A photovoltaic support intelligent inspection method and device based on unmanned aerial vehicle sensing

Abstract

The present application relates to the field of new energy photovoltaic inspection and operation technology, and discloses a photovoltaic support intelligent inspection method and device based on unmanned aerial vehicle sensing, wherein a benchmark digital twin model is constructed by fusing the multi-modal data of the unmanned aerial vehicle and the design parameters, a unified spatial benchmark and structural prior information are provided for the photovoltaic flexible support, and the problem of non-unified data benchmark in complex scenes is effectively solved. A complete spatial form model is constructed by relying on the exposed area features and key feature points of the steel strand, the limitation of the monitoring blind area caused by the photovoltaic panel shielding is broken through, and the precise restoration of the hidden section steel strand form is realized. The vibration monitoring data is input into the model for dynamic correction and state discrimination, the instantaneous sag, tension and vibration changes of the steel strand can be captured in real time, the structural state recognition accuracy is significantly improved, and the abnormal line section is quickly located. The unmanned aerial vehicle carries out targeted review and inspection on the abnormal line section, realizes the intelligent closed loop of alarm and verification, and greatly improves the inspection efficiency and fault diagnosis accuracy.

Description

Technical Field

[0001] This invention relates to the field of new energy photovoltaic inspection and maintenance technology, specifically to an intelligent inspection method and device for photovoltaic supports based on UAV perception. Background Technology

[0002] As the scale of photovoltaic power plant construction expands and construction scenarios become increasingly complex, flexible photovoltaic supports, as an innovative solution, can achieve spans of 30-100 meters or even larger through the synergistic effect of prestressed steel cables and support columns. This significantly improves land utilization and is particularly suitable for complex scenarios with large terrain undulations and poor site regularity, such as mountainous areas, hilly areas, and solar-fishery complementary projects. However, accurate monitoring of steel cable sag remains a technical pain point in the industry. Traditional methods rely on ground equipment such as total stations for single-point static measurements, with single-span measurements taking more than 3 hours and failing to capture high-frequency vibration deformation (0.5-2Hz) of the steel cables caused by wind loads. Actual measurement data from a mountainous photovoltaic project shows that the dynamic measurement error of the total station exceeds 5cm at a wind speed of 8m / s, making it difficult for maintenance personnel to detect safety hazards such as tensioner failure in a timely manner.

[0003] Existing drone technology is mainly used in topographic mapping and other fields, and its application in the inspection of photovoltaic flexible support projects is almost non-existent. The current inspection of photovoltaic flexible support projects faces the following difficulties: Existing drone inspection technology is only suitable for the inspection of fixed tilt supports, and cannot be used to inspect photovoltaic flexible support structure systems where the tilt angle is arranged according to the terrain. Summary of the Invention

[0004] This invention provides a method and device for intelligent inspection of photovoltaic supports based on UAV perception, which solves the problem that existing UAV inspection technology is only suitable for inspecting supports with fixed tilt angles and cannot be used for inspecting flexible photovoltaic support structures with tilt angles that vary with terrain.

[0005] In a first aspect, the present invention provides an intelligent inspection method for photovoltaic (PV) brackets based on UAV perception, applied to an intelligent inspection system for PV brackets, wherein the intelligent inspection system for PV brackets includes a UAV, and the method includes: The multimodal data of the photovoltaic flexible support collected by the UAV are obtained, and a benchmark digital twin model of the photovoltaic flexible support is constructed in combination with the design parameters of the photovoltaic flexible support. Based on the exposed area characteristics of the steel strand of the photovoltaic flexible support in the multimodal data and the coordinates of multiple key feature points of the benchmark digital twin model, the complete spatial morphology model of the steel strand is inverted and constructed. The real-time vibration monitoring data of the steel strand is input into the complete spatial morphology model to dynamically correct and determine the state of the complete morphology curve of the steel strand, and the abnormal segments of the steel strand are determined according to the determination results. The drone is used to perform targeted review and inspection of abnormal segments of the steel strand.

[0006] This invention constructs a baseline digital twin model by fusing multimodal data from unmanned aerial vehicles (UAVs) with design parameters, providing a unified spatial benchmark and structural prior information for flexible photovoltaic (PV) supports, effectively solving the problem of inconsistent data benchmarks in complex scenarios. Based on the characteristics of exposed areas and key feature points of the steel strands, a complete spatial morphology model is constructed, overcoming the monitoring blind spots caused by PV panel shading and achieving accurate restoration of the morphology of concealed steel strand sections. Vibration monitoring data is input into the model for dynamic correction and state discrimination, enabling real-time capture of instantaneous sag, tension, and vibration changes in the steel strands, significantly improving the accuracy of structural state identification and quickly locating abnormal segments. Targeted verification and inspection of abnormal segments are conducted using UAVs, achieving an intelligent closed loop of alarm and verification, greatly improving inspection efficiency and fault diagnosis accuracy.

[0007] In one optional implementation, acquiring the multimodal data of the photovoltaic flexible support collected by the UAV, and constructing a baseline digital twin model of the photovoltaic flexible support in conjunction with the design parameters of the photovoltaic flexible support, includes: The point cloud data and image data corresponding to the exposed area of ​​the steel strands of the photovoltaic flexible support and all columns are collected by the UAV. By combining the point cloud data and the image data, multimodal data of the photovoltaic flexible support is generated; A baseline digital twin model of the photovoltaic flexible support is constructed using the multimodal data and design parameters of the photovoltaic flexible support.

[0008] This invention utilizes drones to precisely collect point cloud and image data of the exposed steel strand areas and all columns of a photovoltaic flexible support system. The two types of data are then fused to form multimodal data, fully preserving structural geometric features and spatial morphological details, effectively overcoming the shortcomings of incomplete information from traditional single data sources. Based on the multimodal data and design parameters, a benchmark digital twin model is collaboratively constructed, establishing a unified and accurate spatial coordinate benchmark and structural topological relationship, providing a reliable foundation for subsequent morphological inversion, dynamic monitoring, and flight path planning. This invention eliminates the need for manual on-site measurements, improving data acquisition efficiency and spatial coverage integrity, and ensuring model accuracy and realism.

[0009] In one optional implementation, the complete spatial morphology model of the steel strand is inverted and constructed based on the exposed area characteristics of the photovoltaic flexible support in the multimodal data and the coordinates of multiple key feature points of the benchmark digital twin model, including: Identify the exposed area characteristics of the steel strands of the photovoltaic flexible support from the multimodal data; Based on the exposed area features, multiple key feature point coordinates are extracted from the benchmark digital twin model, and a preset number of non-collinear key feature point coordinates are randomly selected from all the key feature point coordinates to construct two vectors; Using all the aforementioned vectors, calculate the normal vector of the plane containing the catenary of the steel strand, and based on the normal vector, determine the plane equation of the plane containing the catenary; The coordinates of any of the key feature points are set as the origin of the local coordinate system. Based on the origin of the local coordinate system, the coordinates of two-dimensional points are determined, and the catenary equation is fitted using the coordinates of each of the two-dimensional points. Based on the coordinates of all the key feature points, solve the catenary equation, determine the geometric shape of the catenary equation, and input it into the benchmark digital twin model. Using the endpoint coordinates and tangent direction of the exposed area of ​​the steel strand as boundary conditions, the complete shape curve, maximum mid-span sag, and real-time tension of the steel strand are calculated through the reference digital twin model. The complete spatial shape of the steel strand is constructed by using its complete shape curve, maximum sag at mid-span, and real-time tension.

[0010] This invention identifies the features of exposed areas of steel strands and extracts the coordinates of key feature points. It then constructs vectors and solves the equations of the plane containing the catenary, achieving precise dimensionality reduction from three-dimensional space to a two-dimensional plane, effectively overcoming monitoring blind spots caused by photovoltaic panel shading. By establishing a local coordinate system using feature points and fitting the catenary equations, the spatial geometry of the steel strand can be quickly determined. After inputting the results into a benchmark digital twin model, and using the exposed end points and tangent directions as boundary conditions, the complete shape curve of the concealed area, the maximum sag at mid-span, and real-time tension are accurately inverted, ultimately constructing the complete spatial shape of the steel strand.

[0011] In one optional embodiment, the intelligent inspection system for photovoltaic supports further includes a vibration sensor. The step of inputting real-time vibration monitoring data of the steel strand into the complete spatial morphology model, dynamically correcting and judging the complete morphology curve of the steel strand, and determining abnormal segments of the steel strand based on the judgment result includes: The vibration sensor collects real-time vibration monitoring data of the steel strand. The real-time vibration monitoring data is input into the complete spatial morphology model, and the vibration acceleration, amplitude, and frequency of the real-time vibration monitoring data are used to dynamically correct the complete morphology curve. Based on the dynamically corrected shape curve, the instantaneous sag and real-time tension of the steel strand are calculated. The instantaneous sag, real-time tension, and vibration amplitude of the steel strand are compared with the corresponding preset thresholds. If the comparison result exceeds the threshold range, the position corresponding to the position exceeding the threshold range is determined to be an abnormal state. Based on the topological relationship of the benchmark digital twin model, the steel strand segment corresponding to the abnormal state is located, and the abnormal line segment of the steel strand is generated.

