a space of about 15%-20% (or higher, such as 20%-25%) between the top and bottom of the PV string 512b and the image boundary 1410, to allow for greater tolerance to positional oscillations of the UAV 20, depending on the apparent instability of the UAV 20.

Claims

1. A method for processing electroluminescent EL images of a PV array, comprising: Multiple frames of a PV array sub-segment of the PV array are extracted from the EL image, the PV array sub-segment including one or more PV modules of the PV array; Determine the reference frame with the highest image quality for the PV array sub-segment from the extracted frame; The extracted frame is image-aligned with the reference frame to generate an image-aligned frame by performing the following: The extracted frame is arranged in a stacked arrangement, wherein the corresponding corner points of the PV modules are stacked in a stacked arrangement; and Align the corresponding corner point of each PV module in the extracted frame with the corresponding corner point of the PV module in the reference frame. as well as The image-aligned frame is processed to produce an enhanced image of the PV array sub-segment with a higher resolution than the reference frame.

2. The method as described in claim 1, characterized in that, Extracting the frame from the EL image includes Determine the corresponding corner point of each PV module in the EL image; and The corresponding framework for each PV module is constructed based on the corner points identified for each PV module.

3. The method as described in claim 2, characterized in that, Determining the corresponding corner point of each PV module in the EL image includes Clustering of corresponding corner points of a specific PV module that repeats in different images; and Calculate the average position for each cluster for the corresponding corner point.

4. The method as described in any of the preceding claims, characterized in that, Determining the reference frame with the highest image quality involves evaluating the image quality of each frame based on at least one of the frame's sharpness, signal-to-noise ratio, and integrity.

5. The method as described in claim 1, characterized in that, Processing the image-aligned frames includes grouping the image-aligned frames according to the PV modules in each frame; and performing image averaging on each group of image-aligned frames to obtain a corresponding enhanced frame for each PV module.

6. The method as described in claim 5, characterized in that, The image averaging is based on weighted image stacking averaging and / or a deep convolutional neural network structure.

7. The method as described in claim 5, characterized in that, It also includes associating each augmented frame with a horizontal and a vertical index according to the position of each PV module in the PV array sub-segment; and arranging the augmented frames according to their horizontal and vertical indices to generate the augmented image of the PV array sub-segment.

8. The method as described in claim 5, characterized in that, It also includes scaling the corresponding image intensity for each enhanced frame.

9. The method as described in claim 1, characterized in that, It also includes mapping the enhanced image of the PV array sub-segment onto a base map of the PV array sub-segment, the base map including the geographical location of each PV module.

10. The method as described in claim 9, characterized in that, Mapping the enhanced image onto the base map includes orienting the enhanced image to align the PV array sub-segments in the enhanced image with the PV array sub-segments in the base map.

11. The method as described in claim 10, characterized in that, Each EL image of the PV array sub-segment includes an image identifier, and orienting the enhanced image includes... The image identifier of a particular EL image is associated with its PV module, which is characterized for each enhanced frame in the particular EL image; as well as The orientation of the enhanced image is determined based on the image identifier associated with each enhanced frame.

12. The method as described in claim 9, characterized in that, The geographic location includes GPS coordinates.

13. The method as described in claim 1, characterized in that, The plurality of frames are the coherent frames of the PV array sub-segments.

14. An image processing apparatus for processing EL images of a PV array, comprising: The image processor is configured as Multiple frames of a PV array sub-segment of a PV array are extracted from the EL image, the PV array sub-segment including one or more PV modules of the PV array; Determine the reference frame with the highest image quality for the PV array sub-segment from the extracted frame; The extracted frame is image-aligned with the reference frame to generate an image-aligned frame by performing the following: The extracted frame is arranged in a stacked arrangement, wherein the corresponding corner points of the PV modules are stacked in a stacked arrangement; and Align the corresponding corner point of each PV module in the extracted frame with the corresponding corner point of the PV module in the reference frame. as well as The image-aligned frame is processed to produce an enhanced image of the PV array sub-segment with a higher resolution than the reference frame.

15. The image processing apparatus as claimed in claim 14, characterized in that, The image processor is also configured to extract the frame from the EL image in the following manner: Determine the corresponding corner point of each PV module in the EL image; and The corresponding framework for each PV module is constructed based on the corner points identified for each PV module.

16. The image processing apparatus as claimed in claim 15, characterized in that, The image processor is also configured to determine the corresponding corner point of each PV module in the EL image by: Clustering of corresponding corner points of a specific PV module that repeats in different images; and Calculate the average position for each cluster for the corresponding corner point.

