Deformation detection method for integrally formed photovoltaic inverter box body

Abstract

The invention relates to the technical field of deformation measurement, in particular to a deformation detection method for an integrally-formed photovoltaic inverter box body, and solves the technical problem of misjudgment of box body deformation detection caused by position distortion due to external interference in the prior art. The method comprises the following steps: scanning to obtain sample point cloud data of a box body to be detected and standard point cloud data of a brand new box body; determining an abnormal region according to the offset degree of the sample point cloud data relative to the standard point cloud data; according to the data distribution condition of the abnormal area, identifying a deformation area without interference and a suspected deformation area with interference in the abnormal area; according to the offset degree and the signal reflection intensity of the suspected deformation area, correcting the data distribution condition, and identifying an interference deformation area and a non-deformation interference distortion area in the suspected deformation area; and determining the deformation degree of the to-be-detected box body and taking corresponding maintenance measures according to the proportion of the deformation area to the scanning area and the data distribution condition of the deformation area.

Description

Technical Field

[0001] This invention relates to the field of deformation measurement technology, and specifically to a method for detecting the deformation of an integrally molded photovoltaic inverter housing. Background Technology

[0002] Integrated photovoltaic inverters typically employ a highly efficient integrated design, which allows all functional modules of the inverter (such as DC input, AC output, control module, heat dissipation device, etc.) to be completed in a unified enclosure, thereby improving overall efficiency, reducing installation complexity, and enhancing the stability and reliability of the equipment.

[0003] During transportation, installation, or maintenance, the enclosure may be subjected to external impacts or vibrations, causing local deformation of the enclosure structure. Currently, when detecting enclosure deformation using laser scanning, point cloud data of the object being measured is generally acquired for evaluation. However, the point cloud data may be distorted due to external interference (surface characteristics, ambient light) during acquisition, which may lead to the non-deformed areas being misjudged as deformed areas, affecting the accuracy and reliability of laser scanning or other measurement techniques. Summary of the Invention

[0004] To address the technical problem in existing technologies where positional distortion caused by external interference leads to misjudgments in the detection of enclosure deformation, the present invention aims to provide a deformation detection method for an integrally molded photovoltaic inverter enclosure. The specific technical solution adopted is as follows: Firstly, a deformation detection method for an integrated photovoltaic inverter enclosure is provided, comprising: scanning three-dimensional point cloud data of the outer surface of the enclosure to obtain sample point cloud data of the enclosure to be tested and standard point cloud data of a brand-new enclosure; determining abnormal areas on the enclosure to be tested based on the offset of the sample point cloud data relative to the standard point cloud data; identifying non-interference deformed areas and suspected deformed areas with interference in the abnormal areas based on the data distribution of the sample point cloud data in the abnormal areas; correcting the data distribution based on the offset and signal reflection intensity of the sample point cloud data in the suspected deformed areas to identify interfering deformed areas and non-deformed interference distortion areas in the suspected deformed areas; determining the degree of deformation of the enclosure to be tested and taking corresponding maintenance measures based on the proportion of the deformed area to the scanned area and the data distribution of the deformed area.

[0005] Based on the above technical solution, in the deformation detection method for an integrated photovoltaic inverter enclosure provided by this invention, the abnormal area is first located through offset analysis to distinguish between potential deformation and interference. Then, data distribution characteristics are used to distinguish between interference-free deformation and suspected interference areas, initially isolating the interfered point cloud data. Finally, the data distribution is corrected by combining offset and signal reflection intensity as dual indicators, clearly distinguishing between real deformation with interference and pure interference distortion. This establishes an interference identification and elimination mechanism in the methodological process, optimizing the accuracy of detection of deformation areas due to interference from surface dust, rust, and light during laser scanning detection of enclosure deformation.

[0006] In conjunction with the first aspect above, in one possible implementation, the method for determining the abnormal region on the box to be detected based on the offset of the sample point cloud data relative to the standard point cloud data specifically includes: comparing the sample point cloud data and the standard point cloud data to obtain a set of abnormal data points offset from the standard position; clustering the data points in the abnormal data point set based on the offset of each data point in the abnormal data point set to obtain multiple data point clusters; and determining the abnormal region in the box region corresponding to the abnormal data point set based on the offset and number of data points in each data point cluster.

[0007] In conjunction with the first aspect above, in one possible implementation, the method for obtaining a set of abnormal data points that deviate from the standard position by comparing sample point cloud data and standard point cloud data specifically includes: determining the offset degree of the data points in the sample point cloud data from the standard position based on the Euclidean distance between the data points in the sample point cloud data and the data points in the standard point cloud data, and the distance between the data points in the sample point cloud data and the point cloud plane; and identifying data points in the sample point cloud data whose offset degree is greater than a preset offset degree threshold as abnormal data points.

[0008] In conjunction with the first aspect above, in one possible implementation, determining the abnormal region in the box region corresponding to the abnormal data point set based on the offset and number of data points in each data point cluster specifically includes: determining the detection confidence of each data point cluster based on the offset and number of data points in each data point cluster; and marking the box region corresponding to the data point cluster with a detection confidence greater than a preset confidence threshold as an abnormal region.

[0009] In conjunction with the first aspect mentioned above, in one possible implementation, the method for identifying interference-free deformed regions and interference-suspected deformed regions in an abnormal region based on the data distribution of sample point cloud data in the abnormal region specifically includes: analyzing the data distribution of sample point cloud data in the abnormal region to obtain a first distribution coefficient; if the first distribution coefficient is greater than a preset coefficient threshold, the abnormal region is identified as an interference-free deformed region; if the first distribution coefficient is less than or equal to the preset coefficient threshold, the abnormal region is identified as an interference-suspected deformed region.

[0010] In conjunction with the first aspect mentioned above, in one possible implementation, the method for analyzing the data distribution of sample point cloud data in the abnormal region to obtain the first distribution coefficient specifically includes: analyzing the sample point cloud density of multiple data points in the abnormal region and comparing it with the standard point cloud density of multiple data points in the corresponding region of the new box to determine the point cloud density difference coefficient of the abnormal region; comparing and analyzing the offset direction of multiple data points in the abnormal region to determine the data point continuity coefficient in the abnormal region; and determining the first distribution coefficient based on the point cloud density difference coefficient and the data point continuity coefficient.

[0011] In conjunction with the first aspect above, in one possible implementation, the method for correcting the data distribution based on the offset and signal reflection intensity of sample point cloud data in the suspected deformation region, and identifying the interfering deformation region and the non-deformed interference distortion region in the suspected deformation region, specifically includes: determining the deformation screening factor of the suspected deformation region based on the offset and signal reflection intensity of the sample point cloud data; correcting the first distribution coefficient based on the deformation screening factor to obtain the second distribution coefficient; if the second distribution coefficient is greater than a preset coefficient threshold, marking the suspected deformation region as an interfering deformation region; if the second distribution coefficient is less than or equal to the preset coefficient threshold, marking the suspected deformation region as a non-deformed interference distortion region.

