Optical on-line detection method and system for laser-etched line breakage of photovoltaic module

Abstract

The application provides an optical online detection method and system for broken lines of laser etching lines of a photovoltaic module, and relates to the technical field of photovoltaic module detection. The method comprises the following steps: activating a visible light image acquisition module and a spectrum detection module, performing scanning acquisition, performing spatial feature extraction and contour enhancement processing on a morphological texture image, calculating an explicit broken line criterion, establishing a first detection result, performing multi-segment energy ratio analysis and absorption peak shift detection on a spectrum data set, and after spatial constraint through an etching line connection matrix, establishing a second detection result, performing electrical excitation testing of the photovoltaic module, establishing a third detection result according to an excitation response set, performing interactive verification on the first, second and third detection results, and outputting an online detection result. The technical problem of low accuracy and reliability of broken line detection of laser etching lines of the photovoltaic module in the prior art is solved. The technical effect of effectively improving the accuracy and reliability of broken line monitoring of laser etching lines of the photovoltaic module is achieved.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic module testing technology, specifically to an optical online detection method and system for laser etching line breakage in photovoltaic modules. Background Technology

[0002] During the laser etching process of photovoltaic modules, the high-power laser can cause the etched lines to break or become discontinuous, severely affecting the electrical performance and quality safety of the photovoltaic modules. Currently, the detection of broken lines in laser etching of photovoltaic modules mainly relies on visual inspection and image recognition technologies to acquire and analyze images of the photovoltaic modules to obtain detection results. However, the accuracy of existing technologies is often affected by the production environment, such as temperature, humidity, and dust, making it difficult to effectively detect minute breaks or localized fractures. This leads to misjudgments and omissions in the detection results, thereby reducing the accuracy and reliability of the detection.

[0003] Existing technologies suffer from low accuracy and reliability in detecting broken lines in laser-etched photovoltaic modules. Summary of the Invention

[0004] The purpose of this application is to provide an optical online detection method and system for laser-etched lines breaking in photovoltaic modules, in order to solve the technical problem of low accuracy and reliability in the detection of laser-etched lines breaking in photovoltaic modules in the prior art.

[0005] In view of the above problems, this application provides an optical online detection method and system for laser etching line breakage in photovoltaic modules.

[0006] The first aspect of this application provides an online optical detection method for laser-etched line breaks in photovoltaic modules. The method includes: sequentially activating a visible light image acquisition module and a spectral detection module at a detection station to perform scanning acquisition of the target photovoltaic module, establishing a morphological texture image and a spectral dataset; performing spatial feature extraction and contour enhancement processing on the morphological texture image to establish an etching line connectivity matrix; calculating an explicit break criterion using pixel connectivity, geometric continuity, and local gray-level gradient in the etching line connectivity matrix to establish a first detection result; performing multi-segment energy ratio analysis and absorption peak position shift detection on the spectral dataset, and generating a spectral-spatial coupling vector after spatial constraint through the etching line connectivity matrix; establishing a second detection result based on the spectral-spatial coupling vector; performing an electrical excitation test on the photovoltaic module and recording an excitation response set, the excitation response set including frequency domain phase, amplitude response, and photoelectric coupling response; establishing a third detection result based on the excitation response set; interactively verifying the first, second, and third detection results; and outputting an online detection result.

[0007] Furthermore, the spectral dataset is divided into N overlapping energy segments within the visible to near-infrared band. The energy ratio spectra of adjacent wavelength intervals are calculated based on the division results, constructing a multi-segment energy ratio matrix, where N is an integer greater than 2. Correlation analysis is performed on the multi-segment energy ratio matrix to calculate the energy distribution gradient and spectral stability index, identifying energy discontinuities based on the calculation results. The shift trend of the corresponding absorption peaks is extracted based on these energy discontinuities, establishing a peak position shift vector. A weighted energy shift spectrum is established based on the peak position shift vector, using the local energy ratio change rate in the N overlapping energy segments as weights.

[0008] Furthermore, the weighted energy offset spectrum is spatially matched with the etching line connectivity matrix to limit the spatial projection range of the anomalous energy response, generating a spectral-spatial coupling vector. This vector characterizes the energy perturbation mode and geometric connectivity of the etching path. The spectral-spatial coupling vector is normalized, and the spectral weight coefficients and spatial connectivity coefficients of the coupled components are calculated to establish a weighted coupling feature set. Under the constraint of the etching line connectivity matrix, local statistical parameters of the weighted coupling feature set are calculated using a sliding neighborhood as the computational unit. These local statistical parameters include energy anomaly score, connectivity consistency score, and spectral stability score. Multi-domain fusion judgment is performed on the local statistical parameters, and a second detection result is output.

[0009] Furthermore, the etching line region is located using the etching line connectivity matrix, and a binarized pixel image of the etching line region is obtained; a pixel connectivity index is calculated based on the binarized pixel image, and a topological connectivity graph of the etching line trunk and branches is constructed; a local geometric continuity index is calculated using a line segment fitting algorithm in the topological connectivity graph, the local geometric continuity index being calculated through the direction angle and curvature change rate of adjacent line segments; the gray-level change ratio along the etching line direction and the normal direction is calculated, and a local gray-level gradient vector is calculated; an explicit line break criterion is calculated based on the pixel connectivity index, the local geometric continuity index, and the local gray-level gradient vector.

[0010] Furthermore, a controllable excitation signal is applied to the photovoltaic module, the excitation signal including pulsed light excitation, micro-vibration excitation, and voltage excitation; the response acquisition of the photovoltaic module is performed, and the response acquisition results are established, the response acquisition results including response time nodes, response spectral feature sets, response voltage feature sets, and thermal response feature sets; the response acquisition results are output as an excitation response set.

[0011] Furthermore, time-series differential analysis is performed on the excitation response set to extract the spectral energy difference distribution and voltage response change rate before and after the excitation signal, establishing a dynamic response curve set; multi-domain feature decoupling is performed on the dynamic response curve set to calculate the frequency domain phase drift parameter, amplitude attenuation coefficient, and photoelectric response delay index, and the local temperature rise gradient is extracted using the thermal response feature set; an excitation response feature vector is established based on the frequency domain phase drift parameter, amplitude attenuation coefficient, photoelectric response delay index, and local temperature rise gradient; the excitation response feature vector is used to identify response anomalies and establish a third detection result.

[0012] Furthermore, the first, second, and third detection results are mapped to a unified spatial coordinate system; the detection consistency index of the same etched segment is calculated based on the mapping results, and multi-domain interactive confidence analysis is performed based on the calculation results to complete the interactive verification.

[0013] Furthermore, based on the online detection results, the photovoltaic modules are identified and a splitting identifier is configured; after the online detection results are bound to the unique code of the photovoltaic modules and uploaded, the splitting transmission management of the photovoltaic modules is performed according to the splitting identifier.