17. The image processing apparatus as claimed in claim 14, characterized in that, The image processor is also configured to determine a reference frame with the highest image quality by evaluating the image quality of each frame based on at least one of the frame's sharpness, signal-to-noise ratio, and integrity.

18. The image processing apparatus as claimed in claim 14, characterized in that, The image processor is further configured to process the image-aligned frames by grouping the image-aligned frames according to the PV modules in each frame, and to perform image averaging on each group of image-aligned frames to obtain a corresponding enhanced frame for each PV module.

19. The image processing apparatus as claimed in claim 18, characterized in that, The image processor is also configured to perform image averaging based on weighted image stacking averaging and / or a deep convolutional neural network architecture.

20. The image processing apparatus as claimed in claim 18, characterized in that, The image processor is also configured to associate each enhanced frame with a horizontal and a vertical index according to the position of each PV module in the PV array sub-segment; The enhanced frame is arranged according to its horizontal and vertical indices to generate the enhanced image of the PV array sub-segments.

21. The image processing apparatus as claimed in claim 18, characterized in that, The image processor is also configured to scale the corresponding image intensity for each enhanced frame.

22. The image processing apparatus as claimed in claim 14, characterized in that, The image processor is also configured to map the enhanced image of the PV array sub-segment onto a base map of the PV array sub-segment, the base map including the geographic location of each PV module.

23. The image processing apparatus as claimed in claim 22, characterized in that, The image processor is further configured to map the enhanced image onto the base map by orienting the enhanced image so that the PV array sub-segments in the enhanced image are aligned with the PV array sub-segments in the base map.

24. The image processing apparatus as claimed in claim 23, characterized in that, Each EL image of the PV array sub-segment includes an image identifier, and the image processor is further configured to orient the enhanced image in the following manner: The image identifier of a particular EL image is associated with its PV module, which is characterized for each enhanced frame in the particular EL image; as well as The orientation of the enhanced image is determined based on the image identifier associated with each enhanced frame.

25. The image processing apparatus as claimed in claim 22, characterized in that, The geographic location includes GPS coordinates.

26. The image processing apparatus as claimed in claim 14, characterized in that, The plurality of frames are the coherent frames of the PV array sub-segments.

27. A method for controlling the movement of an air vehicle having a camera for capturing EL images of a PV array, the method comprising: The air vehicle is controlled to fly along the flight path to capture EL images of the corresponding PV array sub-segment of the PV array; Derive corresponding image quality parameters from at least some of the captured EL images, wherein the image quality parameters include SNR scan factor and motion blur scan factor; as well as The flight speed of the air vehicle along the flight path is dynamically adjusted based on the corresponding image quality parameters to capture EL images of the PV array sub-segment: The target flight velocity is derived based on the SNR scanning factor and the motion blur scanning factor; and The current flight speed of the air vehicle is dynamically adjusted to match the target flight speed.

28. The method as described in claim 27, characterized in that, The SNR scan factor depends on the target SNR, the measured SNR, and the estimated number of captured EL images of a specific PV module including the PV array sub-segment.

29. The method as described in claim 27 or 28, characterized in that, The motion blur scanning factor is the ratio of the measured object deflection to the predefined maximum object deflection.

30. The method as described in claim 27, characterized in that, The derived target flight speed includes The minimum of the SNR scan factor and the motion blur scan factor is applied to the current flight speed of the air vehicle to derive the target scan speed; as well as If the target scanning speed is below the maximum flight speed of the air vehicle, then the target scanning speed is selected as the target flight speed.

31. The method as described in claim 30, characterized in that, Also includes: If the target scanning speed exceeds the maximum flight speed of the air vehicle, then the maximum flight speed is selected as the target flight speed.

32. The method as described in claim 27, characterized in that, It also includes adjusting the target's flight speed based on user input factors.

33. The method as described in claim 27, characterized in that, Also includes: Before controlling the air vehicle to fly along the flight path to capture EL images of the corresponding PV array sub-segment of the PV array, EL signals emitted from one or more PV modules of the PV array are detected.

34. The method as described in claim 33, characterized in that, Also includes: Before detecting the EL signal, the air vehicle is maneuvered to an initial position, wherein the yaw axis of the air vehicle and the optical axis of the camera are perpendicular to the ground.

35. The method as described in claim 33, characterized in that, It also includes navigating the air vehicle to the location of the EL signal.

36. The method as described in claim 33, characterized in that, Detecting an EL signal emitted by one or more PV modules of the PV array involves rotating the air vehicle about its yaw axis while simultaneously increasing the optical axis angle of the camera until the EL signal is detected.