[0012] In conjunction with the first aspect above, in one possible implementation, the method for determining the deformation screening factor of suspected deformation regions based on the offset and signal reflection intensity of sample point cloud data specifically includes: determining the signal reflection coefficient of each suspected deformation region based on the average signal reflection intensity of sample point cloud data in each suspected deformation region; determining the offset coefficient of each suspected deformation region based on the average offset of sample point cloud data in multiple suspected deformation regions; and determining the deformation screening factor based on the signal reflection coefficient and the offset coefficient.

[0013] In conjunction with the first aspect mentioned above, in one possible implementation, the method further includes: visually displaying the sample point cloud data of the box to be tested and identifying the deformation area.

[0014] In conjunction with the first aspect above, in one possible implementation, the method of taking corresponding maintenance measures specifically includes: when the deformation degree is severe, determining the processing strategy based on the functional attributes of the deformed area; sending an alarm message to the maintenance terminal; the alarm message includes: the deformed area, the deformation degree, and the processing strategy.

[0015] Secondly, a deformation detection device for a molded photovoltaic inverter enclosure is provided, comprising: a processor and a storage medium; the storage medium includes instructions, and the processor is used to execute the instructions to perform the actions described in the first aspect and any possible implementation thereof. This deformation detection device for the molded photovoltaic inverter enclosure can be an electronic device or a chip within an electronic device.

[0016] Thirdly, a computer-readable storage medium is provided, which stores instructions that, when executed on a deformation detection device for an integrally molded photovoltaic inverter enclosure, cause the deformation detection device for the integrally molded photovoltaic inverter enclosure to perform the actions described in the first aspect and any possible implementation thereof.

[0017] Fourthly, a computer program product containing instructions is provided, which, when running on a deformation detection device for an integrally molded photovoltaic inverter enclosure, causes the deformation detection device for the integrally molded photovoltaic inverter enclosure to perform the actions described in the first aspect and any possible implementation thereof.

[0018] The present invention has the following beneficial effects: First, offset analysis is used to locate abnormal areas and distinguish between potential deformations and interference. Then, data distribution characteristics are used to differentiate between interference-free deformations and suspected interference areas, initially isolating the interfered point cloud data. Finally, the data distribution is corrected by combining offset and signal reflection intensity indicators, clearly distinguishing between real deformations with interference and purely distorted interference. This methodology establishes an interference identification and elimination mechanism, optimizing the accuracy of laser scanning detection of box deformation areas due to interference from surface dust, rust, and light. Attached Figure Description

[0019] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating a method for detecting deformation of an integrally molded photovoltaic inverter housing according to an embodiment of the present invention. Figure 2 A schematic diagram of the appearance of an integrated photovoltaic inverter enclosure provided in one embodiment of the present invention; Figure 3A three-dimensional point cloud diagram of an integrally molded photovoltaic inverter enclosure provided in one embodiment of the present invention; Figure 4 A flowchart illustrating a method for detecting deformation of an integrally molded photovoltaic inverter housing according to an embodiment of the present invention; Figure 5 A flowchart illustrating a method for detecting deformation of an integrally molded photovoltaic inverter housing according to an embodiment of the present invention; Figure 6 A flowchart illustrating a method for detecting deformation of an integrally molded photovoltaic inverter housing according to an embodiment of the present invention; Figure 7 A flowchart illustrating a method for detecting deformation of an integrally molded photovoltaic inverter housing according to an embodiment of the present invention; Figure 8 A flowchart illustrating a method for detecting deformation of an integrally molded photovoltaic inverter housing according to an embodiment of the present invention; Figure 9 A flowchart illustrating a method for detecting deformation of an integrally molded photovoltaic inverter housing according to an embodiment of the present invention; Figure 10 A flowchart illustrating a method for detecting deformation of an integrally molded photovoltaic inverter housing according to an embodiment of the present invention; Figure 11 A flowchart illustrating a method for detecting deformation of an integrally molded photovoltaic inverter housing according to an embodiment of the present invention; Figure 12 A flowchart illustrating a method for detecting deformation of an integrally molded photovoltaic inverter housing according to an embodiment of the present invention; Figure 13 This is a schematic diagram of the hardware structure of a deformation detection device for an integrally molded photovoltaic inverter housing, provided in one embodiment of the present invention. Detailed Implementation

[0021] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a deformation detection method for an integrally molded photovoltaic inverter housing proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0022] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0023] The following describes in detail, with reference to the accompanying drawings, a specific scheme for a deformation detection method for an integrally molded photovoltaic inverter housing provided by the present invention.

[0024] Please see Figure 1 The diagram illustrates a method flowchart for detecting the deformation of an integrally molded photovoltaic inverter housing according to an embodiment of the present invention. This method includes: S1. Scan the three-dimensional point cloud data of the outer surface of the box to obtain the sample point cloud data of the box to be tested and the standard point cloud data of the brand new box.

[0025] The principle of laser scanning for detecting box deformation is based on non-contact 3D measurement technology. A laser beam is emitted from a laser device to scan the box surface. The laser scanner emits a laser beam towards the box surface; after encountering the box, the laser beam reflects back to the scanner. The distance traveled by the laser from emission to return is calculated to obtain the 3D coordinates of the corresponding point in space, thereby establishing a 3D point cloud of the box. For example,... Figure 2 The inverter enclosure shown is illustrated by point cloud data obtained through laser scanning, as shown below. Figure 3 As shown, the 3D point cloud data of the enclosure can be used to represent the true geometry and deformation of the enclosure surface, making it suitable for geometric accuracy inspection of large containers / enclosures.

[0026] In some implementations, the scanning equipment can use a high-precision laser scanner (scanning accuracy ±0.1mm, point cloud density ≥100 points / mm²) and be equipped with a 360° rotating scanning platform to ensure no blind spots on the outer surface of the enclosure. The scanning environment is carried out in a closed testing space, avoiding direct sunlight, airflow disturbances, and vibrations (such as by setting up a shockproof workbench), and the temperature is kept stable at 25±2℃ to reduce the temporary impact of thermal expansion and contraction on the enclosure.

[0027] In some implementations, three brand-new, unused boxes from the same batch can be selected, scanned three times each, and the nine sets of data can be fused using a point cloud registration algorithm to generate a standard point cloud model that eliminates individual differences, which can then be used as a benchmark reference.

[0028] The test chamber is scanned three times to remove noise points. For example, isolated points that deviate from the mean by three times the standard deviation are removed by statistical filtering to obtain the denoised sample point cloud.

[0029] Furthermore, a coordinate system is established for the two types of point clouds. For example, a three-dimensional coordinate system is established with the center of the mounting hole at the bottom of the box as the origin, and coarse registration is completed through the iterated closest points (ICP) algorithm to ensure that the two spatial positions are aligned.

[0030] In some implementations, the point cloud data of the integrated inverter enclosure to be inspected is obtained by laser scanning, while the reflection intensity and fitting residual corresponding to different data points are recorded. At the same time, the standard point cloud data and indicators of the same type of brand-new or smooth and undeformed enclosure are obtained by scanning.

[0031] S2. Determine the abnormal areas on the box to be inspected based on the offset of the sample point cloud data relative to the standard point cloud data.

[0032] In some implementations, outliers deviating from the standard position can be marked first, and spatial connectivity analysis can be performed on these outliers (e.g., by dividing the spatial region using an octree), merging the connected regions into outlier regions. This method is suitable for box structures with many curved surfaces and can better handle offset detection of complex geometries.