[0014] Furthermore, it is determined whether the trust value of the interactive verification is lower than a preset threshold; if the trust value is lower than the preset threshold, an anomaly detection flag is generated, and the corresponding photovoltaic module is diverted to the focus detection area according to the anomaly detection flag for focus re-detection processing.

[0015] A second aspect of this application provides an online optical detection system for laser-etched line breaks in photovoltaic modules. The system includes: a scanning acquisition module for sequentially activating a visible light image acquisition module and a spectral detection module at the detection station to perform scanning acquisition of the target photovoltaic module, establishing a morphological texture image and a spectral dataset; a first detection result establishment module for performing spatial feature extraction and contour enhancement processing on the morphological texture image, establishing an etching line connectivity matrix, calculating explicit line breakage criteria using pixel connectivity, geometric continuity, and local gray-level gradients in the etching line connectivity matrix, and establishing a first detection result; and a second detection result establishment module. The module is used to perform multi-segment energy ratio analysis and absorption peak position shift detection of the spectral dataset, and generate a spectral-spatial coupling vector after spatial constraint through the etching line connectivity matrix, and establish a second detection result based on the spectral-spatial coupling vector; the electrical excitation test module is used to perform electrical excitation test of photovoltaic modules, record the excitation response set, which includes frequency domain phase, amplitude response, and photoelectric coupling response; the detection result output module is used to establish a third detection result based on the excitation response set, perform interactive verification of the first detection result, the second detection result, and the third detection result, and output the online detection result.

[0016] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0017] The method provided in this application sequentially activates the visible light image acquisition module and the spectral detection module at the detection station to perform scanning acquisition of the target photovoltaic module, establishing a morphological texture image and a spectral dataset; spatial feature extraction and contour enhancement processing are performed on the morphological texture image to establish an etching line connectivity matrix; explicit line breakage criteria are calculated using pixel connectivity, geometric continuity, and local gray-level gradient in the etching line connectivity matrix to establish a first detection result; multi-segment energy ratio analysis and absorption peak position shift detection are performed on the spectral dataset, and after spatial constraint through the etching line connectivity matrix, a spectral-spatial coupling vector is generated; a second detection result is established based on the spectral-spatial coupling vector; electrical excitation testing of the photovoltaic module is performed, and an excitation response set is recorded, including frequency domain phase, amplitude response, and photoelectric coupling response; a third detection result is established based on the excitation response set; the first, second, and third detection results are interactively verified, and an online detection result is output. By comprehensively utilizing multiple technologies such as image processing, spectral analysis, and dynamic response analysis, the method effectively avoids false alarms and false negatives, and improves the accuracy and reliability of online monitoring of laser etching line breaks in photovoltaic modules.

[0018] The above description is merely an overview of the technical solution of this application. To enable a clearer understanding of the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this application 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 merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0020] Figure 1 A schematic flowchart of the optical online detection method for laser etching line breakage in photovoltaic modules provided in this application.

[0021] Figure 2 A schematic diagram of the optical online detection system for broken laser-etched lines in photovoltaic modules provided in this application.

[0022] Explanation of reference numerals in the attached figures: 11 for scanning acquisition module, 12 for establishing the first detection result, 13 for establishing the second detection result, 14 for electrical excitation testing module, and 15 for outputting the detection result. Detailed Implementation

[0023] This application provides an optical online detection method and system for laser-etched line breaks in photovoltaic modules, addressing the technical problems of low accuracy and reliability in existing technologies for detecting laser-etched line breaks in photovoltaic modules. By comprehensively utilizing multiple technologies such as image processing, spectral analysis, and dynamic response analysis, it effectively avoids false alarms and missed alarms, thereby improving the accuracy and reliability of online monitoring of laser-etched line breaks in photovoltaic modules.

[0024] The technical solutions of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be understood that the present invention is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. It should also be noted that, for ease of description, only the parts related to the present invention are shown in the accompanying drawings, not all of them.

[0025] Example 1, as Figure 1 As shown, this application provides an optical online detection method for broken laser-etched lines in photovoltaic modules. The optical online detection method for broken laser-etched lines in photovoltaic modules includes:

[0026] The visible light image acquisition module and the spectral detection module are activated sequentially at the detection station to perform scanning and acquisition of the target photovoltaic module, and to establish morphological texture images and spectral datasets.

[0027] Specifically, laser etching line breakage in photovoltaic modules refers to the phenomenon where conductive lines, such as grid lines, formed on the surface of photovoltaic cells during the laser etching process occur with breaks or interruptions. A photovoltaic module inspection station is set up on the photovoltaic module production line. This station is equipped with various inspection devices and supporting facilities, including visible light image acquisition modules and spectral detection modules, for quality inspection of photovoltaic modules after the laser etching process. Furthermore, the inspection station has relatively independent and stable environmental conditions, including suitable temperature, humidity, and light control, which can reduce the interference of external environmental factors on the inspection results.

[0028] The target photovoltaic module refers to the photovoltaic panel currently at the inspection station undergoing laser etching line breakage detection. When the target photovoltaic module is at the inspection station, the visible light image acquisition module and the spectral detection module are activated sequentially. The visible light image acquisition module consists of a high-resolution industrial camera and a precision lens. The high-resolution industrial camera can capture the details of the laser etching lines on the surface of the photovoltaic module, while the precision lens can distinguish the clarity and accuracy of the image, avoiding image information distortion caused by lens distortion. The visible light image acquisition module scans and acquires the surface of the target photovoltaic module through the high-resolution industrial camera and the precision lens, generating a morphological texture image. This morphological texture image includes feature images such as the shape, direction, and surface texture of the etching lines on the surface of the photovoltaic module.

[0029] After acquiring visible light images, the spectral detection module is activated. This module includes a spectrometer and a fiber optic probe. The fiber optic probe accurately guides the light signal to the spectrometer, ensuring the accuracy and reliability of the acquired spectral data. The spectrometer is used to disperse the light reflected or emitted by the photovoltaic module, breaking it down into light signals of different wavelengths and measuring the light intensity corresponding to each wavelength to obtain spectral data. The spectral detection module scans and acquires data from the target photovoltaic module, integrating the acquired spectral data to form a spectral dataset. This spectral data refers to the reflectance spectrum data of the photovoltaic module surface, reflecting the absorption characteristics of the photovoltaic material at different wavelengths.

[0030] The morphological texture images and spectral datasets are complementary; the morphological texture images reflect the physical morphology of the laser etching lines, while the spectral datasets reflect the optical properties of the component surface materials. By combining a visible light image acquisition module and a spectral detection module, the accuracy of laser etching line breakage detection is increased. For example, single image detection may lead to misjudgments due to surface contamination or lighting issues; combining spectral data provides more comprehensive information, thereby improving the accuracy and reliability of laser etching line breakage detection in photovoltaic modules.