[0012] This invention collects real-time vibration data of steel strands using vibration sensors, inputting acceleration, amplitude, and frequency parameters into a complete spatial morphological model to dynamically correct the morphological curve, significantly improving the real-time performance and accuracy of structural condition characterization. Based on the corrected morphological curve, instantaneous sag and real-time tension can be accurately calculated and compared with preset thresholds for rapid identification of structural anomalies. Combined with the topological relationship of a baseline digital twin model, abnormal steel strand segments can be precisely located and abnormal line segments generated, enabling automatic fault identification under concealed and dynamic operating conditions. This invention significantly improves anomaly detection sensitivity and positioning accuracy, providing a reliable basis for subsequent targeted inspections and safety early warnings.

[0013] In one optional implementation, the step of performing targeted review and inspection of abnormal segments of the steel strand using the drone includes: Based on the early warning information corresponding to the abnormal segments of the steel strand, an adaptive inspection route for the UAV is generated. The adaptive inspection route and the early warning information are sent to the UAV, so that the UAV performs targeted review inspection according to the adaptive inspection route; When the drone flies to the area where the abnormal line segment corresponding to the warning information is located, it collects point cloud data and image data of the area where the abnormal line segment of the steel strand is located; Based on the point cloud data and image data of the area where the abnormal line segment is located, the abnormal line segment is verified on-site.

[0014] This invention generates adaptive inspection routes based on early warning information of abnormal steel strand segments, enabling UAVs to efficiently fly to the target area along the optimal path, avoiding interference from complex terrain and support structures, and improving flight safety and inspection efficiency. The UAV performs targeted verification inspections according to the route, accurately collecting point cloud and image data of the abnormal segment area, achieving close-range, multi-angle observation of the abnormal location. On-site verification of abnormal segments based on multimodal data can intuitively verify the structural status, confirm the authenticity of faults, eliminate false alarms, and form a complete closed loop of "early warning—verification—confirmation." This invention significantly improves the reliability of fault diagnosis and the speed of operation and maintenance response, providing a solid guarantee for the safe and stable operation of photovoltaic flexible supports.

[0015] In an optional implementation, the method further includes: Acquire real-time meteorological and environmental data to adjust the inspection time and route of the UAV; Based on the inspection time and inspection route of the UAV, real-time multimodal data of the photovoltaic flexible support collected by the UAV is obtained, and new multimodal data is generated. Based on the new multimodal data, the baseline digital twin model of the photovoltaic flexible support is updated to generate a new baseline digital twin model; Jump to the step of reversing and constructing a complete spatial morphology model of the steel strand based on the exposed area features of the steel strand of the photovoltaic flexible support in the multimodal data and the coordinates of multiple key feature points of the benchmark digital twin model.

[0016] This invention introduces real-time meteorological and environmental data to dynamically adjust the inspection time and flight path of drones, effectively avoiding interference factors such as strong winds and bright sunlight, and improving data acquisition stability and operational safety. Based on the adjusted real-time multimodal data, a new dataset is generated, continuously updating the baseline digital twin model to ensure it always reflects the actual condition of the support structure and terrain changes, guaranteeing the model's timeliness and accuracy. Through iterative inversion of the complete spatial morphology model of the steel strand, the monitoring system achieves self-updating and self-optimization, significantly improving long-term monitoring reliability.

[0017] Secondly, the present invention provides an intelligent inspection device for photovoltaic supports based on UAV perception, applied to an intelligent inspection system for photovoltaic supports, wherein the intelligent inspection system for photovoltaic supports includes a UAV, and the device includes: The acquisition module is used to acquire multimodal data of the photovoltaic flexible support collected by the UAV, and to construct a benchmark digital twin model of the photovoltaic flexible support in combination with the design parameters of the photovoltaic flexible support. The inversion module is used to invert and construct a complete spatial morphology model of the steel strand based on the exposed area features of the steel strand of the photovoltaic flexible support in the multimodal data and the coordinates of multiple key feature points of the benchmark digital twin model. The discrimination module is used to input the real-time vibration monitoring data of the steel strand into the complete spatial morphology model, dynamically correct and discriminate the complete morphology curve of the steel strand, and determine the abnormal segments of the steel strand based on the discrimination results. The inspection module is used to perform targeted review and inspection of abnormal segments of the steel strand using the drone.

[0018] Thirdly, the present invention provides an electronic device, comprising: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the above-described intelligent inspection method for photovoltaic brackets based on UAV perception, or any of its corresponding embodiments.

[0019] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to execute the intelligent inspection method for photovoltaic brackets based on UAV perception as described in the first aspect or any corresponding embodiment.

[0020] Fifthly, the present invention provides a computer program product, including computer instructions, which are used to cause the computer to execute the intelligent inspection method for photovoltaic brackets based on UAV perception as described in the first aspect or any corresponding embodiment. Attached Figure Description

[0021] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0022] Figure 1 This is a schematic diagram of the first process of the intelligent inspection method for photovoltaic brackets based on UAV perception according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the second process of the intelligent inspection method for photovoltaic brackets based on UAV perception according to an embodiment of the present invention; Figure 3 This is a schematic elevation view of a photovoltaic flexible support according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the installation of a vibration sensor according to an embodiment of the present invention; Figure 5 This is a structural block diagram of a photovoltaic bracket intelligent inspection device based on UAV perception according to an embodiment of the present invention; Figure 6 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present invention.

[0023] The attached figures are labeled as follows: 1. Photovoltaic module; 2. Steel strand below the module; 3. Inclined beam; 4. Intermediate column; 5. Intermediate column diagonal brace; 6. Fixing bolt; 7. Intermediate column support rod. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] It is understood that before using the technical solutions disclosed in the various embodiments of the present invention, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in the present invention and their authorization should be obtained in accordance with relevant laws and regulations through appropriate means.

[0026] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0027] This invention provides an intelligent inspection method for photovoltaic (PV) supports based on UAV (unmanned aerial vehicle) perception. By fusing multimodal data from UAVs with design parameters, a baseline digital twin model is constructed, providing a unified spatial benchmark and prior structural information for flexible PV supports, effectively solving the problem of inconsistent data benchmarks in complex scenarios. A complete spatial morphology model is constructed by inverting the characteristics of exposed areas and key feature points of the steel strands, overcoming the monitoring blind spots caused by PV panel shading and achieving accurate restoration of the morphology of concealed steel strand sections. Vibration monitoring data is input into the model for dynamic correction and state discrimination, enabling real-time capture of instantaneous changes in the sag, tension, and vibration of the steel strands, significantly improving the accuracy of structural state identification and quickly locating abnormal segments. Targeted verification inspections of abnormal segments are conducted using UAVs, achieving an intelligent closed loop of alarm and verification, greatly improving inspection efficiency and fault diagnosis accuracy.

[0028] According to an embodiment of the present invention, an embodiment of an intelligent inspection method for photovoltaic brackets based on UAV perception is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0029] This embodiment provides a method for intelligent inspection of photovoltaic (PV) brackets based on UAV perception, applied to an intelligent inspection system for PV brackets. The intelligent inspection system for PV brackets includes a UAV. Figure 1This is a flowchart of an intelligent inspection method for photovoltaic brackets based on UAV perception according to an embodiment of the present invention, as follows: Figure 1 As shown, the process includes the following steps: Step S101: Obtain multimodal data of photovoltaic flexible support collected by UAV, and construct a benchmark digital twin model of photovoltaic flexible support by combining the design parameters of photovoltaic flexible support.

[0030] It should be noted that multimodal data refers to the collective point cloud data, image data, and spatial pose data simultaneously collected by drones equipped with lidar, visible light cameras, etc.; design parameters refer to the original engineering design parameters of photovoltaic flexible supports, such as span, pretension, steel strand properties, column position, and structural topology; and the benchmark digital twin model refers to a digital benchmark model of photovoltaic flexible supports that contains geometric, physical, and topological information, built based on multimodal data and design parameters, providing a unified spatial benchmark and structural reference for the entire process.

[0031] In this embodiment of the invention, a miniature lidar and a high-resolution visible light camera mounted on a drone platform are used to synchronously collect high-precision 3D point cloud and 2D orthophotos of the exposed steel strands (≥250mm gap) and all columns during flight via a hardware synchronous triggering mechanism. Using the collected multimodal data, combined with the initial design parameters of the flexible support (span, pretension), a baseline digital twin model is constructed in the cloud. Specifically, the length of the exposed area of ​​the steel strands can be any value, but in engineering, it is generally 250mm or more for ease of installation.

[0032] Step S102: Based on the exposed area characteristics of the steel strand of the photovoltaic flexible support in the multimodal data and the coordinates of multiple key feature points of the benchmark digital twin model, the complete spatial morphology model of the steel strand is inverted and constructed.

[0033] It should be noted that the exposed area features refer to the location, outline, and geometric features of the ≥250mm section of the steel strand that is only exposed after the photovoltaic panel is installed; the key feature point coordinates refer to the three-dimensional spatial coordinate points on the exposed section of the steel strand extracted from the benchmark digital twin model for the calculation of the catenary; the complete spatial morphology model refers to the digital model constructed through the catenary inversion algorithm, which can characterize the spatial curve, sag, and tension of the entire steel strand (including the hidden area).