[0033] In other implementations, spatial gradient changes in offset can be used to identify anomalous regions. This involves meshing the sample point cloud and standard point cloud (e.g., a 1mm × 1mm 3D mesh) and calculating the average offset of each mesh. The offset gradient (i.e., the rate of change of offset) between adjacent meshes is calculated to generate a gradient field distribution map. When the gradient value continuously exceeds a threshold and the direction is consistent within a certain region, it is marked as a potential anomalous region. Combined with the average offset within the region, if the average offset is greater than 0.3mm, the anomalous region is finally determined. This method can effectively distinguish between continuous deformation (such as bending) and local bulges / depressions, making it suitable for detecting structural deformations of boxes.

[0034] S3. Based on the data distribution of sample point cloud data in the abnormal area, identify the non-interference deformation area and the suspected interference deformation area in the abnormal area.

[0035] In some implementations, since the point cloud normal vectors (reflecting surface orientation) of the actual deformed region exhibit a continuous changing trend, while point cloud offsets caused by interference are often accompanied by disordered jumps in normal vectors, such as chaotic normal vectors at local noise points, the normal vectors of each point within the abnormal region can be calculated and fitted to the local plane using the least squares method. Then, the angle between the normal vectors of adjacent points can be calculated, and the mean absolute value and standard deviation of the angles within the region can be statistically analyzed. The judgment rules include: if the mean angle is <10° and the standard deviation is <5°, it is judged as an interference-free deformed region (the normal vectors change continuously, conforming to the characteristics of physical deformation); if the mean angle is >30° or the standard deviation is >15°, it is judged as a region suspected of interference (the normal vectors change abruptly, possibly due to point cloud disorder caused by interference).

[0036] In other implementations, since the point cloud of a truly deformed region maintains continuous topological connectivity (e.g., the integrity of a triangular mesh), while the point cloud of an interference region often exhibits topological breaks or isolated fragments, a triangular mesh (topological connections between adjacent points in 3D space) can be constructed for the point cloud of anomaly regions. The proportion of elongated triangles (minimum angle < 10°) and the number of isolated points (number of connected edges < 3) in the triangular mesh are calculated. The judgment rules include: if the proportion of elongated triangles is < 10% and the number of isolated points is < 5%, it is judged as an interference-free deformed region (continuous and complete topological structure); if the proportion of elongated triangles is > 30% or the number of isolated points is > 15%, it is judged as a region with suspected interference (broken topological structure, possibly due to interference causing missing or disordered point clouds).

[0037] S4. Based on the offset and signal reflection intensity of the sample point cloud data in the suspected deformation area, correct the data distribution and identify the interference deformation area and the interference distortion area without deformation in the suspected deformation area.

[0038] In some implementations, the laser reflection intensity values ​​of each data point (a parameter built into the scanner, reflecting the light reflection characteristics of the surface material) are extracted, and the mean and standard deviation of the reflection intensity within the region are calculated. The mean reflection intensity of the corresponding region in the standard point cloud is compared. If the difference between the mean reflection intensity of the sample point cloud and the corresponding region in the standard point cloud is greater than twice the standard deviation of the standard region's reflection intensity, it indicates surface interference in that region (such as oil stains or scratches causing abnormal reflectivity). For points with suspected abnormal reflection intensity in the region, the weight of their offset in the region assessment is reduced, for example, by multiplying by a correction factor of 0.3, and spatial continuity and offset consistency are recalculated.

[0039] Disturbed deformation areas can be defined as those with corrected spatial continuity > 0.7 and offset direction standard deviation < 10°, which are judged as being disturbed but still having real deformation (such as stains on the surface of the deformation area). Interference distortion areas without deformation can be defined as: points with corrected spatial continuity < 0.5 and offset mainly from points with abnormal reflection intensity, which are judged as distortions caused by pure interference (such as dust occlusion during scanning).

[0040] S5. Based on the proportion of the deformed area to the scanned area in the box to be tested, and the data distribution of the deformed area, determine the degree of deformation of the box to be tested and take corresponding maintenance measures.

[0041] In some implementations, for any deformed region, the data distribution determines the severity of the local deformation, which can be determined by the average distribution coefficient of the deformed region. Indicates the degree of local deformation, average distribution coefficient A larger value indicates a greater degree of local deformation, and vice versa. The size of the abnormal region corresponds to the area of ​​deformation. Based on these characteristics, the formula for calculating the degree of deformation of this box-shaped sample is as follows: In the formula, This represents the total scanning area of ​​the box to be tested.

[0042] This represents the area of ​​the deformed region within the chamber to be tested.

[0043] This represents the proportion of the deformed area to the scanned area; the larger the value, the larger the deformed area.

[0044] This is the average distribution coefficient for all regions that experience localized deformation.

[0045] This is a normalization function used to normalize values ​​to the range [0,1].

[0046] This represents the degree of deformation of the enclosure to be inspected. The percentage of the scanned area relative to the deformed region is also considered. The larger the value, the more important it is to consider the severity of the deformation. ,Right now The larger the deformation, the more severe the deformation of the test chamber, and the more necessary the maintenance.

[0047] Because integrated inverter enclosures house numerous functional components, severe deformation of the enclosure may affect the functionality of these components. Therefore, selection is based on experience. The corresponding cabinet sample was found to be severely deformed and was replaced. If necessary, the function of each internal component can be tested individually, with a focus on the cabinet structure, internal electrical components, and heat dissipation system to ensure stable operation of the inverter.

[0048] Based on the above technical solution, the abnormal areas are first located through offset analysis to distinguish between potential deformation and interference. Then, data distribution characteristics are used to differentiate between interference-free deformation and suspected interference areas, initially isolating the interfered point cloud data. Finally, the data distribution is corrected by combining offset and signal reflection intensity as dual indicators, clearly distinguishing between real deformation with interference and pure interference distortion. An interference identification and elimination mechanism is established from a methodological perspective, optimizing the accuracy of laser scanning detection of box deformation areas due to interference from surface dust, rust, and light.

[0049] In one possible implementation, combining Figure 1 ,like Figure 4As shown, the method in S2 above can be specifically implemented through the following steps S21 to S23, which are explained in detail below: S21. Compare the sample point cloud data and the standard point cloud data to obtain a set of abnormal data points that are offset from the standard position.

[0050] Sample point cloud data may contain local deformation features of the box, while standard point cloud data represents the normal shape of the box. Therefore, the point cloud shapes corresponding to the two are superimposed and overlapped through a coordinate system. If the position of the data point at the corresponding position is different from that of the standard point, it indicates that the box position corresponding to this data point may have been deformed, and it is an abnormal data point.

[0051] In some implementations, the standard point cloud can be converted into a three-dimensional distance field (the shortest distance from each spatial point to the surface of the standard point cloud), and the distance value distribution of the sample point cloud in the distance field can be calculated. When the distance value of the sample point exceeds the threshold (e.g., 0.5 mm), it is marked as an outlier.

[0052] S22. Based on the offset of each data point in the abnormal data point set, cluster the data points in the abnormal data point set to obtain multiple data point clusters.