[0031] Spatial feature extraction and contour enhancement are performed on the morphological texture image to establish an etched line connectivity matrix. The explicit line breakage criterion is calculated using the pixel connectivity, geometric continuity, and local gray-level gradient in the etched line connectivity matrix to establish a first detection result.

[0032] Furthermore, the explicit line breakage criterion is calculated using the pixel connectivity, geometric continuity, and local gray-level gradient in the etching line connectivity matrix to establish a first detection result, including: locating the etching line region using the etching line connectivity matrix and obtaining a binary pixel map of the etching line region; calculating the pixel connectivity index based on the binary pixel map and constructing a topological connectivity graph of the etching line trunk and branches; calculating the local geometric continuity index using a line segment fitting algorithm in the topological connectivity graph, the local geometric continuity index being calculated through the direction angle and curvature change rate of adjacent line segments; calculating the gray-level change ratio along the etching line direction to the normal direction, and calculating the local gray-level gradient vector; and calculating the explicit line breakage criterion based on the pixel connectivity index, the local geometric continuity index, and the local gray-level gradient vector.

[0033] Specifically, the morphological texture image is converted to grayscale, and edge detection algorithms, such as Canny edge detection, are used to extract spatial features. Alternatively, feature extraction can be performed based on a region growing algorithm. Region growing starts from a seed point and merges adjacent pixels with similar properties into a region until no more pixels can be merged. By selecting a pixel on the etching line as the seed point and defining similarity criteria, such as grayscale value difference less than a grayscale value threshold and texture similarity greater than a texture threshold, adjacent pixels are merged into the region according to the similarity criteria starting from the seed point until no more pixels can be merged, thus obtaining the spatial features of the etching line region.

[0034] The morphological texture image is then enhanced using contour enhancement algorithms to make the etching lines clearer and more prominent. For example, histogram equalization enhancement is employed. By calculating the gray-level histogram of the morphological texture image, the frequency of each gray level is statistically analyzed, and the cumulative distribution function of the gray levels is calculated based on the histogram. This cumulative distribution function is then mapped to obtain new gray values, which are then used to replace the original gray values ​​in the morphological texture image, resulting in the enhanced morphological texture image. Combining the extracted spatial features of the etching line region, connected regions in the enhanced morphological texture image are labeled, with each connected region assigned a displacement identifier. Specifically, scanning begins from the top left corner of the morphological texture image, pixel by pixel. When an unlabeled pixel is encountered, it is marked with a new identifier, and a depth-first search or breadth-first search algorithm is used to find all pixels identical to that unlabeled pixel and mark them with the same identifier. This process continues until all pixels are labeled. Based on the pixel marking results, a matrix of size equal to the total number of pixels in the image is created. Initially, all elements are set to 0. For each pixel, if it belongs to a connected region of the same marking, it is considered connected. The corresponding element in the matrix is ​​then set to 1. The resulting matrix is ​​the etching line connectivity matrix. The rows and columns of the etching line connectivity matrix correspond to the pixels in the morphological texture map, and the values ​​of the matrix elements indicate whether the corresponding pixels are connected.

[0035] By traversing the etching line connectivity matrix, pixels belonging to the same connected region are marked, thus locating the etching line region. After locating the etching line region, a binary pixel map of the etching line region is obtained. This binary pixel map simplifies the pixel values ​​in the image to only two states: 0 and 1. The etching line region is white with a pixel value of 1, and the background region is black with a pixel value of 0. The pixel connectivity index refers to the degree of connection between pixels in the etching line. For each etching line pixel in the binary pixel map, the number of pixels belonging to the etching line region within its four-neighborhood is counted, with a maximum of eight neighboring pixels analyzed. For example, if three pixels within a pixel's four-neighborhood belong to the etching line region, it indicates that the pixel has high local connectivity. By combining the connectivity of all etching line pixels, the pixel connectivity index of the entire etching line region is obtained.

[0036] A topological connectivity graph of the etching line trunk and branches is constructed based on pixel connectivity metrics. This topological connectivity graph is a graphical structure using nodes and edges to represent the connections between etching lines. Each node represents an etching line pixel, and edges represent the connections between nodes. When constructing the topological connectivity graph of the etching line trunk and branches, the endpoints and intersections of the etching lines are first identified as nodes. The connections between nodes are then determined based on pixel connectivity, forming edges. For example, if the pixel connectivity between two endpoints is high, it indicates a continuous etching line connection between the two nodes, and an edge is used to connect the two nodes in the topological connectivity graph. The trunk in the topological connectivity graph represents the main direction of the etching line, while branches may be defects caused by broken lines or incomplete etching. Geometric analysis is performed on the etched lines in the topology connection diagram. Using line segment fitting algorithms, such as the least squares method, adjacent etched line segments are fitted to calculate the direction angle and rate of curvature change between them, obtaining local geometric continuity indices. The direction angle reflects the degree of turning between adjacent line segments; if the direction angle is too large, it indicates a significant direction shift between adjacent etched line segments, potentially indicating a risk of line breakage. The rate of curvature change reflects how quickly the curvature of the etched line changes; a large rate of curvature change indicates a possible abrupt change or bending in the etched line. For example, if the direction angle between two adjacent line segments exceeds the implementation requirements, and the rate of curvature change shows an abrupt change at a certain point, it indicates poor geometric continuity of the etched line at that location.

[0037] The ratio of grayscale change along the etching line direction to that along the normal direction is calculated. The grayscale change along the etching line direction reflects the grayscale change of the etching line itself, while the grayscale change along the normal direction reflects the grayscale difference between the etching line and the surrounding background. The ratio of grayscale change reflects the variation of grayscale values ​​in different directions and is used to determine whether the grayscale changes of the etching line at different locations are abnormal. Based on the grayscale change ratio, a local grayscale gradient vector is calculated. The direction of the local grayscale gradient vector represents the direction of grayscale change, and its magnitude represents the amplitude of the grayscale change. When there is a break in the etching line, the local grayscale gradient vector will show a significant change.

[0038] By combining pixel connectivity metrics, local geometric continuity metrics, and local grayscale gradient vectors, a comprehensive explicit break-line criterion is calculated, and the linear break-line criterion is used as the first detection result. For example, the pixel connectivity metrics, local geometric continuity metrics, and local grayscale gradient vectors are normalized. Based on practical experience, a weight is assigned to each of the three metrics. The three metrics are then weighted and summed to obtain the explicit break-line criterion. When the explicit break-line criterion exceeds a set threshold, it is determined that the etched line portion has a linear break-line defect.

[0039] By integrating information from multiple aspects such as space, geometry, and grayscale, it is possible to comprehensively and accurately determine whether there are visible broken lines, thereby improving the accuracy and reliability of photovoltaic module testing.