[0034] In this embodiment of the invention, multimodal data collected by UAV is used to identify the exposed features of the unshaded steel strands of the photovoltaic flexible support. Based on this, the three-dimensional coordinates of multiple key feature points on the exposed section of the steel strands are extracted from the completed benchmark digital twin model.

[0035] Subsequently, using these key feature points as constraints, and combining the catenary mechanical model with spatial geometric algorithms, the morphology of the middle section of the steel strand, which was originally obscured by the photovoltaic panel and could not be directly measured, was calculated and inverted. Finally, a complete spatial morphology model was constructed, including the exposed and concealed sections, which can fully reflect the true spatial orientation, sag, and stress state of the steel strand, thus achieving a blind-spot-free and accurate characterization of the full-span structural morphology of the steel strand.

[0036] Step S103: Input the real-time vibration monitoring data of the steel strand into the complete spatial morphology model, dynamically correct and judge the complete morphology curve of the steel strand, and determine the abnormal line segment of the steel strand based on the judgment result.

[0037] It should be noted that real-time vibration monitoring data refers to dynamic time-series data such as vibration acceleration, amplitude, and frequency of the steel strand collected by ground vibration sensors; dynamic correction refers to the instantaneous updating of the static morphological curve in the complete spatial morphological model using real-time vibration data to make it conform to the actual motion state; state discrimination refers to comparing the corrected sag, tension, and vibration parameters with safety thresholds to determine whether the structure is in a normal working state; abnormal segments refer to steel strand segments whose sag, tension, and vibration amplitude exceed the threshold range and are judged to have safety risks.

[0038] In this embodiment of the invention, real-time vibration monitoring data of the steel strand, collected in real time by vibration sensors, is input into a pre-constructed complete spatial morphological model of the steel strand. Dynamic parameters such as vibration acceleration, amplitude, and frequency are used to correct the original complete morphological curve based on static geometric inversion in real time, ensuring that the morphological curve accurately reflects the instantaneous deformation state under external forces such as wind loads. Based on this, the vibration amplitude corresponding to the corrected instantaneous sag, real-time tension, and spatial attitude is assessed and compared with preset thresholds. When parameters exceed the threshold range, the corresponding location is determined to be in an abnormal state. Combining the spatial location and topological relationship of the baseline digital twin model, abnormal segments of the steel strand with safety risks are precisely located and identified, providing accurate targets for subsequent targeted review and inspection.

[0039] Step S104: Use a drone to perform targeted verification and inspection of abnormal segments of the steel strand.

[0040] It should be noted that targeted verification and inspection refers to close-range, targeted verification and inspection conducted by drones along adaptive routes to the target area based on the location of abnormal line segments.

[0041] In this embodiment of the invention, based on the determined location information of abnormal steel strand segments, an adaptive inspection route for UAVs that is adapted to complex terrain and flexible support structures is automatically planned, and the UAVs are controlled to fly to the abnormal section to perform directional and close-range verification and inspection operations.

[0042] This embodiment provides a method for intelligent inspection of photovoltaic (PV) brackets based on UAV perception, applied to an intelligent inspection system for PV brackets. The intelligent inspection system for PV brackets includes a UAV. Figure 2 This is a flowchart of an intelligent inspection method for photovoltaic brackets based on UAV perception according to an embodiment of the present invention, as follows: Figure 2 As shown, the process includes the following steps: Step S201: Obtain multimodal data of photovoltaic flexible support collected by UAV, and construct a benchmark digital twin model of photovoltaic flexible support by combining the design parameters of photovoltaic flexible support.

[0043] In some optional implementations, step S201 above includes: Step S2011: Obtain point cloud data and image data corresponding to the exposed area of ​​the steel strands of the photovoltaic flexible support and all columns collected by the UAV.

[0044] It should be noted that the exposed area of ​​the steel strands refers to the limited area (usually ≥250mm) where the steel strands are not obstructed after the photovoltaic panels are installed, and is only exposed on both sides of the support structure. This is a structural part that can be directly observed. The column refers to the vertical support component of the photovoltaic flexible support system, used to support the steel strands and photovoltaic modules, and is a key load-bearing structure of the support system. Point cloud data refers to a spatial dataset composed of a large number of three-dimensional coordinate points collected by a drone equipped with a lidar system, which can accurately reflect the surface geometry and spatial position of the support system. Image data refers to high-definition image data captured by a drone equipped with a visible light camera, used for structural feature identification, area positioning, and appearance observation.

[0045] In embodiments of the present invention, such as Figure 3 As shown in the image, the photovoltaic flexible support structure is viewed from the perspective of a drone. The specific components include the photovoltaic module 1, the steel strands below the module 2, the inclined beam 3, the intermediate column 4, the intermediate column diagonal brace 5, the fixing bolt 6, and the intermediate column support rod 7.

[0046] Specifically, using a miniature lidar and a high-resolution visible light camera mounted on a drone, and through a hardware synchronous triggering mechanism, high-precision 3D point cloud and 2D orthophotos of the exposed areas of the steel strands (≥250mm gaps) and all columns are simultaneously acquired during flight inspection. This yields point cloud data and image data. The spatial registration error of the point cloud is controlled to the centimeter level.

[0047] Step S2012: Combine point cloud data and image data to generate multimodal data of photovoltaic flexible support.

[0048] In this embodiment of the invention, the point cloud data of the exposed area of ​​the steel strand and the column and the image data acquired at the same time and from the same perspective are spatially registered, temporally synchronized and feature-associated, so that the three-dimensional geometric position information and the two-dimensional visual texture information correspond one-to-one, forming integrated data with high-precision spatial structure and clear visual features, namely, the multimodal data of photovoltaic flexible support.

[0049] Step S2013: Using the multimodal data and design parameters of the photovoltaic flexible support, a benchmark digital twin model of the photovoltaic flexible support is constructed.

[0050] In this embodiment of the invention, based on the fused multimodal data of the photovoltaic flexible support, and relying on the three-dimensional geometric contour, spatial position, and appearance feature information of the support carried by the multimodal data, combined with the predetermined structural dimensions, span layout, component materials, installation arrangement, and other design parameters of the photovoltaic flexible support, a digital replica of the overall structure, steel strands, columns, and other components of the support is completed in the cloud through 3D modeling and topology association algorithms, thus constructing a baseline digital twin model of the photovoltaic flexible support. The baseline digital twin model includes the following three layers of information: Geometric reference layer: precise location of all columns, spatial curve of exposed steel strands.

[0051] Physical reference layer: design tension of steel strand, material properties (linear density, elastic modulus).

[0052] Topological reference layer: the connection relationship between columns and steel strands, and the spatial correspondence of each span.

[0053] Step S202: Based on the exposed area characteristics of the steel strand of the photovoltaic flexible support in the multimodal data and the coordinates of multiple key feature points of the benchmark digital twin model, the complete spatial morphology model of the steel strand is inverted and constructed.

[0054] In some optional implementations, step S202 above includes: Step S2021: Identify the exposed area characteristics of the steel strands of the photovoltaic flexible support from the multimodal data.

[0055] In this embodiment of the invention, multimodal data is analyzed and processed to identify and extract the exposed area features of the steel strands of the photovoltaic flexible support.

[0056] Step S2022: Based on the exposed area features, extract the coordinates of multiple key feature points from the benchmark digital twin model, and randomly select a preset number of non-collinear key feature point coordinates from all the key feature point coordinates to construct two vectors.

[0057] It should be noted that the preset number of non-collinear key feature point coordinates refers to the three-dimensional coordinates of a pre-set number (usually ≥3) of key feature points that are not on the same straight line. These coordinates are used to construct vectors and solve for the plane containing the catenary, ensuring calculation accuracy. Vectors refer to spatially directed line segments constructed based on the coordinates of non-collinear key feature points. These vectors are used to characterize the spatial positional relationship between feature points and are the basis for calculating the normal vector of the plane containing the catenary.

[0058] In this embodiment of the invention, based on the exposed area characteristics of the steel strand, the coordinates of four key feature points are extracted from the benchmark digital twin model. Among them, the exposed feature point P1 near the left column is... x 1, y 1, z 1), Exposed P2 near the midsection ( x 2, y 2, z 2) and the right column near the mid-span P3 ( x 3, y 3, z 3), P4 near the right end of the column x 4, y 4, z 4) These four key feature points belong to a single steel strand, so they are coplanar. The core of the spatial catenary equation is dimensionality reduction: first, project the three-dimensional points onto the plane where the catenary is located, convert them into two-dimensional coordinates, then fit the two-dimensional catenary equation, and finally restore it to a three-dimensional spatial equation.

[0059] Since the four points are coplanar, the equation of the plane can be calculated using the three-point method. Ax + By + Cz + D Given that (P1, P2, P4) = 0, select three non-collinear points (e.g., (P1, P2, P4)) and construct two vectors: , .