[0053] Since interfering point cloud data generally corresponds to a large offset and the local number of points is small, we can perform adaptive clustering based on the offset size for all abnormal data points. Data points with similar offsets can be grouped into one cluster. Interfering point clouds have poor concentration, and the corresponding data point clusters are smaller. Therefore, we can filter the point cloud based on different data point clusters.

[0054] In some implementations, the density-based spatial clustering of applications with noise (DBSCAN) algorithm can be used. This algorithm does not require a preset number of clusters and can adaptively identify point clusters of arbitrary shapes, making it suitable for irregular deformation areas that may occur on the surface of the container. The core parameter is set as follows: the neighborhood radius is determined based on the point cloud density. For example, when the point cloud density is 100 points / mm², the neighborhood radius is 0.8 mm, ensuring that adjacent outliers can be grouped into the same cluster.

[0055] Traverse the set of outlier data points. For each unclassified point, search for all other outlier points within its neighborhood radius. If the number of points in the neighborhood is greater than or equal to the minimum threshold within the cluster (e.g., 10), then set that point as the core point and recursively expand the points in its neighborhood to form a data point cluster.

[0056] For non-core points (i.e., the number of points in their neighborhood is less than the minimum threshold within the cluster), if they are in the neighborhood of a core point, they are assigned to the cluster of that core point; otherwise, they are marked as noise points and can be directly removed.

[0057] The result is multiple data point clusters, each representing a spatially contiguous region of outlier aggregation.

[0058] S23. Based on the offset and number of data points in each data point cluster, determine the abnormal area in the box area corresponding to the abnormal data point set.

[0059] In some implementations, dual filtering conditions are set, including: filtering by offset to ensure that the overall offset of the region is significant and to exclude point clusters with slight fluctuations; and filtering by the actual area corresponding to the size of the point cluster to exclude local interference areas that are too small.

[0060] For point clusters that meet the conditions, the region boundary is determined in the following way: calculate the minimum circumscribed cuboid of the point cluster, and use the boundary of the cuboid as the initial boundary of the abnormal region; smooth the boundary (e.g., by using Gaussian filtering) to remove the sharp corners to fit the continuous features of the actual deformed region.

[0061] Based on the above technical solution, by directly comparing the sample with the standard point cloud, all outlier points deviating from the standard position can be quickly screened out, avoiding the omission of potential deformation areas. Offset-based clustering analysis can aggregate outlier points with similar spatial locations and offset characteristics, effectively isolating the influence of noise points on the results. Combining offset and data point quantity as dual indicators to determine outlier regions makes the region boundaries clearer, providing structured data support for subsequent deformation judgment. It can effectively identify both continuous deformation (such as large-area depressions) and local deformation (such as bulges), and is applicable to different types of box deformation detection scenarios.

[0062] In one possible implementation, combining Figure 4 ,like Figure 5 As shown, the method in S21 above can be specifically implemented through the following S211 to S212, which are explained in detail below: S211. Based on the Euclidean distance between the data points in the sample point cloud data and the data points in the standard point cloud data, and the distance between the data points in the sample point cloud data and the point cloud plane, determine the offset degree of the data points in the sample point cloud data from the standard position.

[0063] In some implementations, data points The formula for calculating the offset from the standard position is as follows: In the formula, For data points coordinate data, For standard point cloud and data points The coordinates of the nearest point.

[0064] Represents data points The larger the Euclidean distance relative to the standard point, the higher the value of the corresponding data point. The greater the offset, the more abnormal the data points.

[0065] For data points With its location in the point cloud plane The straight-line distance, at the same time Represents data points Relative to the plane The offset value is larger, and the greater the unevenness of the point position, the more likely it is to be an abnormal data point. By constraining the plane distance of the point cloud, it is possible to effectively distinguish between true deformation points (deviating from the corresponding standard point and also deviating from the overall surface trend) and local interference points (only a single point deviates without destroying the overall surface trend).

[0066] For data points The offset from the standard position, when averaged, is a quantification of the balance between local and overall trend offsets. Specifically, the greater the Euclidean distance of a data point relative to the standard point... At the same time, combined with data points Relative to the plane offset ,Right now The larger the value, the greater the deviation of the data point from the standard position.

[0067] Euclidean distance reflects the direct positional deviation between points, capturing subtle local shifts; point cloud planar distance (usually the perpendicular distance from a sample point to the local fitting plane of the standard point cloud) reflects the degree of deviation of a point from the overall surface trend, better reflecting the overall characteristics of physical deformation (e.g., the bending of a box surface causes a local area to deviate from the original plane, rather than the shift of a single discrete point). The combination of the two means that the offset includes not only absolute positional deviation but also deviation relative to the overall trend, more accurately quantifying the difference between real deformation (such as depressions and bulges) and normal surface fluctuations.

[0068] S212. Data points in the sample point cloud data whose offset is greater than a preset offset threshold are identified as abnormal data points.

[0069] In some implementations, the threshold can be dynamically adjusted based on factors such as the material of the enclosure (e.g., the stiffness difference between metal and plastic), the usage scenario (e.g., the difference in environmental interference between outdoor and indoor environments), and the precision of the equipment, taking into account both flexibility and standardization. For example, the threshold can be set to 0.5mm for basic equipment and 0mm for precision equipment.

[0070] The offset calculated based on dual distances already contains information about the overall trend deviation. Therefore, threshold screening can more accurately identify points that truly deviate from the normal structure, avoiding over-screening (missed detection) or under-screening (false detection) caused by a single distance threshold.

[0071] Based on the above technical solution, by using dual distance fusion to calculate offset and threshold filtering, the accuracy of offset measurement is improved while the anti-interference capability is enhanced. It effectively distinguishes between real deformation and local interference, providing high-quality basic data for subsequent abnormal area identification, and ultimately improving the overall reliability of box deformation detection.

[0072] In one possible implementation, combining Figure 4 ,like Figure 6 As shown, the method in S23 above can be specifically implemented through the following S231 to S232, which are explained in detail below: S231. Determine the detection confidence level of each data point cluster based on the offset and number of data points in each data point cluster.

[0073] In some implementations, the formula for calculating the detection confidence of data point clusters is as follows: In the formula, For data point clusters The average offset of all data points in the dataset. Since the offset of true deformation is usually structural and continuous, the offset will not be extremely large, while interference points (such as noise and surface stains) often exhibit localized, discrete, and large offsets. Therefore, the data point cluster... The larger the average offset of all data points in the cluster, the greater the offset, and the more likely the data points in the cluster are to be interfering point cloud data points.

[0074] For data point clusters The data scale, i.e., the number of data points. It is the largest of all data point clusters.

[0075] Represents a cluster of data points The relative scale. Since true deformation is a physical change in the box structure, it is usually manifested as a large number of points and a large scale in a continuous region; while interference points (such as isolated noise) are mostly discretely distributed, with fewer points and a smaller scale. Therefore, the smaller the relative scale, the fewer data points are contained in the cluster, satisfying the discreteness of the interference point cloud.

[0076] For a certain data point cluster The detection confidence is determined by a normalization function. The correlation between offset and scale is normalized to a confidence level in the [0, 1] interval, thereby enabling the fusion judgment of multi-dimensional features.