[0040] Perform multi-segment energy ratio analysis and absorption peak position shift detection on the spectral dataset, and generate a spectral-spatial coupling vector after spatial constraint through the etching line connectivity matrix. Establish a second detection result based on the spectral-spatial coupling vector.

[0041] Further, performing multi-segment energy ratio analysis and absorption peak position shift detection on the spectral dataset includes: dividing the spectral dataset into N overlapping energy segments within the visible to near-infrared band; calculating the energy ratio spectrum of adjacent wavelength intervals based on the division results; constructing a multi-segment energy ratio matrix, where N is an integer greater than 2; performing correlation analysis between spectral segments on the multi-segment energy ratio matrix; calculating the energy distribution gradient and spectral segment stability index; identifying energy discontinuity characteristics based on the calculation results; extracting the shift trend of the corresponding absorption peak based on the energy discontinuity characteristics; and establishing a peak position shift vector; using the local energy ratio change rate in the N overlapping energy segments as weights, and establishing a weighted energy shift spectrum based on the peak position shift vector.

[0042] Specifically, the spectral dataset is processed by dividing the spectral data from the visible to near-infrared bands into N overlapping energy segments based on spectral characteristics and the required wavelength range, where N is an integer greater than 2. This overlapping energy segment division allows for a comprehensive analysis of energy variation characteristics across different wavelength ranges, avoiding missed detections due to coarse segmentation. For example, if the spectral range is 400-1000 nm, each 100 nm segment is considered a basic interval, with a certain overlap width, such as 20 nm, ensuring partial wavelength overlap between adjacent energy segments. The energy ratio spectrum of adjacent wavelength intervals is calculated based on the segmentation results. This energy ratio spectrum is obtained by calculating the spectral energy ratio of adjacent wavelength intervals. For example, if two adjacent wavelength intervals A and B have spectral energies EA and EB respectively, then the energy ratio of the adjacent intervals is R = EA / EB. By calculating the energy ratios of all adjacent wavelength intervals, a multi-segment energy ratio matrix is ​​constructed, where the matrix elements are the energy ratios of corresponding adjacent intervals.

[0043] Correlation analysis algorithms are used to analyze the correlation between spectral segments of a multi-segment energy ratio matrix. For example, the Pearson correlation coefficient method is used to calculate the correlation coefficient between the energy ratio sequences of two spectral segments. If the correlation coefficient between the two spectral segments is close to 1, it indicates that the energy change trends of the two spectral segments are highly consistent; close to -1, it indicates that the energy change trends of the two spectral segments are opposite; and close to 0, it indicates no significant correlation. In normal photovoltaic modules, the spectral responses of different wavelength bands are usually relatively stable and consistent. If there are broken etching lines or uneven etching, the responses of different wavelength bands will fluctuate irregularly, leading to a decrease in the correlation between spectral segments. Based on the correlation analysis results, the energy distribution gradient is obtained by calculating the rate of change of the energy ratio of adjacent spectral segments in the wavelength direction. That is, the difference value is calculated for the energy ratio sequences of adjacent spectral segments in the multi-segment energy ratio matrix. The magnitude of the difference value reflects the rate of energy change between adjacent spectral segments. The larger the difference value, the larger the energy distribution gradient, indicating a more drastic energy change. Simultaneously, the correlation degree between each spectral band and other spectral bands is determined based on the correlation analysis results, and the variance or standard deviation of the energy ratio within each spectral band is calculated to obtain the spectral band stability index. This index reflects the consistency of each spectrum within a continuous wavelength range; the smaller the variance or standard deviation, the smaller the fluctuation in the energy ratio of the spectral band, and the higher the spectral band stability index. Based on the energy distribution gradient and the spectral band stability index, energy discontinuity characteristics are identified. Energy discontinuity characteristics refer to spectral bands with abnormally large energy distribution gradients or whose spectral band stability deviates significantly from the normal range.

[0044] An absorption peak is a specific wavelength position in a spectrum where energy absorption is strong. When material properties or structure change, the position of the absorption peak may shift. By utilizing energy discontinuity characteristics, the corresponding wavelength range in the spectral data is obtained. By comparing this energy discontinuity with the spectral energy characteristics of the absorption peak position under normal conditions, it can be determined whether the absorption peak has shifted towards shorter or longer wavelengths. The difference in wavelength between the normal and shifted positions of the absorption peak is calculated. Using the normal position as the starting point and the shifted position as the ending point, the shift trend of the corresponding absorption peak is extracted. The shift direction is used as the vector direction, and the shift amount as the vector magnitude, to establish a peak position shift vector.

[0045] The energy ratios in a multi-segment energy ratio matrix are differentially calculated, and the difference results are used as the local energy ratio change rate, which reflects the speed of change of the energy ratio within an energy segment. The local energy ratio change rates in N overlapping energy segments are used as weights, and a weighted energy shift spectrum is established based on the peak position shift vector. The peak position shift vector is multiplied by the corresponding local energy ratio change rate weight to obtain the weighted energy shift spectrum for each energy segment. This weighted energy shift spectrum not only includes the shift information of the absorption peak but also incorporates the change rate of different energy segments, achieving a more accurate representation of spectral energy perturbations and further improving the accuracy and reliability of laser etching line breakage defects in photovoltaic modules.

[0046] By combining the etching line connectivity matrix with the weighted energy shift spectrum of the spectral data for spatial constraints, a spectral-spatial coupling vector is generated. This vector simultaneously reflects the anomalous characteristics of the photovoltaic module in both the spectral and spatial dimensions, providing more comprehensive and accurate detection results. A second detection result is then generated based on this vector. This second result integrates both spectral and spatial analysis results, enabling a more comprehensive, accurate, and reliable detection of laser etching line breaks in photovoltaic modules compared to single spectral or spatial analysis. This provides reliable data support for the quality assessment and fault diagnosis of photovoltaic modules.

[0047] Furthermore, after spatial constraints are applied through the etching line connectivity matrix, a spectral-spatial coupling vector is generated. This includes: spatially matching the weighted energy offset spectrum with the etching line connectivity matrix to limit the spatial projection range of the anomalous energy response, generating a spectral-spatial coupling vector that characterizes the energy perturbation mode and geometric connectivity of the etching path; normalizing the spectral-spatial coupling vector, calculating the spectral weight coefficients and spatial connectivity coefficients of the coupled components, and establishing a weighted coupling feature set; under the constraints of the etching line connectivity matrix, calculating the local statistical parameters of the weighted coupling feature set using the sliding neighborhood as the calculation unit, the local statistical parameters including energy anomaly score, connectivity consistency score, and spectral stability score; and performing multi-domain fusion judgment on the local statistical parameters to output a second detection result.