[0060] Step S2023: Using all vectors, calculate the normal vector of the plane containing the catenary of the steel strand, and determine the plane equation of the plane containing the catenary based on the normal vector.

[0061] It should be noted that the plane containing the catenary refers to the spatial plane corresponding to the catenary formed by the axis of the steel strand in its natural suspension state. It is the key carrier for reducing the three-dimensional shape calculation to two-dimensional calculation. The normal vector refers to the spatial vector perpendicular to the plane containing the catenary. It is used to accurately determine the spatial orientation of the plane containing the catenary and then derive the plane equation. The plane equation refers to the mathematical equation established based on the normal vector and the coordinates of key feature points. It is used to accurately describe the spatial position of the plane containing the catenary and provide a benchmark for subsequent two-dimensional fitting.

[0062] In this embodiment of the invention, the normal vector of the plane is calculated based on the two vectors mentioned above:

[0063] After obtaining the three parameters A, B, and C in the above formula, substitute them into the coordinates of P1 to calculate D in the plane equation: D = ( Ax 1+ By 1+ Cz 1) Therefore, the plane equation of the plane containing the final catenary is obtained as follows: Ax 1+ By 1+ Cz 1+D=0.

[0064] Step S2024: Set the coordinates of any key feature point as the origin of the local coordinate system, determine the coordinates of two-dimensional points based on the origin of the local coordinate system, and fit the catenary equation using the coordinates of each two-dimensional point.

[0065] It should be noted that the origin of the local coordinate system refers to a local two-dimensional coordinate system established by arbitrarily selecting one of the key feature points as the origin. This system is used to convert the three-dimensional coordinates of the key feature points into two-dimensional coordinates, simplifying the catenary fitting calculation. The two-dimensional point coordinates refer to the two-dimensional coordinates obtained by converting the three-dimensional coordinates of the key feature points through the local coordinate system, which are adapted to the two-dimensional fitting requirements of the catenary equation. The catenary equation refers to the mathematical equation constructed based on the hyperbolic cosine function, which is used to characterize the curve shape of the steel strand under natural suspension. Its parameters are determined by the fitting of the two-dimensional point coordinates.

[0066] In this embodiment of the invention, the spatial catenary is a planar curve, and a local coordinate system is established within this plane. u , v ), to three-dimensional points ( x , y , z Transform into two-dimensional points ( u , v ).

[0067] Choose the origin: Take P1 as the origin O' of the local coordinate system. x 1, y 1, z 1) Sure u Axis direction: take vector As u The positive direction of the axis, u The coordinates are the projection lengths from point O': (i=1,2,3,4), (P1's u 1=0) Sure v Axial direction: perpendicular to the plane uThe direction of the axis is v axis, v Coordinates are from point to u The perpendicular distance to the axis (which can be calculated via vector projection, from point to...) u The distance between the axis and the line = the magnitude of the vector × the sine of the angle between the vector and the line), ultimately yielding four two-dimensional points. u 1, v 1),( u 2, v 2),( u 3, v 3),( u 4, v 4).

[0068] For two-dimensional points ( u i , v i Fitting the catenary equation: The general solution for the catenary of steel strand is:

[0069] In the formula, cosh( t ) for and t The relevant hyperbolic cosine function = , a , b , u 0 is the parameter to be determined, where, a The "opening size" of the catenary is determined. a The larger the value, the flatter the curve. u 0: The x-coordinate of the curve's axis of symmetry (the catenary about the line ( u = u 0) Symmetrical); b The vertical translation of the curve.

[0070] Step S2025: Based on the coordinates of all key feature points, solve the catenary equation, determine the geometric shape of the catenary equation, and input it into the benchmark digital twin model.

[0071] It should be noted that the geometric shape of the catenary equation refers to the curve shape of the exposed section of the steel strand obtained by solving the catenary equation, which intuitively reflects the spatial orientation and bending state of the exposed section.

[0072] In this embodiment of the invention, it is assumed that the coordinates of four points are known as ( u 1, v 1),( u 2, v 2),( u 3, v 3),( u 4, v4) Substituting these equations into the general equation for the catenary, we obtain a system of four nonlinear equations. Solving this system yields the parameters. a , b , u The three parameters determine the geometry of the equation, and the geometry of the equation is then input into the baseline digital twin model.

[0073] Step S2026: Using the endpoint coordinates and tangent direction of the exposed area characteristics of the steel strand as boundary conditions, calculate the complete shape curve, maximum mid-span sag, and real-time tension of the steel strand through the benchmark digital twin model.

[0074] It should be noted that the boundary conditions refer to the endpoint coordinates (spatial position constraints) and tangent direction (curve direction constraints) of the exposed area of ​​the steel strand, which are used to invert and calculate the shape of the hidden section blocked by the photovoltaic panel; the complete shape curve refers to the complete spatial curve of the entire span of the steel strand (exposed section + hidden section) obtained by combining the shape of the exposed section catenary and the inversion results of the hidden section.

[0075] In this embodiment of the invention, a parametric catenary / parabolic mechanical model is implanted with a benchmark digital twin model. The endpoint coordinates and tangent directions of the exposed areas on both sides of the steel strand are used as forced boundary conditions to automatically fit and calculate the theoretical three-dimensional shape of the steel strand in the area shaded by the photovoltaic panel. This serves as a "digital mirror" of the concealed area. The position of any point on the steel strand can be calculated using the theoretical three-dimensional shape of the steel strand, including the maximum sag at mid-span, the precise height at any position, and the real-time tension. The precise height at any position can be used to draw the complete shape curve of the steel strand.

[0076] Specifically, in photovoltaic flexible support systems, the steel strands sag naturally under their own weight, forming a smooth curve, which is mechanically known as a "caten." Although the mathematical form of the catenary is quite complex, its shape is essentially determined by three core parameters: the degree of curvature... a , location of the center of symmetry u 0. Vertical height offset b Once these three parameters are determined, the shape of the entire curve is uniquely determined. The specific spatial position of the steel strand can then be calculated using the coordinates of the aforementioned four feature points. The two inner points (P2, P3): located near the mid-span, are the actual points where the steel strand passes through the edge of the photovoltaic panel opening. These two points are control points that the catenary must strictly pass through, directly determining the "must-pass path" of the curve at critical locations. The two outer points (P1, P4): located near the column, are connected to the two inner points. The direction of the line connecting these two points to the inner points can approximately reflect the "direction" or "tangential direction" of the steel strand at P2 and P3. This is analogous to knowing the height of a curve at a certain point and whether it curves upwards or downwards at that point, which is crucial for determining the shape of the curve.

[0077] With the location and orientation information provided by these four points, the catenary can be determined through mathematical "curve fitting". a , b , u There are three core parameters. This process is similar to finding an ideal catenary that is optimal when the following conditions are met: Passing through points P2 and P3: This is the most basic requirement, ensuring that the curve at the opening of the photovoltaic panel is consistent with the actual observation.

[0078] The direction at P1 and P4 and , The lines are aligned in the same direction: This uses the additional information provided by the outer points to constrain the local shape of the curve and avoid unreasonable bending.

[0079] Since the information from four points exceeds the information from three unknown parameters, it is usually impossible to find a curve that perfectly satisfies all the conditions simultaneously. Therefore, the actual approach used is "optimization"—finding a catenary that best approximates the position and direction implied by the four points overall. Specifically, this involves continuously adjusting the three parameters and repeatedly performing trial calculations until the sum of errors in the following two aspects is minimized: The deviation between the calculated height and the measured height of the curve at points P2 and P3; The calculated trend of the curve at points P2 and P3 is the same as , Deviation in the direction of the connecting line.

[0080] This process of "repeated trial and error and gradual approximation" is called "iterative optimization" in mathematics. Computers can complete thousands of trial calculations in a very short time to find the optimal solution.

[0081] once a , b , u With the three parameters determined, the equation for the entire catenary is completely determined. At this point, even if the middle section of the steel strand is completely shaded by the photovoltaic panel, the position of any point on it can be calculated, including: Maximum sag at mid-span: the lowest point of the steel strand, which is the core indicator for judging structural safety.

[0082] Precise height at any position: It can draw the complete shape curve of the entire steel strand.

[0083] Real-time tension calculation: By analyzing the curvature of the curve, the current tension of the steel strand can be deduced, providing a mechanical basis for structural health assessment. Using this method, the complete spatial morphology of the entire steel strand can be reconstructed with high precision using only four points on the 500mm exposed sections at both ends, completely solving the monitoring blind spot problem caused by photovoltaic panel shading.

[0084] Step S2027: Using the complete shape curve of the steel strand, the maximum sag at mid-span, and the real-time tension, construct the complete spatial shape of the steel strand.

[0085] It should be noted that the maximum sag at mid-span refers to the maximum vertical distance between the mid-span position of the steel strand and the line connecting the two end supports, which is a key parameter characterizing the slack and stress state of the steel strand; real-time tension refers to the tensile force borne by the steel strand in the current state, calculated by combining the complete shape curve with the mechanical model, reflecting the load-bearing state of the steel strand; the complete spatial shape of the steel strand refers to a digital model that integrates the complete shape curve of the steel strand, the maximum sag at mid-span, real-time tension, and other parameters, which can comprehensively and accurately characterize the spatial position, shape, and stress state of the steel strand throughout the entire span.