[0077] S232. Mark the box area corresponding to the cluster of data points with a detection confidence level greater than the preset confidence threshold as an abnormal area.

[0078] To achieve adaptive differentiation between deformation clusters and interference clusters, a threshold filtering method based on quantized confidence levels is used (e.g., ...). This reduces false positives (misidentifying interference as deformation) and avoids false negatives (misidentifying deformation as interference), ultimately resulting in accurate anomaly region annotations.

[0079] Based on the above technical solution, the confidence level and threshold are calculated by multi-dimensional feature fusion to locate abnormal areas, thereby solving the core problem of difficulty in distinguishing between interference clusters and real deformation clusters in laser scanning detection.

[0080] In one possible implementation, combining Figure 1 ,like Figure 7 As shown, the method in S3 above can be specifically implemented through the following steps S31 to S33, which are explained in detail below: S31. Analyze the data distribution of sample point cloud data in the abnormal area to obtain the first distribution coefficient.

[0081] Since point cloud offset will produce the above-mentioned abnormal areas, and point cloud distortion will also cause point cloud offset in the corresponding areas, the abnormal areas not only include the box deformation area but also the area where these interfering factors exist. Therefore, it is necessary to filter the point cloud corresponding to the abnormal area based on the difference in point cloud distribution between the two.

[0082] In some implementations, the DBSCAN clustering algorithm is used to cluster the regional point cloud, and the proportion of the number of points in the largest cluster to the total number of points in the region is counted. The closer the proportion is to 1, the stronger the spatial continuity of the point cloud.

[0083] For each point within the region, calculate its offset vector (spatial position difference) relative to the standard point cloud. Perform principal component analysis on all offset vectors and calculate the variance contribution rate of the first principal component. The closer the variance contribution rate of the first principal component is to 1, the more concentrated the offset direction is.

[0084] The density of the point cloud within the calculation area is calculated, and the coefficient of variation (standard deviation / mean) of the density is calculated. The closer the coefficient of variation is to 0, the more uniform the point cloud density is.

[0085] A weighted formula is used to integrate multi-dimensional indicators (the weights can be adjusted according to the scenario), for example: First distribution coefficient = 0.4 × the proportion of the number of points in the largest cluster to the total number of points in the region + 0.3 × the variance contribution rate of the first principal component + 0.3 × (1 - the coefficient of variation of density).

[0086] At this point, the range of the first distribution coefficient is (0, 1), and the closer it is to 1, the more the distribution conforms to the true deformation characteristics.

[0087] S32. If the first distribution coefficient is greater than the preset coefficient threshold, the abnormal area is marked as an interference-free deformation area.

[0088] In some implementations, the preset coefficient threshold can be set according to the box material and detection accuracy. For example, the threshold for metal boxes is set to 0.7, and the threshold for plastic boxes is set to 0.6.

[0089] S33. If the first distribution coefficient is less than or equal to the preset coefficient threshold, the abnormal area is marked as a suspected deformation area with interference.

[0090] Not all anomalous regions that do not meet this threshold condition originate from external interference factors (such as point cloud distortion caused by ambient light). In actual detection, some anomalous regions may simultaneously exhibit a mixture of local deformation of the box and surface interference. For example, if the box bulges or depressions while its surface has high light reflection intensity, it will cause point cloud distortion. In such mixed scenarios, surface interference will significantly alter the microscopic distribution characteristics of the point cloud, specifically manifesting as abnormally sparse point cloud density, deviating from the standard density, or reduced data point continuity, leading to an underestimation of the first distribution coefficient, with its value falling below the threshold.

[0091] If only this threshold is used for screening, such mixed areas will be incorrectly classified as non-deformation areas, resulting in missed deformation detections and thus reducing the reliability of the detection results. In other words, the above screening steps will produce errors. Therefore, it is necessary to analyze the reflection intensity and regional change characteristics of the remaining abnormal areas for accurate screening.

[0092] Based on the above technical solution, since the point cloud of real deformation has characteristics such as spatial continuity, consistency of offset direction, and uniform density, by quantifying the distribution characteristics of point cloud data and setting a threshold, it is possible to effectively distinguish between two types of regions: non-interference deformation regions and interference-suspected deformation regions, thereby reducing the risk of misjudging interference as deformation or missing deformation as interference.

[0093] In one possible implementation, combining Figure 7 ,like Figure 8 As shown, the method in S31 above can be specifically implemented through the following S311 to S313, which are explained in detail below: S311. Analyze the sample point cloud density of multiple data points in the abnormal area and compare it with the standard point cloud density of multiple data points in the corresponding area of ​​the new box to determine the point cloud density difference coefficient of the abnormal area.

[0094] Generally, local deformation of the box can present surface shapes such as bulges, depressions, and bends. The point cloud data anomalies caused by these have microscopic continuity. At the same time, the point cloud distribution density in the anomaly area remains unchanged. However, due to the discontinuity of point cloud anomalies caused by point cloud distortion, the distribution density of the point cloud becomes sparser.

[0095] In some implementations, the formula for calculating the point cloud density difference coefficient is as follows: In the formula, Abnormal area Point cloud density, This represents the point cloud density under normal standard conditions.

[0096] True deformation (such as bulges and depressions) is a physical change in the box structure and does not change the point cloud density. ≈ ),therefore As it approaches 0, its reciprocal approaches its maximum value.

[0097] Point cloud distortion (such as point cloud anomalies caused by interference or noise) can make the point cloud density sparser. Less than ),therefore It is significantly greater than 0, and its reciprocal is significantly smaller than the actual deformation region.

[0098] For extremely small positive values, such as 0.001, because There is a possibility that the result will be 0. By adding a very small positive value, we can avoid the case where the reciprocal is meaningless. In addition, the effect of the minimum value on the calculation result can be ignored.

[0099] The point cloud density difference coefficient effectively suppresses the interference of point cloud distortion (sparse density) through density proximity screening, highlighting the characteristic of constant density under true deformation.

[0100] S312. Compare and analyze the offset direction of multiple data points in the abnormal region to determine the continuity coefficient of data points in the abnormal region.

[0101] In some implementations, the formula for calculating the data point continuity coefficient is as follows: In the formula, and These are abnormal regions. The maximum and minimum principal component vectors in principal component analysis (PCA).

[0102] Point clouds with real deformation have microscopic continuity (such as point clouds in raised regions being continuously distributed along the deformation direction), so the angle between the principal component vectors of PCA is small and the cosine value approaches 1.

[0103] Point cloud distortion (such as point cloud disorder caused by interference) lacks continuity, the angle between the principal component vectors of PCA is large, and the cosine value is significantly less than 1.

[0104] The data point continuity coefficient effectively suppresses the interference of point cloud distortion (discontinuity) through spatial continuity filtering, highlighting the characteristics of continuous distribution of true deformation.

[0105] S313. Determine the first distribution coefficient based on the point cloud density difference coefficient and the data point continuity coefficient.

[0106] In some implementations, the first distribution coefficient The calculation formula is as follows: In the formula, For a certain abnormal region The data point distribution coefficient.

[0107] The true deformation region simultaneously satisfies the conditions of near-normal density and continuous point cloud, therefore The value is significantly larger than expected.