[0048] Specifically, spatial matching is performed between the weighted energy shift spectrum and the etching line connectivity matrix. Abnormal energy responses in the weighted energy shift spectrum are precisely projected onto the spatial range of the etching line connectivity matrix, determining the spatial relationship between the location of energy disturbances and the etching path, thus generating a spectral-spatial coupling vector. For example, if a laser etching line breaks at a certain point in a photovoltaic module, the corresponding spectral energy may exhibit abnormal changes. Spatial matching associates the energy anomaly with the specific etching line location. The spectral-spatial coupling vector simultaneously characterizes the energy disturbance mode and geometric connectivity of the etching path. The energy disturbance mode refers to the abnormal spectral energy distribution caused by the breakage defect or uneven etching, while geometric connectivity reflects the spatial continuity and integrity of the etching line.

[0049] The generated spectral-spatial coupling vector is normalized using a minimum-maximum normalization method to eliminate biases caused by differences in different spectral bands or spatial variations. The normalized spectral-spatial coupling vector contains characteristics related to energy perturbations and spatial consistency. For each spectral component in the normalized spectral-spatial coupling vector, the spectral weight coefficient of the coupling component is calculated. This spectral weight coefficient reflects the relative importance of different spectral bands in the coupling vector. It can be calculated by statistically analyzing the frequency of different values ​​of each spectral band in the coupling vector and using the information entropy calculation formula: Information Entropy = -∑(frequency × log(frequency)). After obtaining the information entropy of multiple spectral bands, the information entropy is normalized, and the resulting value is the spectral weight coefficient of each spectral band. Simultaneously, the spatial connectivity coefficient is calculated. This coefficient reflects the degree of spatial connectivity between different locations on the etching line and is related to the pixel connectivity in the etching line connectivity matrix. For each location in the normalized spectral-spatial coupling vector, the number of connected pixels within a certain range (e.g., a 4×4 neighborhood) is statistically analyzed. Dividing the number of connected pixels by the total number of pixels in the neighborhood yields the connectivity ratio, which is the spatial connectivity coefficient for that location. Based on the calculated spectral weighting coefficient and spatial connectivity coefficient, a weighted coupling feature set is established. This set comprehensively considers spectral and spatial information, further enhancing the reliability and accuracy of the detection results.

[0050] Under the constraint of the etching line connectivity matrix, local statistical parameters of the weighted coupling feature set are calculated using a sliding neighborhood as the computational unit. The sliding neighborhood refers to a local region selected within the etching line connectivity matrix, and sliding analysis is performed according to a preset step size. For each sliding local region, energy anomaly score, connectivity consistency score, and spectral stability score are calculated to obtain local statistical parameters. Energy anomaly analysis reflects the degree of energy change anomaly within the sliding neighborhood. Statistical methods are used to calculate the mean or variance of the deviation as the degree of anomaly; a higher score indicates a more pronounced anomaly. The connectivity consistency score reflects the stability of etching line connectivity within the sliding neighborhood. It is obtained by dividing the actual number of connected pixels by the theoretical total number of connected pixels; a higher score indicates better connectivity. The spectral stability score reflects the stability of spectral band energy changes within the sliding neighborhood. It is obtained by calculating the standard deviation of multiple spectral band energies; a higher score indicates a more stable spectral band. Finally, the energy anomaly score, connectivity consistency score, and spectral stability score from the local statistical parameters are fused across multiple domains to generate a second detection result. For example, by setting judgment rules and thresholds, local statistical parameters are comprehensively evaluated. When the energy anomaly score exceeds the energy anomaly threshold, and the connectivity consistency score and spectral stability are lower than the corresponding thresholds, it is determined that there is an anomaly in the region, and a second detection result is output.

[0051] By combining spectral data with spatial geometric features, not only can energy anomalies be identified, but their spatial location can also be pinpointed. This improves the accuracy and reliability of identifying minute broken wire defects or complete local etching. Furthermore, by combining local statistical analysis and multi-domain fusion judgment, more accurate and reliable detection results can be output.

[0052] Perform electrical excitation tests on photovoltaic modules and record the excitation response set, which includes frequency domain phase, amplitude response, and photoelectric coupling response.

[0053] Furthermore, an electrical excitation test is performed on the photovoltaic module, and an excitation response set is recorded, including: applying a controllable excitation signal to the photovoltaic module, the excitation signal including pulsed light excitation, micro-vibration excitation, and voltage excitation; performing response acquisition of the photovoltaic module, establishing response acquisition results, the response acquisition results including response time nodes, response spectral feature sets, response voltage feature sets, and thermal response feature sets; and outputting the response acquisition results as an excitation response set.

[0054] Specifically, a controllable excitation signal is applied to the photovoltaic (PV) module. A controllable excitation signal means that the excitation parameters can be precisely adjusted. The applied excitation signals include pulsed light excitation, micro-vibration excitation, and voltage excitation. Pulsed light excitation simulates rapid changes in actual illumination by emitting short pulses of light, thus exciting the PV module's response characteristics under rapid light variations. Micro-vibration excitation applies minute mechanical vibrations to the PV module, simulating mechanical stress and vibration conditions in the actual working environment, and detecting potential wire breakage problems caused by mechanical stress. Voltage excitation tests the electrical characteristics of the PV module by applying a voltage change of a certain amplitude. After applying the excitation signal, the PV module's response is comprehensively collected, including multiple response acquisition results such as response time points, response spectral feature sets, response voltage feature sets, and thermal response feature sets. Response time points refer to the time points at which the PV module's response changes after the excitation signal is applied, used to analyze the PV module's response speed and stability to different excitation signals. The response spectral feature set refers to the spectral data extracted through frequency domain analysis, reflecting the frequency response characteristics of the PV module to different excitation signals, and spectral changes can be monitored in real time using a spectrophotometer. The response voltage characteristic set reflects the electrical response of a photovoltaic (PV) module under voltage excitation, including electrical parameters such as response current and power. This data can be acquired using a multimeter to reflect the electrical performance of the PV module. The thermal response characteristic set refers to the temperature change of the PV module under different excitation conditions, which can be acquired using temperature sensors such as thermal imagers or thermocouples. The acquired response results are integrated to form an excitation response set, which includes frequency domain phase, amplitude response, and photoelectric coupling response. Frequency domain phase refers to the phase difference between the response signal and the excitation signal after the PV module is excited, which can be obtained through frequency domain analysis of the response signal using Fourier transform. The amplitude response reflects the intensity of the PV module's response to different excitation signals. The photoelectric coupling response refers to the interaction between light and electrical signals in the PV module, which can be obtained by simultaneously applying light and electrical excitation signals and measuring the electrical response of the PV module. For example, a light source with a wavelength of 500nm is selected, the pulse duration is 200ms, the power is 100mW, and the excitation time of each pulse is measured for 1 second; during the micro-vibration excitation process, a small-amplitude vibration with a frequency of 50Hz and an amplitude of 0.1mm is applied for a duration of 5 seconds, and the vibration source is placed on the back or side of the photovoltaic module to excite the photovoltaic module to undergo slight deformation; for voltage excitation, an adjustable DC voltage signal is selected, which varies from 0V to 40V in 5V steps, and the duration of each step is 1 second, to test the electrical stability of the photovoltaic module under different voltages.