[0086] In this embodiment of the invention, based on the calculated complete shape curve of the steel strand (covering the entire span of the exposed and concealed sections), and combined with the two core mechanical and geometric parameters of the maximum sag at mid-span of the steel strand (reflecting its relaxation degree and geometric deformation characteristics) and real-time tension (reflecting its load-bearing stress state), the curve shape, sag, and tension are organically integrated through parameter correlation and model fusion algorithms to construct a digital model that can comprehensively and accurately reflect the entire span of the steel strand's spatial position, geometric shape, and stress state, i.e., the complete spatial shape of the steel strand.

[0087] Step S203: Input the real-time vibration monitoring data of the steel strand into the complete spatial morphology model, dynamically correct and judge the complete morphology curve of the steel strand, and determine the abnormal line segment of the steel strand based on the judgment result.

[0088] In some optional implementations, the intelligent inspection system for photovoltaic brackets also includes a vibration sensor, and step S203 includes: Step S2031: Obtain real-time vibration monitoring data of the steel strand collected by the vibration sensor.

[0089] It should be noted that a vibration sensor refers to a device used to collect data related to the vibration of steel strands. It can capture the vibration state of steel strands in real time and output core parameters such as vibration acceleration, amplitude, and frequency. It is the core component for obtaining real-time vibration data. Real-time vibration monitoring data refers to various data collected in real time by vibration sensors that reflect the vibration state of steel strands. These mainly include vibration acceleration, amplitude, and frequency. It is the core basis for dynamic correction and state judgment.

[0090] In embodiments of the present invention, such as Figure 4 As shown, high-frequency vibration sensors (such as accelerometers) are deployed directly and rigidly connected to rigid components adjacent to the steel strand at key mechanical nodes (such as mid-span).

[0091] The real-time vibration monitoring data of the steel strand collected by the vibration sensor is transmitted to the intelligent inspection system of the photovoltaic support via network transmission. The network performs continuous monitoring 24 / 7 with a sampling rate of ≥500Hz, directly capturing full-band vibration time history data in the range of 0.1-100Hz, which serves as a high-fidelity and high-reliability reference vibration source.

[0092] Step S2032: Input the real-time vibration monitoring data into the complete spatial morphology model, and use the vibration acceleration, amplitude and frequency of the real-time vibration monitoring data to dynamically correct the complete morphology curve.

[0093] It should be noted that vibration acceleration refers to the rate of change of velocity during the vibration of the steel strand, which is the core parameter reflecting the vibration intensity of the steel strand. It is used to dynamically correct the complete morphological curve and reflect the intensity of its vibration. Amplitude refers to the maximum distance the steel strand deviates from its equilibrium position during vibration, and it is a key indicator for judging its vibration amplitude and whether there is any abnormality. Frequency refers to the number of vibrations of the steel strand per unit time. Combined with acceleration and amplitude, it can comprehensively reflect its vibration state and provide a basis for morphological correction and anomaly judgment. Dynamic correction refers to the operation of using real-time vibration monitoring data (acceleration, amplitude, frequency) to adjust the complete morphological curve of the steel strand in real time so that the curve is consistent with the actual vibration state of the steel strand.

[0094] In this embodiment of the invention, the photovoltaic support intelligent inspection system uses real-time vibration monitoring data as dynamic input to input a complete spatial morphology model, thereby correcting and verifying the theoretical morphology in real time, and thus indirectly and accurately restoring the dynamic spatial posture and sag of the entire steel strand.

[0095] Specifically, the vibration acceleration, amplitude, and frequency of the steel strand are identified from real-time vibration monitoring data. Through a preset dynamic correction algorithm, combined with the above three types of vibration parameters, the original complete shape curve of the steel strand in the model is adjusted and updated in real time to eliminate the deviation between static modeling and actual vibration conditions. This allows the corrected complete shape curve to accurately match the actual spatial orientation and deformation degree of the steel strand under the current vibration state.

[0096] Step S2033: Calculate the instantaneous sag and real-time tension of the steel strand based on the dynamically corrected morphological curve.

[0097] It should be noted that instantaneous sag refers to the vertical sag distance of the steel strand at the current moment after dynamic correction, reflecting its instantaneous geometric shape and is one of the core parameters for judging whether there is an anomaly; real-time tension refers to the tension that the steel strand bears at the current moment after dynamic correction, reflecting its stress state, and together with instantaneous sag, it serves as the basis for judging anomalies.

[0098] In this embodiment of the invention, the complete shape curve of the steel strand after dynamic correction of vibration parameters is used as the calculation basis. Based on the catenary mechanics theory and spatial geometric relationship, combined with the constraint conditions of the support structure, the instantaneous sag and real-time tension of the steel strand at the current moment are solved.

[0099] Step S2034: The instantaneous sag, real-time tension, and vibration amplitude of the steel strand are compared with the corresponding preset thresholds.

[0100] It should be noted that the vibration amplitude, i.e. the specific value of the amplitude, is a direct reflection of the vibration intensity. It is used together with the instantaneous sag and real-time tension for threshold comparison to determine the condition of the steel strand. The preset threshold refers to the reasonable range of various parameters (instantaneous sag, real-time tension, and vibration amplitude) set according to engineering design standards and safety specifications. It is the benchmark for judging whether the steel strand is abnormal.

[0101] In this embodiment of the invention, the instantaneous sag, real-time tension, and vibration amplitude of the steel strand, obtained through dynamic calculation, are used as evaluation indicators. Each indicator is matched against a pre-set threshold based on the photovoltaic flexible support design specifications and structural safety limits, and a comparative analysis is conducted. By comparing each indicator with its reasonable value range, the current degree of geometric deformation, load-bearing stress state, and vibration intensity of the steel strand are quantitatively evaluated, and the structural operating condition is comprehensively checked from three dimensions: morphology, mechanics, and vibration.

[0102] Step S2035: If the comparison result exceeds the threshold range, the position corresponding to the position exceeding the threshold range is determined as an abnormal state. Based on the topological relationship of the benchmark digital twin model, the steel strand segment corresponding to the abnormal state is located, and the abnormal line segment of the steel strand is generated.

[0103] It should be noted that an abnormal state refers to a state in which the instantaneous sag, real-time tension, or vibration amplitude of the steel strand exceeds a preset threshold, posing a potential structural safety hazard; the topological relationship of the benchmark digital twin model refers to the spatial position, connection relationship, and layout logic of each component (such as steel strand and column) in the benchmark model, which is used to accurately locate the specific location of the abnormal section.

[0104] In this embodiment of the invention, when any parameter—instantaneous sag, real-time tension, or vibration amplitude—exceeds the corresponding preset threshold range, the spatial location where the parameter exceeds the limit is immediately marked as a structural anomaly. Simultaneously, relying on the overall topological connection relationship of the photovoltaic flexible support and the spatial location association information of the components embedded in the benchmark digital twin model, the specific steel strand segment to which the anomaly belongs is accurately mapped and locked, completing the matching and positioning from the parameter exceeding point to the physical structural segment, and ultimately delineating and generating a clearly defined abnormal steel strand segment.

[0105] Step S204: Use a drone to perform targeted verification and inspection of abnormal segments of the steel strand.

[0106] In some optional implementations, step S204 above includes: Step S2041: Generate an adaptive inspection route for the UAV based on the early warning information corresponding to the abnormal segments of the steel strand.

[0107] It should be noted that the early warning information refers to the prompt information generated based on the abnormal status of the steel strand, which includes the abnormal location, abnormal type and risk level, and is used to trigger subsequent review and inspection, providing a basis for fault handling; the adaptive inspection route refers to the optimal flight path automatically planned by combining the location of the abnormal steel strand segment, complex terrain and equipment performance, which can flexibly adapt to the on-site environment, ensure that the UAV accurately reaches the target area and improve inspection efficiency.

[0108] In this embodiment of the invention, the invention constructs a three-dimensional monitoring network that combines air and ground monitoring with dynamic and static monitoring, integrating direct measurement and macroscopic scanning to achieve comprehensive diagnosis from high-frequency dynamics to long-term evolution.

[0109] It is worth mentioning that, while deploying high-frequency vibration sensors at key mechanical nodes of the steel strands serves as static monitoring of the photovoltaic flexible support, using drones for macroscopic morphology and auxiliary inspection serves as dynamic monitoring. Therefore, when early warning information corresponding to abnormal segments of the steel strands is obtained, an adaptive inspection route for the drone can be planned, facilitating control of the drone to proceed to the abnormal segments of the steel strands for re-inspection according to the early warning information.

[0110] Step S2042: The adaptive inspection route and early warning information are sent to the UAV so that the UAV can perform targeted review inspection according to the adaptive inspection route.

[0111] It should be noted that targeted review and inspection refers to precise inspection of abnormal segments of steel strands. It focuses on abnormal areas, verifies the authenticity of abnormalities through close observation and data collection, avoids misjudgment, and forms a closed-loop management system.