[0108] The interference area either has large density differences or poor continuity. The value is significantly smaller than expected.

[0109] Furthermore, the normalized value can be adjusted by applying a threshold. Perform threshold filtering (e.g., setting a normalized threshold) (For true deformation), it can achieve quantitative differentiation between the two types of regions.

[0110] Based on the above technical solution, the essential difference between real deformation and interference areas is accurately characterized by multi-dimensional fusion of density features and continuity features. This provides a quantitative basis for the classification of abnormal areas (deformation without interference / suspected interference), effectively solving the technical pain point of difficulty in distinguishing between deformation and interference in laser scanning, and improving the accuracy and reliability of box deformation detection.

[0111] In one possible implementation, combining Figure 8 ,like Figure 9 As shown, the method in S4 above can be specifically implemented through the following steps S41 to S44, which are explained in detail below: S41. Determine the deformation screening factor for suspected deformation areas based on the offset and signal reflection intensity of the sample point cloud data.

[0112] In some implementations, the spatial offset (Euclidean distance) of all sample points within the suspected deformation area relative to the standard point cloud is calculated, and the average value is taken to reflect the intensity of physical deformation in the area.

[0113] Extract the laser reflection intensity (a built-in parameter of the scanner, reflecting the light reflection characteristics of the surface material) from all points within the suspected area, and calculate the average reflection intensity of the area. Simultaneously, obtain the average reflection intensity of the standard area (a normal, undisturbed box area). Calculate the reflection intensity difference rate.

[0114] The design incorporates a fusion formula that combines offset and reflection intensity difference rate to obtain the deformation screening factor.

[0115] S42. Based on the deformation screening factor, the first distribution coefficient is corrected to obtain the second distribution coefficient.

[0116] The screening factor for the deformed region with interference is significantly larger than that for the pure distorted region. Therefore, the distribution coefficient of all abnormal regions that do not meet the conditions can be corrected based on the screening factor, so that the size of the distribution coefficient can more accurately reflect the deformation characteristics of the abnormal region, thereby improving the accuracy of the deformation region screening.

[0117] In some implementations, the formula for calculating the second distribution coefficient is as follows: In the formula, Suspected deformation area The first distribution coefficient.

[0118] Suspected deformation area The deformation screening factor has a value range of [0,1].

[0119] Suspected deformation area The second distribution coefficient, since the deformation screening factor highlights the significant characteristics of the mixed anomaly region, can be used as a weight to make the characteristics of the mixed anomaly region more apparent when the distribution coefficient is weighted. Larger areas showed no significant change, while areas covered only by dust or oil showed no change. Deformation screening factor The larger the value (the higher the probability of actual deformation), the greater the correction to the first distribution coefficient, making... It better reflects the characteristics of real deformation.

[0120] Based on the description in S5, when calculating the average distribution coefficient of the deformation region, the first distribution coefficient is used for the deformation region without interference, and the modified second part coefficient is used for the deformation region with interference.

[0121] S43. If the second distribution coefficient is greater than the preset coefficient threshold, the suspected deformation area is marked as an interference deformation area.

[0122] Furthermore, the normalized result can be adjusted using a threshold. Perform with the first distribution coefficient Using the same threshold for screening, it is possible to quantitatively distinguish between two types of regions: deformed regions with interference and deformed regions with interference.

[0123] If the second distribution coefficient is greater than the preset coefficient threshold, it indicates that there is real deformation in the region and the surface interference does not completely cover the deformation characteristics, and it is then judged as a deformed region with interference.

[0124] S44. If the second distribution coefficient is less than or equal to the preset coefficient threshold, the suspected deformation area is marked as an interference distortion area without deformation.

[0125] If the second distribution coefficient is less than or equal to the preset coefficient threshold, it means that the point cloud anomaly in the region is entirely caused by surface interference (such as stains or reflections) and there is no real deformation. In this case, it is determined to be an interference distortion region without deformation.

[0126] Based on the above technical solution, a deformation screening factor is obtained through the fusion analysis of offset (reflecting deformation intensity) and reflection intensity (reflecting the degree of surface interference). This effectively distinguishes between two types of scenarios: those with surface interference but actual deformation (such as depressions covered by stains) and those with point cloud anomalies caused by pure interference (such as false offsets caused by reflections). This avoids misjudging real deformation as interference or pure interference as deformation. The first distribution coefficient may be underestimated due to surface interference (such as stains causing a deterioration in the continuity of point cloud features). After correction by the deformation screening factor, the second distribution coefficient more closely matches the true deformation attributes of the region, making the judgment results more reliable and significantly reducing the misjudgment rate in interference scenarios.

[0127] In one possible implementation, combining Figure 9 ,like Figure 10 As shown, the method in S41 described above can be specifically implemented through the following steps S411 to S413, which are explained in detail below: S411. Determine the signal reflection coefficient of each suspected deformation region based on the average signal reflection intensity of the sample point cloud data in each suspected deformation region.

[0128] The deformation region itself may cause changes in the laser scanning angle, which may lead to attenuation of the reflected signal, thus reducing the reflection intensity of the region. In the case of a mixture of deformation and distortion, the two may work together, resulting in a severe weakening of the reflected signal. At the same time, the superposition of distortions will lead to more severe thickness shifts, resulting in a larger fitting residual (offset) when fitting the point cloud of the corresponding region.

[0129] In some implementations, the signal reflection coefficient The calculation formula is as follows: In the formula, This refers to the average reflection intensity of all data points within the anomalous region.

[0130] Actual deformation (such as bulges or depressions) will change the laser scanning angle, causing attenuation of the reflected signal. reduce, The larger the value, the more likely it is to be a mixed anomaly region.

[0131] S412. Determine the offset coefficient of each suspected deformation region based on the average offset of the sample point cloud data in multiple suspected deformation regions.

[0132] In some implementations, the offset coefficient The calculation formula is as follows: In the formula, This represents the average data point offset of all data points in the suspected deformation area.

[0133] This represents the minimum average data point offset across all suspected deformation regions.

[0134] The significance of regional offset is quantified by the ratio of regional offset to minimum offset. The larger the ratio, the greater the degree of data point offset in the corresponding abnormal region. Since the offset of true deformation is usually significantly greater than the false offset caused by interference (such as the random offset of noise points), it is more likely to be a mixed abnormal region.

[0135] S413. Determine the deformation screening factor based on the signal reflection coefficient and the offset coefficient.

[0136] In some implementations, the deformation screening factor is calculated using the following formula: In the formula, It serves as a deformation screening factor for a certain abnormal region, integrating two core features: the degree of reflection attenuation and the significance of offset.

[0137] Mixed anomaly regions (containing both deformation and interference) will simultaneously exhibit significant attenuation of reflection intensity and significant shift, therefore The values ​​are significantly larger. In contrast, the pure interference region (without actual deformation) only exhibits abnormal reflection intensity or a small offset. The value is significantly smaller. (Through...) Threshold filtering can achieve quantitative differentiation between mixed abnormal regions and pure interference regions.

[0138] Based on the above technical solution, by integrating the reflection intensity attenuation and the significance of the offset in multiple dimensions, the essential difference between the real deformation area with interference and the pure interference area is accurately characterized. This provides a quantitative basis for the secondary screening of abnormal areas (distinguishing between interference deformation and pure interference), effectively solving the problem of misjudgment caused by the superposition of deformation and interference in laser scanning, and improving the accuracy and reliability of box deformation detection.