[0055] By applying different types of excitation signals to photovoltaic modules and collecting their response data, a comprehensive evaluation of the electrical performance, photoelectric effect, mechanical properties, and thermal response of photovoltaic modules can be achieved. This improves the comprehensiveness, accuracy, and reliability of laser etching line breakage detection, ensuring the quality and performance of photovoltaic modules.

[0056] A third detection result is established based on the stimulus response set. The first, second, and third detection results are then interactively verified, and an online detection result is output.

[0057] Furthermore, establishing a third detection result based on the excitation response set includes: performing time-series differential analysis on the excitation response set to extract the spectral energy difference distribution and voltage response change rate before and after the excitation signal, and establishing a dynamic response curve set; performing multi-domain feature decoupling on the dynamic response curve set to calculate the frequency domain phase drift parameter, amplitude attenuation coefficient, and photoelectric response delay index, and extracting the local temperature rise gradient using the thermal response feature set; establishing an excitation response feature vector based on the frequency domain phase drift parameter, amplitude attenuation coefficient, photoelectric response delay index, and local temperature rise gradient; and using the excitation response feature vector to identify response anomalies and establish a third detection result.

[0058] Specifically, time-series difference analysis is performed on the spectral energy and voltage response data in the excitation response set to extract the spectral energy difference distribution and voltage response change rate before and after the excitation signal. For example, for the response spectral feature set, the difference in light intensity at each wavelength point before and after excitation is calculated to obtain the spectral energy difference distribution, which intuitively reflects the impact of excitation on the absorption and conversion of light at different wavelengths. For the response voltage feature set, the voltage difference between adjacent time points is calculated to obtain the voltage response change rate, reflecting the dynamic change of voltage under excitation. Based on the spectral energy difference distribution and voltage response change rate, a dynamic response curve set is constructed with time as the horizontal axis and the spectral energy difference or voltage response change rate as the vertical axis. Multi-domain feature decoupling is performed on the dynamic response curve set, and frequency domain analysis is performed on the time domain response data through Fourier transform to obtain the frequency domain phase drift parameter, which reflects the time delay of the signal at different frequencies. Voltage response data of photovoltaic modules at different time points are extracted from the dynamic response curve set, and the maximum amplitude V is obtained by extracting the amplitude of the voltage response data within a certain time period. max The amplitude attenuation coefficient is obtained by exponentially fitting the response signal. The amplitude attenuation formula is V(t) = V max ⋅e −αtWhere α is the attenuation coefficient and t is time, the amplitude attenuation coefficient is obtained by fitting using the least squares method. Simultaneously, by analyzing the time interval from the application of the excitation signal to the photovoltaic module generating a photoelectric response, a photoelectric time delay index is obtained. For each time point in the thermal response feature set, the local temperature rise gradient is calculated using the finite difference method; this local temperature rise gradient reflects the rate of temperature change in different regions of the photovoltaic module.

[0059] Frequency domain phase drift parameters, amplitude attenuation coefficients, photoelectric response delay indicators, and local temperature rise gradients are integrated to construct an excitation response feature vector. This vector synthesizes multiple response characteristics of the photovoltaic module under the action of an excitation signal, and the preset weights can be set according to actual needs. Finally, the excitation response feature vector is compared with the normal response feature vector to identify response anomalies. When one or more features exceed the normal range in the normal response feature vector, a defect is indicated, and a third detection result is obtained. The third detection result not only reflects the dynamic response of the photovoltaic module under multiple excitations but also identifies potential defects in the photovoltaic module under actual operating conditions.

[0060] Finally, the first, second, and third detection results were cross-validated. By comparing and comprehensively analyzing the detection results from different sources, the accuracy and reliability of the online detection results of laser etching line breaks in photovoltaic modules were further improved, reducing misjudgments or omissions.

[0061] Furthermore, the first detection result, the second detection result, and the third detection result are interactively verified, including: mapping the first detection result, the second detection result, and the third detection result to a unified spatial coordinate system; calculating the detection consistency index of the same etched segment based on the mapping result; performing multi-domain interactive confidence analysis based on the calculation result to complete the interactive verification.

[0062] Specifically, a specific intersection point of the photovoltaic module is used as the origin of the coordinate system. A two-dimensional planar index coordinate system is constructed based on the actual length and width of the module. For the defect location information in the first detection result, which is expressed in pixels, the first detection result is converted into a unified spatial coordinate system using geometric conversion methods such as the principle of similar triangles, based on the actual distance between the optical imaging device and the photovoltaic module, the imaging scale, and the setting of the origin. For the second detection result, fixed feature points in the image, such as the edges of etching lines, are extracted and registered with their spatial positions in the spectral data, and then mapped to the unified spatial coordinate system. For the third detection result, fixed feature points in the image are fixed and matched with their positions in the electrical excitation test to achieve the mapping of the third detection result. Based on the mapping results, the detection results at the same locations are analyzed to evaluate the matching degree of the three detection results. A consistency index for the same etched segment is calculated. This consistency index can be obtained by comparing the consistency of the judgments made by each detection method regarding the same etched path. For example, it can be calculated using overlap, matching degree, or correlation coefficient. The closer the correlation coefficient is to 1, the higher the consistency. This means that if all three detection results simultaneously identify either a broken etched line defect or the absence of a quality defect in the current photovoltaic module, the consistency of the detection results is high. Through multi-domain interactive confidence analysis, the credibility of each detection result is integrated to evaluate the reliability of the final detection result. Different detection methods have different credibility. Based on the expected performance and historical performance data of each detection method, an independent confidence level is calculated for each method, such as the clarity of the optical image, the signal strength of the spectral data, and the response amplitude of the electrical excitation test. The consistency score is weighted based on the independent confidence level of each detection method to obtain the final interactive verification confidence value. Interactive verification is then completed, and the online detection result is output. If the interactive verification confidence value is high, it indicates that the three detection results are consistent, and the detection result is reliable. If the trust value of the interactive verification is low, it indicates that there is a problem with the etched segment, which leads to errors or inconsistencies in the results of different detection methods. Further analysis of the photovoltaic module is required.

[0063] By using spatial mapping and consistency analysis, the advantages and disadvantages of multiple different detection methods can be comprehensively considered, and a more reliable detection conclusion can be obtained through confidence analysis. This can effectively cope with changes under different environmental conditions, reduce false alarms and false negatives, and improve the accuracy and reliability of online detection of photovoltaic modules.