[0112] In this embodiment of the invention, precise verification and inspection commands are issued to the UAV to ensure targeted and efficient execution of the inspection task. Specifically, the system-pre-planned adaptive inspection route (adapting to the location of abnormal segments and avoiding complex terrain and support obstructions), along with early warning information including the abnormal location, abnormal parameters, and risk level, are simultaneously sent to the UAV control system. After receiving the command, the UAV uses the adaptive inspection route as its flight basis, autonomously flying along the planned path to accurately reach the area where the abnormal steel strand segment is located. It then focuses on the abnormal part to conduct targeted, close-range verification and inspection, avoiding ineffective inspection work and ensuring that the inspection focuses on the core target. This provides precise flight support and task guidance for subsequent on-site verification of abnormalities and fault confirmation.

[0113] Step S2043: When the UAV flies to the area where the abnormal line segment corresponding to the warning information is located, collect point cloud data and image data of the area where the abnormal line segment of the steel strand is located.

[0114] In this embodiment of the invention, the macroscopic form of the UAV and its auxiliary inspection (dynamic, periodic, full coverage) are as follows: LiDAR macroscopic scanning: UAVs regularly perform full-coverage flights, acquiring full-field 3D point clouds using high-precision LiDAR. Through multi-period point cloud automatic registration and change detection algorithms, long-term, slow changes (millimeter-level accuracy) in steel strand sag and overall cable net morphology are quantitatively analyzed to identify structural slack or displacement.

[0115] Air-Ground Collaborative Closed-Loop Verification: When the ground-based sensor network detects instantaneous sag anomalies, real-time tension anomalies, vibration spectrum anomalies, or amplitude exceeding limits in the early warning information, the system automatically generates a verification command. After receiving the command, the UAV flies to the coordinates of the area where the abnormal steel strand segment carried in the early warning information is located. Using its onboard high-frame-rate visible light camera, it performs close-range hovering photography or slow-speed tracking scans on the suspected abnormal steel strand segment to obtain intuitive visual evidence, completing the intelligent closed loop of "data alarm - visual verification".

[0116] Point cloud data and image data of the area where abnormal segments of steel strands are located are extracted from the miniature lidar and high-resolution visible light camera carried by the drone.

[0117] Step S2044: Based on the point cloud data and image data of the area where the abnormal line segment is located, conduct on-site verification of the abnormal line segment.

[0118] It should be noted that on-site verification refers to confirming the actual state of abnormal line segments by comparing point cloud and image data, eliminating false alarms, ensuring the accuracy of anomaly judgment, and guaranteeing the closed loop of inspection.

[0119] In this embodiment of the invention, point cloud data and image data of the area where the abnormal line segment is located, collected by a drone, are used to verify and confirm the abnormal state of the abnormal line segment of the steel strand on-site, eliminate misjudgments, and form an inspection closed loop. Specifically, the point cloud data and image data collected by the drone after flying to the area of ​​the abnormal line segment are the core basis. The point cloud data can accurately reflect the spatial shape and actual deformation details of the abnormal line segment, while the image data can intuitively present the appearance, damage, and other visual characteristics of the abnormal line segment. By fusing and analyzing the two types of data, comparing and verifying them, and combining the normal state parameters of the steel strand and the benchmark digital twin model, it is verified whether the abnormal state of the abnormal line segment really exists, the abnormal type is identified (e.g., excessive relaxation, abnormal vibration), and the degree of risk is determined. This avoids ineffective handling due to parameter misjudgment, ensures the accuracy of abnormal judgment, and provides reliable on-site data support for subsequent fault handling.

[0120] In some alternative implementations, the method further includes: Step S205: Obtain real-time meteorological data and real-time environmental data, and adjust the inspection time and inspection route of the UAV.

[0121] It should be noted that real-time meteorological data refers to meteorological information such as temperature, wind force, precipitation, and sunshine collected in real time. Its core purpose is to determine whether meteorological conditions are suitable for drone inspection, thereby adjusting the inspection time and flight path. Real-time environmental data refers to surrounding environmental data such as terrain, obstacle distribution, and air humidity collected in real time within the inspection area. This data is used to optimize drone inspection flight paths and avoid flight risks. Drone inspection time refers to the specific time period suitable for drone inspection operations, determined by combining real-time meteorological and environmental data, ensuring inspection safety and accurate data collection. Drone inspection flight path refers to the drone flight path planned based on real-time meteorological and environmental data and the inspection target (abnormal area of ​​the steel strand). This path can be flexibly adjusted to ensure efficient and safe inspection.

[0122] In this embodiment of the invention, by collecting meteorological and environmental data of the work area in real time, the impact of external working conditions on UAV flight and data collection is comprehensively assessed. Based on real-time meteorological elements such as wind speed, sunlight, and weather conditions, as well as real-time environmental elements such as site terrain and obstacle distribution, the working time and flight inspection route of the UAV are dynamically optimized and corrected. This avoids areas with severe weather and complex obstacles, selects suitable inspection times, and plans safe and reasonable flight paths to reduce external environmental interference, ensure UAV flight safety, and improve the completeness and accuracy of subsequent multimodal data collection.

[0123] Step S206: Based on the inspection time and route of the UAV, acquire real-time multimodal data of the photovoltaic flexible support collected by the UAV, and generate new multimodal data.

[0124] It should be noted that real-time multimodal data refers to the collection of various types of data, such as point clouds, images, and locations, collected by the UAV during the adjusted inspection time and along the flight path. It combines accuracy and completeness and is used for model updates.

[0125] In this embodiment of the invention, the UAV is controlled to fly autonomously along the planned path during the appropriate operating period, according to the adjusted and determined UAV inspection time and route, and simultaneously collect point cloud data and image data of photovoltaic flexible supports, steel strands, columns and other components. The original sensor data collected on site is spatiotemporally registered, denoised and preprocessed, and feature fusion and integration are performed to form updated multimodal data.

[0126] Step S207: Based on the new multimodal data, update the benchmark digital twin model of the photovoltaic flexible support and generate a new benchmark digital twin model.

[0127] In this embodiment of the invention, new multimodal data, re-collected and fused on-site, is used as the basis for comparison, verification, and parameter iteration with the original photovoltaic flexible support reference digital twin model. By replacing aged, unrealistic geometric parameters, appearance features, and spatial topology information in the model, the three-dimensional morphology and layout relationship of the overall support structure, columns, and steel strand components are synchronously corrected and iteratively optimized, completing the dynamic update of the original reference digital twin model and generating a new reference digital twin model that fits the current actual working conditions on site.

[0128] Step S208: Jump to the step of reversing and constructing a complete spatial morphology model of the steel strand based on the exposed area features of the photovoltaic flexible support steel strand in multimodal data and the coordinates of multiple key feature points of the benchmark digital twin model.

[0129] In this embodiment of the invention, after the dynamic update of the baseline digital twin model is completed, the program flow automatically jumps back to the steel strand morphology inversion modeling stage. Based on the updated multimodal data, the exposed area characteristics of the photovoltaic flexible support steel strand are identified, and the coordinates of multiple key feature points are extracted from the new baseline digital twin model. Then, through algorithms such as spatial vector operations, plane equation solving, local coordinate transformation, and catenary equation fitting, the complete spatial morphology model of the steel strand is reconstructed. This forms a cyclic iterative monitoring mechanism adapted to meteorological environments, continuously updating the geometric morphology and stress parameters of the steel strand, achieving uninterrupted dynamic monitoring and closed-loop inspection of the operating status of the photovoltaic flexible support steel strand.

[0130] Specifically, this invention formulates a scientific and efficient inspection strategy to guide drones in completing periodic general surveys and targeted detailed surveys, achieving a complete closed loop from "data collection" to "status assessment" and then to "response decision-making": 1. Inspection task generation and dynamic scheduling Inspection tasks are automatically generated based on the following three types of information: The baseline digital twin model contains static information such as the precise coordinates of all columns collected by drones, the design span and initial shape of the steel strands, and the layout of the photovoltaic panel array, which is used to plan the basic inspection path and key observation points.

[0131] Historical monitoring data: Based on the obtained data on the vibration characteristics and sag evolution curves of each steel strand segment, high-risk areas are identified and their inspection frequency is increased.

[0132] Real-time meteorological and environmental data: Combine real-time information such as wind speed, wind direction, and temperature collected by drones to dynamically adjust the timing of inspections (such as avoiding flying during periods of strong wind) and observation parameters.

[0133] Task types are divided into two categories: Periodic full-coverage survey: The entire site is scanned by lidar and visible light imaging at a preset cycle (such as weekly / monthly) to monitor the long-term slow changes in the sag of steel strands, the overall shape of the cable net, and macroscopic anomalies on the surface of photovoltaic modules.

[0134] Emergency review and detailed investigation: When the ground sensor network triggers an early warning (such as abnormal vibration spectrum, amplitude exceeding the limit, etc.), the system automatically generates a review task, assigns a drone to fly to the designated coordinates for close-range observation, and obtains high-resolution images and infrared thermal images as diagnostic evidence.