37. The method as described in claim 36, characterized in that, The optical axis angle of the camera is increased from 0° to 70°.

38. The method as described in claim 36, characterized in that, The optical axis angle of the camera increases at a decreasing pitch rate.

39. The method as described in claim 36, characterized in that, The air vehicle rotates at a decreasing yaw rate.

40. The method as described in claim 36, characterized in that, Also includes: Before rotating the air vehicle, maneuver the air vehicle to a predefined altitude.

41. The method as described in claim 27, characterized in that, Also includes: The camera's field of view (FOV) is aligned with the corresponding PV array sub-segment in the following manner: Determine the corresponding key points of the reference PV module in the corresponding PV array sub-segment; Derive the target alignment point to be aligned with the corresponding PV array sub-segment from the corresponding key points of the FOV for the camera. Perform a perspective transformation to align the corresponding key points with the target alignment points; as well as The aerial vehicle is manipulated to the corresponding PV array sub-segment based on the perspective transformation.

42. The method as described in claim 41, characterized in that, Aligning the camera's field of view (FOV) with the corresponding PV array sub-segment further includes maneuvering the air vehicle to an appropriate altitude, wherein the corresponding PV array sub-segment is within the camera's FOV at the appropriate altitude at a predefined size ratio.

43. The method as described in claim 42, characterized in that, The corresponding PV array sub-segment occupies 80% to 90% of the camera's FOV at the predefined size ratio.

44. The method as described in claim 27, characterized in that, It also includes dynamically adjusting the camera's focus based on the measured image sharpness.

45. The method as described in claim 27, characterized in that, The air vehicle also includes a light source aligned with the optical axis of the camera, and the method further includes supplying power to the light source in addition to capturing the EL image of the PV array.

46. ​​An air transport vehicle, comprising: A camera used to capture EL images of the PV array; Propulsion equipment for actuating the movement of the air vehicle; as well as The controller is communicatively coupled to the camera and the propulsion device, and is configured to: The air vehicle is controlled to fly along the flight path to capture EL images of the corresponding PV array sub-segment of the PV array; Derive corresponding image quality parameters from at least some of the captured EL images, wherein the image quality parameters include SNR scan factor and motion blur scan factor; as well as The flight speed of the air vehicle along the flight path is dynamically adjusted based on the corresponding image quality parameters to capture the EL image of the PV array sub-segment: The target flight velocity is derived based on the SNR scanning factor and the motion blur scanning factor; and The current flight speed of the air vehicle is dynamically adjusted to match the target flight speed.

47. The air transport vehicle as claimed in claim 46, characterized in that, The SNR scan factor depends on the target SNR, the measured SNR, and the estimated number of captured EL images of a specific PV module including the PV array sub-segment.

48. The air transport vehicle as claimed in claim 46 or 47, characterized in that, The motion blur scanning factor is the ratio of the measured object deflection to the predefined maximum object deflection.

49. The air transport vehicle as claimed in claim 46, characterized in that, The controller is also configured to derive the target flight speed in the following manner: The minimum of the SNR scan factor and the motion blur scan factor is applied to the current flight speed of the air vehicle to derive the target scan speed; as well as If the target scanning speed is below the maximum flight speed of the air vehicle, then the target scanning speed is selected as the target flight speed.

50. The air transport vehicle as described in claim 49, characterized in that, The controller is also configured to select the maximum flight speed as the target flight speed if the target scanning speed exceeds the maximum flight speed of the air vehicle.

51. The air transport vehicle as described in claim 46, characterized in that, The controller is also configured to adjust the target's flight speed based on user input factors.

52. The air transport vehicle as described in claim 46, characterized in that, The controller is also configured to: Detecting EL signals emitted by one or more PV modules from the PV array, the PV array having an array axis, and the one or more PV modules being aligned with the array axis and including flat surfaces: Determine the array axis of the PV array; as well as The camera is controlled to capture the EL image of the PV array along the array axis, while the propulsion device is dynamically adjusted to align the optical axis of the camera to a flat surface perpendicular to the one or more PV modules.

53. The air transport vehicle as described in claim 52, characterized in that, The controller is also configured to set the air vehicle to an initial position before locating the EL signal by: Dynamically adjust the propulsion device to set the yaw axis of the air vehicle perpendicular to the ground; and The optical axis of the camera is dynamically adjusted to be perpendicular to the ground.

54. The air transport vehicle as described in claim 52, characterized in that, The controller is also configured to dynamically adjust the propulsion device to navigate the air vehicle to the location of the EL signal.