[0064] Furthermore, after outputting the online detection results, the process includes: identifying the photovoltaic modules based on the online detection results and configuring a splitting identifier; binding the online detection results with the unique code of the photovoltaic modules and uploading them, and then managing the splitting transmission of the photovoltaic modules based on the splitting identifier.

[0065] Specifically, photovoltaic (PV) modules are categorized and identified based on online testing results. If the online testing indicates a broken laser etching line, the module is marked as faulty; otherwise, it is identified as a normal module. Corresponding categorization identifiers are configured based on the categorization results. These identifiers can be digital codes, QR codes, or RFID tags, ensuring that each PV module's specific status can be clearly identified during subsequent processing. The online testing results are bound to a unique code for each PV module. This unique code refers to the serial number or barcode assigned during the PV module's production process, ensuring the traceability of the testing data. The PV modules are then managed through categorization based on these identifiers. For example, faulty PV modules are automatically guided to the inspection or repair area, while normal modules proceed to the next process flow, achieving real-time and efficient quality management of the PV modules.

[0066] Furthermore, the interactive verification of the first detection result, the second detection result, and the third detection result is performed, and the online detection result is output. This also includes: determining whether the trust value of the interactive verification is lower than a preset threshold; if the trust value is lower than the preset threshold, a detection anomaly identifier is generated, and the corresponding photovoltaic module is diverted to the focus detection area according to the detection anomaly identifier for focus re-detection processing.

[0067] Specifically, based on historical testing data, production requirements, or expert experience, a preset threshold for the trust value is set. This preset threshold serves as a critical value to determine the reliability of the test results. The trust value of the interactive verification is compared with the preset threshold. When the trust value of the interactive verification is less than the preset threshold, it indicates that there is a significant inconsistency or error in the current test results. At this point, a test anomaly identifier is generated, and the corresponding photovoltaic modules are diverted to the focus testing area based on the test anomaly identifier. The photovoltaic modules in the focus testing area are then re-tested, improving the accuracy of the test anomaly detection, enhancing the reliability and accuracy of the output online test results, and ensuring the quality of photovoltaic module production.

[0068] Example 2, based on the same inventive concept as the optical online detection method for broken laser etching lines in photovoltaic modules in the foregoing examples, such as... Figure 2 As shown, this application provides an optical online detection system for broken laser-etched lines in photovoltaic modules, wherein the optical online detection system for broken laser-etched lines in photovoltaic modules includes:

[0069] The scanning acquisition module 11 is used to sequentially activate the visible light image acquisition module and the spectral detection module at the detection station to perform scanning acquisition of the target photovoltaic module, and establish a morphological texture image and a spectral dataset. The first detection result establishment module 12 is used to perform spatial feature extraction and contour enhancement processing on the morphological texture image, establish an etching line connectivity matrix, calculate the explicit line breakage criterion using the pixel connectivity, geometric continuity and local gray-level gradient in the etching line connectivity matrix, and establish a first detection result. The second detection result establishment module 13 is used to perform multi-segment energy ratio analysis and absorption peak position shift detection on the spectral dataset, and generate a spectral-spatial coupling vector after spatial constraint through the etching line connectivity matrix, and establish a second detection result based on the spectral-spatial coupling vector. The electrical excitation test module 14 is used to perform electrical excitation test on the photovoltaic module, record the excitation response set, which includes frequency domain phase, amplitude response and photoelectric coupling response. The detection result output module 15 is used to establish a third detection result based on the excitation response set, perform interactive verification of the first detection result, the second detection result and the third detection result, and output the online detection result.

[0070] Furthermore, the second detection result establishment module 13 is also used to: divide the spectral dataset into N overlapping energy segments in the visible to near-infrared band range; calculate the energy ratio spectrum of adjacent wavelength intervals based on the division results; construct a multi-segment energy ratio matrix, where N is an integer greater than 2; perform correlation analysis between spectral segments on the multi-segment energy ratio matrix; calculate the energy distribution gradient and spectral segment stability index; identify energy discontinuity features based on the calculation results; extract the shift trend of the corresponding absorption peak based on the energy discontinuity features; and establish a peak position shift vector; using the local energy ratio change rate in the N overlapping energy segments as weights, establish a weighted energy shift spectrum based on the peak position shift vector.

[0071] Furthermore, the second detection result establishment module 13 is also used to: spatially match the weighted energy offset spectrum with the etching line connectivity matrix to limit the spatial projection range of the abnormal energy response, and generate a spectral-spatial coupling vector, wherein the spectral-spatial coupling vector characterizes the energy perturbation mode and geometric connectivity of the etching path; normalize the spectral-spatial coupling vector, calculate the spectral weight coefficient and spatial connectivity coefficient of the coupling components, and establish a weighted coupling feature set; under the constraint of the etching line connectivity matrix, calculate the local statistical parameters of the weighted coupling feature set using the sliding neighborhood as the calculation unit, wherein the local statistical parameters include the energy anomaly score, connectivity consistency score, and spectral stability score; perform multi-domain fusion judgment on the local statistical parameters, and output the second detection result.

[0072] Furthermore, the first detection result establishment module 12 is also used to: locate the etching line region using the etching line connectivity matrix and obtain a binary pixel map of the etching line region; calculate the pixel connectivity index based on the binary pixel map and construct a topological connectivity graph of the etching line trunk and branches; calculate the local geometric continuity index using a line segment fitting algorithm in the topological connectivity graph, wherein the local geometric continuity index is calculated by the direction angle and curvature change rate of adjacent line segments; calculate the gray-level change ratio of the etching line along the line direction to the normal direction and calculate the local gray-level gradient vector; and calculate the explicit line breakage criterion based on the pixel connectivity index, the local geometric continuity index and the local gray-level gradient vector.

[0073] Furthermore, the electrical excitation test module 14 is also used to: apply a controllable excitation signal to the photovoltaic module, the excitation signal including pulsed light excitation, micro-vibration excitation, and voltage excitation; perform response acquisition of the photovoltaic module, establish response acquisition results, the response acquisition results including response time nodes, response spectral feature sets, response voltage feature sets, and thermal response feature sets; and output the response acquisition results as an excitation response set.

[0074] Furthermore, the detection result output module 15 is also used to: perform time series differential analysis on the excitation response set, extract the spectral energy difference distribution and voltage response change rate before and after the excitation signal, and establish a dynamic response curve set; perform multi-domain feature decoupling of the dynamic response curve set, calculate the frequency domain phase drift parameter, amplitude attenuation coefficient, and photoelectric response delay index, and extract the local temperature rise gradient using the thermal response feature set; establish an excitation response feature vector based on the frequency domain phase drift parameter, amplitude attenuation coefficient, photoelectric response delay index, and local temperature rise gradient; and use the excitation response feature vector to identify response anomalies and establish a third detection result.