[0135] 2. Pre-defined route generation based on structural topology For periodic survey tasks, route planning no longer relies on real-time dynamic calculations, but instead performs offline pre-planning based on the structural topological relationships in the baseline digital twin model: Waypoint generation: Using the top of each column as the key waypoint, insert several intermediate observation points (equidistant or densified according to curvature) between the two columns along the direction of the steel strand to ensure that the UAV can collect data of the exposed section from the best perspective (the sensor axis is at an angle of 60° to 90° with the direction of the steel strand).

[0136] Flight path optimization: Under the premise of ensuring a safe distance (avoiding obstacles such as photovoltaic panels and pillars), a smooth, flyable flight path along the shape of the steel strand is obtained based on the above content, covering the entire field.

[0137] Route database management: The generated routes are bound to the corresponding digital twin model versions. When the model changes significantly due to structural modifications or long-term deformation, the routes are regenerated and the route database is updated.

[0138] 3. Autonomous execution and multimodal data acquisition by unmanned aerial vehicles (UAVs) After receiving mission instructions, the UAV automatically takes off and performs inspections according to a preset route. During flight, RTK positioning and visual-assisted navigation ensure track accuracy. Onboard sensors such as lidar, visible light camera, and infrared thermal imager are triggered synchronously through hardware to achieve precise spatiotemporal alignment of multimodal data.

[0139] 4. Data post-processing and model iterative updates The data collected by the drone is initially processed at the edge node and then uploaded to the cloud. The cloud platform performs the following operations: Morphology update: Using the catenary inversion algorithm, combined with the feature point data of the exposed area of ​​the newly collected steel strand, the dynamic morphology of the hidden area of ​​the steel strand is updated, and compared with the historical model to generate a sag change curve.

[0140] Anomaly detection: Detects surface defects (microcracks, dirt) of photovoltaic panels through image recognition algorithms, identifies hot spot anomalies through infrared analysis, and identifies structural displacements through point cloud comparison.

[0141] Model evolution: Newly accumulated real data (including normal and faulty samples) is used to train and optimize the diagnostic model, continuously improving the accuracy of early warning.

[0142] 5. Closed-loop feedback and operation and maintenance decision-making The system pushes the analysis results to the operations and maintenance terminal, including: Status assessment report: Current sag, estimated tension, trend of change, and safety level of each steel strand segment.

[0143] Warning information: Issue warnings for abnormal points that exceed dynamic thresholds, and attach verification images as supporting evidence.

[0144] Maintenance recommendations: Based on the type and severity of the anomaly, maintenance recommendations are automatically generated (such as "tensioner needs adjustment" or "local rust prevention treatment for steel strand"), and linked to spare parts inventory and personnel scheduling information to achieve intelligent dispatch of maintenance work orders.

[0145] Ultimately, based on the dynamic spatial attitude of the entire steel strand, the UAV can closely follow the spatial orientation of the steel strand and maintain a smooth three-dimensional flight trajectory with the best observation angle and safe distance.

[0146] The present invention has the following beneficial effects: First, in terms of core sensing capabilities, this invention achieves a qualitative leap from "partial blind testing" to "full-area visibility." Traditional methods are helpless due to photovoltaic panel obstruction, relying solely on indirect inference or costly contact-based point deployment. This invention innovatively utilizes the extremely small segments inevitably exposed at both ends of the steel strand to obtain its precise spatial coordinates through high-precision visual measurement using drones. Based on this, these discrete "spatial points" can be input into a simplified catenary mechanical model, and the algorithm automatically inverts and calculates the three-dimensional spatial curve of the entire completely obscured steel strand. This is analogous to accurately drawing the outline of the entire underwater bridge using only the two exposed sections of the piers, completely eliminating monitoring blind spots.

[0147] Secondly, in terms of the value of monitoring data, this invention upgrades from a "static snapshot" to a "dynamic movie." Most existing inspection technologies can only provide static data at a specific moment, failing to capture dynamic processes such as wind-induced vibration. This invention integrates two types of time-series data: high-frequency vibration signals collected by a fixed sensor network and macroscopic morphological change point clouds acquired through periodic UAV inspections. These two types of data can be fused and analyzed on a unified timeline, reproducing the dynamic changes in the sag of steel strands under wind loads, much like playing a movie, and simultaneously calculating the real-time fluctuations in their internal tension. This allows maintenance personnel not only to know "what the structure looks like now," but also "how it moves in the wind and how the forces change," providing crucial dynamic evidence for early warning.

[0148] Third, in terms of engineering implementation and economic benefits, this invention achieves a transformation from "high cost and low efficiency" to "high cost-effective automation." Traditional high-precision monitoring relies on densely deployed fixed sensors, with deployment costs for a single power station often exceeding one million yuan, and maintenance is difficult. This invention uses a mobile and flexible UAV as the main data acquisition platform, enabling it to automatically adapt to the complex spatial posture of mountainous terrain and support structures through intelligent flight path planning, achieving efficient full coverage. This significantly reduces the system's initial investment and long-term maintenance costs, with calculations showing a reduction of related operation and maintenance costs by more than 30%. Simultaneously, through accurate early anomaly warnings, it can effectively prevent major economic losses such as batch damage to components caused by excessive slack or breakage of steel strands, resulting in significant economic and safety benefits.

[0149] This embodiment also provides an intelligent inspection device for photovoltaic brackets based on UAV perception. This device is used to implement the above embodiments and preferred embodiments, and details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0150] This embodiment provides an intelligent inspection device for photovoltaic (PV) brackets based on UAV perception, applied to an intelligent inspection system for PV brackets. The intelligent inspection system for PV brackets includes a UAV, such as... Figure 5 As shown, it includes: The acquisition module 301 is used to acquire multimodal data of the photovoltaic flexible support collected by the UAV, and to construct a benchmark digital twin model of the photovoltaic flexible support in combination with the design parameters of the photovoltaic flexible support. The inversion module 302 is used to invert and construct a complete spatial morphology model of the steel strand based on the exposed area features of the steel strand of the photovoltaic flexible support in multimodal data and the coordinates of multiple key feature points of the benchmark digital twin model. The discrimination module 303 is used to input the real-time vibration monitoring data of the steel strand into the complete spatial morphology model, dynamically correct and discriminate the complete morphology curve of the steel strand, and determine the abnormal segments of the steel strand based on the discrimination results. Inspection module 304 is used to perform targeted review and inspection of abnormal segments of steel strands using drones.

[0151] In some optional implementations, the acquisition module 301 includes: The first acquisition unit is used to acquire point cloud data and image data corresponding to the exposed area of ​​the steel strands of the photovoltaic flexible support and all columns, collected by the UAV. The combining unit is used to combine point cloud data and image data to generate multimodal data for photovoltaic flexible supports; Model units are constructed to build a baseline digital twin model of the photovoltaic flexible support using multimodal data and design parameters.

[0152] In some alternative implementations, the inversion module 302 includes: The identification unit is used to identify the exposed area features of the steel strands of the photovoltaic flexible support from multimodal data; The optional unit is used to extract the coordinates of multiple key feature points from the benchmark digital twin model based on the features of the exposed area, and to select a preset number of non-collinear key feature point coordinates from all the key feature point coordinates to construct two vectors; The first calculation unit is used to calculate the normal vector of the plane containing the catenary of the steel strand using all vectors, and to determine the plane equation of the plane containing the catenary based on the normal vector. The fitting unit is used to set the coordinates of any key feature point as the origin of the local coordinate system, determine the coordinates of two-dimensional points based on the origin of the local coordinate system, and fit the catenary equation using the coordinates of each two-dimensional point. The solving unit is used to solve the catenary equation based on the coordinates of all key feature points, determine the geometric shape of the catenary equation, and input it into the benchmark digital twin model. The second calculation unit is used to calculate the complete shape curve, maximum mid-span sag, and real-time tension of the steel strand using the endpoint coordinates and tangent direction of the exposed area characteristics of the steel strand as boundary conditions through a reference digital twin model. The morphology building unit is used to construct the complete spatial morphology of the steel strand by using the complete morphological curve of the steel strand, the maximum sag at mid-span, and the real-time tension.

[0153] In some optional implementations, the discrimination module 303 includes: The second acquisition unit is used to acquire real-time vibration monitoring data of the steel strand collected by the vibration sensor. The input unit is used to input real-time vibration monitoring data into the complete spatial morphology model, and to dynamically correct the complete morphology curve using the vibration acceleration, amplitude and frequency of the real-time vibration monitoring data. The third calculation unit is used to calculate the instantaneous sag and real-time tension of the steel strand based on the dynamically corrected shape curve. The comparison unit is used to compare the instantaneous sag, real-time tension, and vibration amplitude of the steel strand with the corresponding preset thresholds. The positioning unit is used to determine the position corresponding to the threshold range as an abnormal state if the comparison result exceeds the threshold range, and to locate the steel strand segment corresponding to the abnormal state by combining the topological relationship of the benchmark digital twin model, and generate the abnormal line segment of the steel strand.