55. The air transport vehicle as described in claim 52, characterized in that, The controller is also configured to locate the EL signal by dynamically adjusting the propulsion device to rotate the air vehicle about the yaw axis of the air vehicle, while increasing the optical axis angle of the camera until the EL signal emitted by the one or more PV modules of the PV array is located.

56. The air transport vehicle as described in claim 55, characterized in that, The optical axis angle of the camera is increased from 0° to 70°.

57. The air transport vehicle as claimed in claim 55, characterized in that, The optical axis angle of the camera increases at a decreasing pitch rate.

58. The air transport vehicle as described in claim 55, characterized in that, The air vehicle rotates at a decreasing yaw rate.

59. The air transport vehicle as described in claim 55, characterized in that, The controller is also configured to dynamically adjust the propulsion device to maneuver the air vehicle to a predefined altitude before rotating it.

60. The air transport vehicle as claimed in claim 52, characterized in that, The controller is also configured to align the camera's field of view (FOV) with the corresponding PV array sub-segment in the following manner: Determine the corresponding key points of the reference PV module in the corresponding PV array sub-segment; Derive the target alignment point to be aligned with the corresponding PV array sub-segment from the corresponding key points of the FOV for the camera. Perform a perspective transformation to align the corresponding key points with the target alignment points; as well as The propulsion equipment is dynamically adjusted to maneuver the air vehicle to the corresponding PV array sub-segment based on the perspective transformation.

61. The air transport vehicle as claimed in claim 60, characterized in that, The controller is also configured to align the camera's field of view (FOV) with the corresponding PV array sub-segment by dynamically adjusting the propulsion device to maneuver the air vehicle to an appropriate altitude, wherein the corresponding PV array sub-segment is within the camera's FOV at the appropriate altitude at a predefined size ratio.

62. The air transport vehicle as claimed in claim 61, characterized in that, The corresponding PV array sub-segment occupies 80% to 90% of the camera's FOV at the predefined size ratio.

63. The air transport vehicle as described in claim 52, characterized in that, The controller is also configured to dynamically adjust the camera's focus based on the measured image sharpness.

64. The air transport vehicle as described in claim 52, characterized in that, It also includes a light source aligned with the optical axis of the camera, and the controller is further configured to supply power to the light source except when capturing the EL image of the PV array.

65. A method for obtaining an enhanced image of a PV array sub-segment from an EL image of a PV array sub-segment captured by an aerial vehicle with a camera, the method comprising: The air vehicle is controlled to fly along the flight path to capture EL images of the corresponding PV array sub-segment of the PV array; Derive the corresponding image quality parameters from at least some of the captured EL images; The flight speed of the air vehicle along the flight path is dynamically adjusted based on the corresponding image quality parameters in order to capture the EL image of the PV array sub-segment; Multiple frames of the PV array sub-segment are extracted from the EL image, the PV array sub-segment including one or more PV modules of the PV array; Determine the reference frame with the highest image quality for the PV array sub-segment from the extracted frame; The extracted frame is image-aligned with the reference frame to generate an image-aligned frame by performing the following: The extracted frame is arranged in a stacked arrangement, wherein the corresponding corner points of the PV modules are stacked in a stacked arrangement; and Align the corresponding corner point of each PV module in the extracted frame with the corresponding corner point of the PV module in the reference frame. as well as The image-aligned frame is processed to produce an enhanced image of the PV array sub-segment with a higher resolution than the reference frame.

66. A system for capturing and processing EL images of PV array sub-segments of a PV array, comprising: Air transportation, including: A camera used to capture EL images of the PV array; Propulsion equipment for actuating the movement of the air vehicle; and The controller is communicatively coupled to the camera and the propulsion device, and is configured to: The air vehicle is controlled to fly along the flight path to capture EL images of the corresponding PV array sub-segment of the PV array; Derive the corresponding image quality parameters from at least some of the captured EL images; and The flight speed of the air vehicle along the flight path is dynamically adjusted based on the corresponding image quality parameters to capture the EL image of the PV array sub-segment; and Image processing equipment, including: The image processor is configured as follows: Multiple frames of a PV array sub-segment of a PV array are extracted from the EL image, the PV array sub-segment including one or more PV modules of the PV array; Determine the reference frame with the highest image quality for the PV array sub-segment from the extracted frame; The extracted frame is image-aligned with the reference frame to generate an image-aligned frame by performing the following: The extracted frame is arranged in a stacked arrangement, wherein the corresponding corner points of the PV modules are stacked in a stacked arrangement; and Align the corresponding corner point of each PV module in the extracted frame with the corresponding corner point of the PV module in the reference frame; and The image-aligned frame is processed to produce an enhanced image of the PV array sub-segment with a higher resolution than the reference frame.

Images