[0139] In one possible implementation, combining Figure 1 ,like Figure 11 As shown, the above method may also include the following S6, which will be explained in detail below: S6. Visualize the point cloud data of the sample box to be tested and mark the deformation area.

[0140] In some implementations, professional point cloud visualization software (such as CloudCompare and ParaView) or self-developed 3D visualization modules are used, which need to support point cloud rendering, region annotation, interactive operation, and multi-data layer overlay.

[0141] Region types are distinguished by color coding, for example: Unaffected deformation areas are rendered in red, and the point cloud density is set to high to highlight spatial continuity; Areas with interfering deformation are rendered in orange with 50% opacity to indicate the presence of surface interference; The interference distortion area is rendered in gray, the point cloud density is reduced, and it is visually distinguished from the effective distortion area.

[0142] Furthermore, it can also support interactive feature design, including: Multi-view control: Supports free rotation, zoom, and panning, and allows one-click switching between preset top / side / bottom views to fully observe the deformation area; Information Query: Click on any point or area to bring up the attribute panel, which displays key parameters such as the average offset, reflection intensity, and distribution coefficient of the area; Multi-data layer comparison: Supports overlaying sample point clouds with standard point clouds (standard point clouds are set to semi-transparent blue) to intuitively compare the differences between deformed and normal areas.

[0143] Based on the above technical solution, complex point cloud data and deformed areas are transformed into intuitive 3D graphics. Technicians can quickly understand the location, range, and degree of deformation without analyzing numerical values, lowering the technical threshold. This also facilitates subsequent physical inspection, repair, or cause tracing, improving problem-solving efficiency.

[0144] In one possible implementation, combining Figure 1 ,like Figure 12As shown, the method for taking the corresponding maintenance measures in S5 above can be specifically implemented through the following S51 to S52, which are explained in detail below: S51. When the deformation degree is severe, the processing strategy is determined according to the functional attributes of the deformed area.

[0145] In some implementations, the photovoltaic inverter enclosure is pre-divided into functional zones, clearly defining the criticality and failure impact of each zone, including: Electrical component installation area: This area includes the inverter core module, wiring terminals, etc. Deformation can easily lead to loose electrical connections and insulation failure. Heat dissipation functional area: including heat dissipation fins, vents, etc. Deformation can easily block the heat dissipation channel and cause the equipment to overheat; Structural support area: This includes the box frame, fixing holes, etc. Deformation can easily damage the overall structural stability, leading to box collapse or component detachment.

[0146] Regional processing strategies include: If the electrical component installation area is severely deformed, immediately remove the electrical components in that area and replace them with cabinet modules of the same specifications; simultaneously test the insulation and conductivity of the components, and replace any damaged components as well.

[0147] The heat dissipation functional area was severely deformed. After cleaning the heat dissipation channels, the heat dissipation structure was repaired using sheet metal technology. After repair, the heat dissipation efficiency was tested. If it was less than 80% of the design value, the heat dissipation module was replaced.

[0148] If the structural support area is severely deformed and the structural safety factor is <0.8, the enclosure should be replaced directly; if the safety factor is between 0.8 and 1.0, the structural strength should be retested after carbon fiber reinforcement, and the enclosure can continue to be used only after it meets the standards.

[0149] S52. Send an alarm message to the maintenance terminal.

[0150] The alarm messages include: the area of ​​deformation, the degree of deformation, and the handling strategy.

[0151] Some implementations support cross-platform collaboration between mobile and PC management systems. Hierarchical access control is also implemented; for example, technical experts can view all data and policy configurations, while frontline maintenance personnel can only see execution-level operation procedures and real-time alarms.

[0152] Based on the above technical solutions, by making differentiated decisions based on functional attributes and providing real-time alarms, we can formulate exclusive handling strategies for severe deformation of different functional areas (electrical area, heat dissipation area, and structural area) of the enclosure. This not only prevents the risk of short circuits and leakage caused by delayed handling in critical functional areas (such as electrical component areas), but also avoids excessive maintenance of non-critical areas, thereby improving resource utilization and maintenance effectiveness, and achieving precise maintenance and efficient response in severe deformation scenarios.

[0153] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0154] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0155] In this embodiment of the invention, the deformation detection device for an integrated photovoltaic inverter housing can be divided into functional units according to the above method example. For example, each function can be divided into its own functional unit, or two or more functions can be integrated into one processing unit. The integrated unit can be implemented in hardware or software. It should be noted that the unit division in this embodiment is illustrative and only represents one logical functional division; other division methods may be used in actual implementation.

[0156] This invention also provides a hardware structure diagram of a deformation detection device for an integrated photovoltaic inverter housing, see [link / reference]. Figure 13 The deformation detection device 1300 for the integrated photovoltaic inverter housing includes a processor 1301, and optionally, a memory 1302 connected to the processor 1301.

[0157] In the first possible implementation, see Figure 13 The deformation detection device 1300 for the integrated photovoltaic inverter enclosure also includes a transceiver 1303. The processor 1301, memory 1302, and transceiver 1303 are connected via a bus. The transceiver 1303 is used to communicate with other devices or communication networks. Optionally, the transceiver 1303 may include a transmitter and a receiver. The device in the transceiver 1303 that performs the receiving function can be considered as a receiver, which is used to execute the receiving steps in the embodiments of the present invention. The device in the transceiver 1303 that performs the transmitting function can be considered as a transmitter, which is used to execute the transmitting steps in the embodiments of the present invention.

[0158] Based on the first possible implementation method Figure 13 The structural diagram shown can be used to illustrate the structure of the deformation detection device for the integrally molded photovoltaic inverter housing involved in the above embodiments.

[0159] in, Figure 13The diagram also illustrates the system chip in the deformation detection device for the integrated photovoltaic inverter enclosure. In this case, the actions performed by the aforementioned deformation detection device for the integrated photovoltaic inverter enclosure can be implemented by this system chip. The specific actions performed can be found above and will not be repeated here.

[0160] In implementation, each step of the method provided in this embodiment can be completed by integrated logic circuits in the processor or by instructions in software form. The steps of the method disclosed in this embodiment can be directly manifested as being executed by a hardware processor, or being executed by a combination of hardware and software modules in the processor.

[0161] The processor in this invention may include, but is not limited to, at least one of the following: a central processing unit (CPU), a microprocessor, a digital signal processor (DSP), a microcontroller unit (MCU), or an artificial intelligence processor, etc., which are various computing devices that run software. Each computing device may include one or more cores for executing software instructions to perform calculations or processing. The processor may be a standalone semiconductor chip or integrated with other circuits into a single semiconductor chip. For example, it may be integrated with other circuits (such as encoding / decoding circuits, hardware acceleration circuits, or various bus and interface circuits) to form a System-on-a-Chip (SoC), or it may be integrated as a built-in processor within an ASIC. The ASIC with the integrated processor may be packaged separately or together with other circuits. In addition to the cores for executing software instructions to perform calculations or processing, the processor may further include necessary hardware accelerators, such as field-programmable gate arrays (FPGAs), programmable logic devices (PLDs), or logic circuits that implement dedicated logic operations.