[0154] In some alternative implementations, the inspection module 304 includes: The early warning unit is used to generate an adaptive inspection route for the UAV based on the early warning information corresponding to the abnormal segments of the steel strand. The execution unit is used to send adaptive inspection routes and early warning information to the UAV, so that the UAV can perform targeted review inspections according to the adaptive inspection routes. The data acquisition unit is used to collect point cloud data and image data of the area where the abnormal line segment of the steel strand is located when the UAV flies to the area where the abnormal line segment corresponding to the warning information is located. The verification unit is used to perform on-site verification of abnormal line segments based on point cloud data and image data of the area where the abnormal line segments are located.

[0155] In some alternative embodiments, the device further includes: The adjustment unit is used to acquire real-time meteorological and environmental data and adjust the inspection time and inspection route of the UAV. The data acquisition unit is used to acquire real-time multimodal data of the photovoltaic flexible support collected by the drone based on the inspection time and inspection route of the drone, and generate new multimodal data. The update unit is used to update the baseline digital twin model of the photovoltaic flexible support based on the new multimodal data, and generate a new baseline digital twin model. The jump execution unit is used to jump to the execution of the steps of inverting and constructing a complete spatial morphology model of the steel strand based on the exposed area features of the steel strand of the photovoltaic flexible support in multimodal data and the coordinates of multiple key feature points of the benchmark digital twin model.

[0156] The intelligent inspection device for photovoltaic supports based on UAV perception provided in this embodiment of the invention can execute the intelligent inspection method for photovoltaic supports based on UAV perception provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects for executing the method. Further functional descriptions of the above modules and units are the same as in the corresponding embodiments described above, and will not be repeated here.

[0157] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.

[0158] The following is a detailed reference. Figure 6 This diagram illustrates a structural schematic suitable for implementing an electronic device according to embodiments of the present invention. The electronic device may include a processor (e.g., a central processing unit, graphics processor, etc.) 401, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 402 or a program loaded from memory 408 into random access memory (RAM) 403. The RAM 403 also stores various programs and data required for the operation of the electronic device. The processor 401, ROM 402, and RAM 403 are interconnected via a bus 404. An input / output (I / O) interface 405 is also connected to the bus 404.

[0159] Typically, the following devices can be connected to I / O interface 405: input devices 406 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 407 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; memory devices 408 including, for example, magnetic tapes, hard disks, etc.; and communication devices 409. Communication device 409 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 6 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown, and more or fewer devices may be implemented or have instead.

[0160] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 409, or installed from a memory 408, or installed from a ROM 402. When the computer program is executed by the processor 401, it performs the functions defined in the intelligent inspection method for photovoltaic brackets based on UAV perception according to embodiments of the present invention.

[0161] Figure 6 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0162] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the intelligent inspection method for photovoltaic brackets based on UAV perception shown in the above embodiments is implemented.

[0163] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0164] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for intelligent inspection of photovoltaic support structures based on UAV perception, characterized in that, The method is applied to an intelligent inspection system for photovoltaic (PV) mounting systems, which includes a drone, and includes: The multimodal data of the photovoltaic flexible support collected by the UAV are obtained, and a benchmark digital twin model of the photovoltaic flexible support is constructed in combination with the design parameters of the photovoltaic flexible support. Based on the exposed area characteristics of the steel strand of the photovoltaic flexible support in the multimodal data and the coordinates of multiple key feature points of the benchmark digital twin model, the complete spatial morphology model of the steel strand is inverted and constructed. The real-time vibration monitoring data of the steel strand is input into the complete spatial morphology model to dynamically correct and determine the state of the complete morphology curve of the steel strand, and the abnormal segments of the steel strand are determined according to the determination results. The drone is used to perform targeted review and inspection of abnormal segments of the steel strand.

2. The method according to claim 1, characterized in that, The process of acquiring multimodal data of the photovoltaic flexible support collected by the UAV and constructing a baseline digital twin model of the photovoltaic flexible support in combination with the design parameters of the photovoltaic flexible support includes: The point cloud data and image data corresponding to the exposed area of ​​the steel strands of the photovoltaic flexible support and all columns are collected by the UAV. By combining the point cloud data and the image data, multimodal data of the photovoltaic flexible support is generated; A baseline digital twin model of the photovoltaic flexible support is constructed using the multimodal data and design parameters of the photovoltaic flexible support.

3. The method according to claim 1, characterized in that, Based on the exposed area characteristics of the steel strands of the photovoltaic flexible support in the multimodal data and the coordinates of multiple key feature points of the benchmark digital twin model, a complete spatial morphology model of the steel strands is inverted and constructed, including: Identify the exposed area characteristics of the steel strands of the photovoltaic flexible support from the multimodal data; Based on the exposed area features, multiple key feature point coordinates are extracted from the benchmark digital twin model, and a preset number of non-collinear key feature point coordinates are randomly selected from all the key feature point coordinates to construct two vectors; Using all the aforementioned vectors, calculate the normal vector of the plane containing the catenary of the steel strand, and based on the normal vector, determine the plane equation of the plane containing the catenary; The coordinates of any of the key feature points are set as the origin of the local coordinate system. Based on the origin of the local coordinate system, the coordinates of two-dimensional points are determined, and the catenary equation is fitted using the coordinates of each of the two-dimensional points. Based on the coordinates of all the key feature points, solve the catenary equation, determine the geometric shape of the catenary equation, and input it into the benchmark digital twin model. Using the endpoint coordinates and tangent direction of the exposed area of ​​the steel strand as boundary conditions, the complete shape curve, maximum mid-span sag, and real-time tension of the steel strand are calculated through the reference digital twin model. The complete spatial shape of the steel strand is constructed by using its complete shape curve, maximum sag at mid-span, and real-time tension.

4. The method according to claim 3, characterized in that, The intelligent inspection system for photovoltaic supports also includes vibration sensors. The system inputs real-time vibration monitoring data of the steel strand into the complete spatial morphology model, dynamically corrects and determines the state of the complete morphology curve of the steel strand, and identifies abnormal segments of the steel strand based on the determination results. The vibration sensor collects real-time vibration monitoring data of the steel strand. The real-time vibration monitoring data is input into the complete spatial morphology model, and the vibration acceleration, amplitude, and frequency of the real-time vibration monitoring data are used to dynamically correct the complete morphology curve. Based on the dynamically corrected shape curve, the instantaneous sag and real-time tension of the steel strand are calculated. The instantaneous sag, real-time tension, and vibration amplitude of the steel strand are compared with the corresponding preset thresholds. If the comparison result exceeds the threshold range, the position corresponding to the position exceeding the threshold range is determined to be an abnormal state. Based on the topological relationship of the benchmark digital twin model, the steel strand segment corresponding to the abnormal state is located, and the abnormal line segment of the steel strand is generated.

5. The method according to claim 1, characterized in that, The targeted review and inspection of abnormal segments of the steel strand using the drone includes: Based on the early warning information corresponding to the abnormal segments of the steel strand, an adaptive inspection route for the UAV is generated. The adaptive inspection route and the early warning information are sent to the UAV, so that the UAV performs targeted review inspection according to the adaptive inspection route; When the drone flies to the area where the abnormal line segment corresponding to the warning information is located, it collects point cloud data and image data of the area where the abnormal line segment of the steel strand is located; Based on the point cloud data and image data of the area where the abnormal line segment is located, the abnormal line segment is verified on-site.

6. The method according to claim 1, characterized in that, The method further includes: Acquire real-time meteorological and environmental data to adjust the inspection time and route of the UAV; Based on the inspection time and inspection route of the UAV, real-time multimodal data of the photovoltaic flexible support collected by the UAV is obtained, and new multimodal data is generated. Based on the new multimodal data, the baseline digital twin model of the photovoltaic flexible support is updated to generate a new baseline digital twin model; Jump to the step of reversing and constructing a complete spatial morphology model of the steel strand based on the exposed area features of the steel strand of the photovoltaic flexible support in the multimodal data and the coordinates of multiple key feature points of the benchmark digital twin model.

7. A photovoltaic support intelligent inspection device based on UAV perception, characterized in that, This is applied to an intelligent inspection system for photovoltaic (PV) mounting systems, which includes a drone. The device includes: The acquisition module is used to acquire multimodal data of the photovoltaic flexible support collected by the UAV, and to construct a benchmark digital twin model of the photovoltaic flexible support in combination with the design parameters of the photovoltaic flexible support. The inversion module is used to invert and construct a complete spatial morphology model of the steel strand based on the exposed area features of the steel strand of the photovoltaic flexible support in the multimodal data and the coordinates of multiple key feature points of the benchmark digital twin model. The discrimination module is used to input the real-time vibration monitoring data of the steel strand into the complete spatial morphology model, dynamically correct and discriminate the complete morphology curve of the steel strand, and determine the abnormal segments of the steel strand based on the discrimination results. The inspection module is used to perform targeted review and inspection of abnormal segments of the steel strand using the drone.

8. An electronic device, characterized in that, include: The system includes a memory and a processor, which are interconnected. The memory stores computer instructions, and the processor executes the computer instructions to perform the intelligent inspection method for photovoltaic brackets based on UAV perception, as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the intelligent inspection method for photovoltaic brackets based on UAV perception as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, It includes computer instructions for causing a computer to execute the intelligent inspection method for photovoltaic brackets based on UAV perception, as described in any one of claims 1 to 6.

Images