[0162] The memory in the embodiments of the present invention may include at least one of the following types: read-only memory (ROM) or other types of static storage devices capable of storing static information and instructions; random access memory (RAM) or other types of dynamic storage devices capable of storing information and instructions; or electrically erasable programmable read-only memory (EEPROM). In some scenarios, the memory may also be a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but is not limited thereto.

[0163] This invention also provides a computer-readable storage medium including instructions that, when run on a computer, cause the computer to perform any of the methods described above.

[0164] This invention also provides a computer program product containing instructions that, when run on a computer, cause the computer to perform any of the methods described above.

[0165] This invention also provides a chip, which includes a processor and an interface circuit. The interface circuit is coupled to the processor. The processor is used to run computer programs or instructions to implement the above-described method. The interface circuit is used to communicate with other modules outside the chip.

[0166] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented using software programs, implementation can be, in whole or in part, in the form of a computer program product. This computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device containing one or more servers, data centers, etc., that can be integrated with the medium. The available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state disks (SSDs)).

[0167] Although the invention has been described herein in conjunction with various embodiments, those skilled in the art, by reviewing the accompanying drawings and the disclosure, will understand and implement other variations of the disclosed embodiments in carrying out the claimed invention. In this invention, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude a plurality. A single processor or other unit can implement several of the functions listed in this invention.

[0168] Although the invention has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made therein without departing from the spirit and scope of the invention. Accordingly, this specification and drawings are merely illustrative of the invention and are to be considered as covering any and all modifications, variations, combinations, or equivalents within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its spirit and scope. Thus, if such modifications and modifications of the invention fall within the scope of the invention and its equivalents, the invention is also intended to include such modifications and modifications.

Claims

1. A method for detecting the deformation of an integrally molded photovoltaic inverter housing, characterized in that, include: Scan the three-dimensional point cloud data of the outer surface of the box to obtain the sample point cloud data of the box to be tested and the standard point cloud data of the brand new box. Based on the offset of the sample point cloud data relative to the standard point cloud data, the abnormal area on the box to be detected is determined; Based on the data distribution of sample point cloud data in the abnormal region, identify the non-interference deformation region and the suspected interference deformation region in the abnormal region. Based on the offset and signal reflection intensity of the sample point cloud data in the suspected deformation area, the data distribution is corrected to identify the interference deformation area and the interference distortion area without deformation in the suspected deformation area. Based on the proportion of the deformed area to the scanned area in the box to be tested, and the data distribution of the deformed area, the degree of deformation of the box to be tested is determined and corresponding maintenance measures are taken.

2. The deformation detection method for an integrally molded photovoltaic inverter housing according to claim 1, characterized in that, The step of determining the abnormal region on the box to be detected based on the offset of the sample point cloud data relative to the standard point cloud data includes: By comparing the sample point cloud data and the standard point cloud data, a set of abnormal data points offset from the standard position is obtained; Based on the offset of each data point in the abnormal data point set, the data points in the abnormal data point set are clustered to obtain multiple data point clusters; Based on the offset and number of data points in each data point cluster, the abnormal region in the box area corresponding to the abnormal data point set is determined.

3. The deformation detection method for an integrally molded photovoltaic inverter housing according to claim 2, characterized in that, The comparison of the sample point cloud data and the standard point cloud data yields a set of abnormal data points offset from the standard position, including: Based on the Euclidean distance between the data points in the sample point cloud data and the data points in the standard point cloud data, and the distance between the data points in the sample point cloud data and the point cloud plane, the offset degree of the data points in the sample point cloud data from the standard position is determined. In the sample point cloud data, data points with an offset greater than a preset offset threshold are identified as abnormal data points.

4. The deformation detection method for an integrally molded photovoltaic inverter housing according to claim 2, characterized in that, The step of determining the abnormal region within the box region corresponding to the abnormal data point set based on the offset and number of data points in each data point cluster includes: The detection confidence level of each data point cluster is determined based on the offset and number of data points in each cluster. The data point clusters with a detection confidence level greater than a preset confidence threshold are identified as abnormal areas within the corresponding box region.

5. The deformation detection method for an integrally molded photovoltaic inverter housing according to claim 1, characterized in that, The step of identifying interference-free deformation regions and suspected deformation regions with interference within the abnormal region based on the data distribution of sample point cloud data in the abnormal region includes: Analyze the data distribution of the sample point cloud data in the abnormal region to obtain the first distribution coefficient; If the first distribution coefficient is greater than a preset coefficient threshold, the abnormal region is marked as an interference-free deformation region; If the first distribution coefficient is less than or equal to the preset coefficient threshold, the abnormal region is identified as a suspected deformation region with interference.

6. The deformation detection method for an integrally molded photovoltaic inverter housing according to claim 5, characterized in that, The analysis of the data distribution of sample point cloud data in the abnormal region yields a first distribution coefficient, including: The sample point cloud density of multiple data points in the abnormal region is analyzed and compared with the standard point cloud density of multiple data points in the corresponding region of the new box to determine the point cloud density difference coefficient of the abnormal region. By comparing and analyzing the offset directions of multiple data points in the abnormal region, the continuity coefficient of the data points in the abnormal region is determined. The first distribution coefficient is determined based on the point cloud density difference coefficient and the data point continuity coefficient.

7. The deformation detection method for an integrally molded photovoltaic inverter housing according to claim 6, characterized in that, The step of correcting the data distribution based on the offset and signal reflection intensity of the sample point cloud data in the suspected deformation region, and identifying the interference-affected deformation region and the interference distortion region without deformation in the suspected deformation region, includes: Based on the offset and signal reflection intensity of the sample point cloud data, the deformation screening factor of the suspected deformation region is determined; Based on the deformation screening factor, the first distribution coefficient is corrected to obtain the second distribution coefficient; If the second distribution coefficient is greater than the preset coefficient threshold, the suspected deformation region is identified as a deformation region with interference. If the second distribution coefficient is less than or equal to the preset coefficient threshold, the suspected deformation region is identified as an interference distortion region without deformation.

8. The deformation detection method for an integrally molded photovoltaic inverter housing according to claim 7, characterized in that, The step of determining the deformation screening factor for the suspected deformation region based on the offset and signal reflection intensity of the sample point cloud data includes: The signal reflection coefficient of each suspected deformation region is determined based on the average signal reflection intensity of the sample point cloud data in each suspected deformation region. Based on the average offset of the sample point cloud data in multiple suspected deformation regions, determine the offset coefficient of each suspected deformation region; The deformation screening factor is determined based on the signal reflection coefficient and the offset coefficient.

9. The deformation detection method for an integrally molded photovoltaic inverter housing according to claim 1, characterized in that, Also includes: The sample point cloud data of the box to be tested is visualized and the deformation area is identified.

10. The deformation detection method for an integrally molded photovoltaic inverter housing according to claim 1, characterized in that, The corresponding maintenance measures include: When the degree of deformation is severe, the processing strategy is determined based on the functional attributes of the deformed area; Send an alarm message to the maintenance terminal; the alarm message includes: the deformed area, the degree of deformation, and the processing strategy.

Images