[0075] Furthermore, the detection result output module 15 is also used to: map the first detection result, the second detection result, and the third detection result to a unified spatial coordinate system; calculate the detection consistency index of the same etched segment based on the mapping result; perform multi-domain interactive confidence analysis based on the calculation result; and complete interactive verification.

[0076] Furthermore, the detection result output module 15 is also used for: identifying the photovoltaic module based on the online detection result and configuring the splitting identifier; binding the online detection result with the unique code of the photovoltaic module and uploading it, and then managing the splitting transmission of the photovoltaic module based on the splitting identifier.

[0077] Furthermore, the detection result output module 15 is also used to: determine whether the trust value of the interactive verification is lower than a preset threshold; if the trust value is lower than the preset threshold, generate a detection anomaly identifier, and divert the corresponding photovoltaic module to the attention detection area according to the detection anomaly identifier, and perform attention re-detection processing.

[0078] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0079] Obviously, those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. An optical online detection method for laser-etched line breaks in photovoltaic modules, characterized in that, The method includes: The visible light image acquisition module and the spectral detection module are activated sequentially at the detection station to perform scanning acquisition of the target photovoltaic module and establish morphological texture image and spectral dataset. Spatial feature extraction and contour enhancement processing are performed on the morphological texture image to establish an etching line connectivity matrix. The explicit line breakage criterion is calculated using the pixel connectivity, geometric continuity and local gray-level gradient in the etching line connectivity matrix to establish a first detection result. Perform multi-segment energy ratio analysis and absorption peak position shift detection on the spectral dataset, and generate a spectral-spatial coupling vector after spatial constraint through the etching line connectivity matrix. Establish a second detection result based on the spectral-spatial coupling vector. Perform electrical excitation tests on photovoltaic modules and record the excitation response set, which includes frequency domain phase, amplitude response, and photoelectric coupling response; A third detection result is established based on the stimulus response set, and the first, second, and third detection results are interactively verified to output the online detection result. After spatial constraint through the etching line connectivity matrix, a spectral-spatial coupling vector is generated, including: The weighted energy offset spectrum is spatially matched with the etching line connectivity matrix to limit the spatial projection range of the abnormal energy response, generating a spectral-spatial coupling vector. The spectral-spatial coupling vector characterizes the energy perturbation mode and geometric connectivity of the etching path. The spectral-spatial coupling vector is normalized, and the spectral weight coefficients and spatial connectivity coefficients of the coupled components are calculated to establish a weighted coupling feature set. Under the constraint of the etching line connectivity matrix, the local statistical parameters of the weighted coupling feature set are calculated using the sliding neighborhood as the computational unit. The local statistical parameters include the energy anomaly score, connectivity consistency score, and spectral band stability score. The local statistical parameters are subjected to multi-domain fusion determination, and a second detection result is output. The explicit line breakage criterion is calculated using pixel connectivity, geometric continuity, and local gray-level gradient in the etching line connectivity matrix, establishing the first detection result, including: The etching line region is located using the etching line connectivity matrix, and a binarized pixel map of the etching line region is obtained. Pixel connectivity index is calculated based on the binarized pixel map, and a topological connectivity graph of the etch line trunk and branches is constructed. In the topological connectivity graph, a line segment fitting algorithm is used to calculate the local geometric continuity index, which is calculated by the directional angle and the rate of change of curvature of adjacent line segments. Calculate the ratio of grayscale change along the etching line direction to that along the normal direction, and calculate the local grayscale gradient vector; The explicit line break criterion is calculated based on the pixel connectivity index, local geometric continuity index, and local gray-level gradient vector. A third detection result is established based on the stimulus response set, including: Time series difference analysis was performed on the excitation response set to extract the spectral energy difference distribution and voltage response change rate before and after the excitation signal was applied, and a dynamic response curve set was established. Perform multi-domain feature decoupling of the dynamic response curve set, calculate frequency domain phase drift parameters, amplitude attenuation coefficient, photoelectric response delay index, and extract local temperature rise gradient using thermal response feature set; An excitation response feature vector is established based on frequency domain phase drift parameters, amplitude attenuation coefficient, photoelectric response delay index, and local temperature rise gradient. The excitation response feature vector is used to identify response anomalies and establish a third detection result.

2. The optical online detection method for broken laser-etched lines in photovoltaic modules as described in claim 1, characterized in that, Performing multi-segment energy ratio analysis and absorption peak position shift detection on the spectral dataset includes: The spectral dataset is divided into N overlapping energy segments in the visible to near-infrared band. The energy ratio spectrum of adjacent wavelength intervals is calculated for the division results, and a multi-segment energy ratio matrix is ​​constructed, where N is an integer greater than 2. Correlation analysis between spectral segments is performed on the multi-segment energy ratio matrix to calculate the energy distribution gradient and spectral segment stability index, and energy discontinuity characteristics are identified based on the calculation results; Based on the energy discontinuity feature, the shift trend of the corresponding absorption peak is extracted, and a peak position shift vector is established; Using the rate of change of local energy ratios in N overlapping energy segments as weights, a weighted energy shift spectrum is established based on the peak position shift vector.

3. The optical online detection method for broken laser-etched lines in photovoltaic modules as described in claim 1, characterized in that, Perform electrical excitation tests on photovoltaic modules and record the excitation response set, including: A controllable excitation signal is applied to the photovoltaic module, the excitation signal including pulsed light excitation, micro-vibration excitation, and voltage excitation; Perform response acquisition of photovoltaic modules and establish response acquisition results, which include response time nodes, response spectral feature sets, response voltage feature sets, and thermal response feature sets; The collected response results are output as an excitation response set.

4. The optical online detection method for broken laser-etched lines in photovoltaic modules as described in claim 1, characterized in that, The first detection result, the second detection result, and the third detection result are interactively verified, including: Map the first detection result, the second detection result, and the third detection result to a unified spatial coordinate system; The detection consistency index of the same etched segment is calculated based on the mapping results. Multi-domain interactive confidence analysis is performed based on the calculation results to complete the interactive verification.

5. The optical online detection method for broken laser-etched lines in photovoltaic modules as described in claim 1, characterized in that, After outputting the online detection results, the following are included: Based on the online detection results, the photovoltaic modules are identified for current diversion, and current diversion identifiers are configured. After the online detection results are uploaded and bound to the unique code of the photovoltaic module, the photovoltaic module's transmission is managed according to the diversion identifier.

6. The optical online detection method for broken laser-etched lines in photovoltaic modules as described in claim 1, characterized in that, The system also includes interactive verification of the first, second, and third detection results, outputting online detection results, and further includes: Determine whether the trust value of the interactive verification is lower than a preset threshold; If the trust value is lower than the preset threshold, an anomaly detection flag is generated. Based on the anomaly detection flag, the corresponding photovoltaic module is diverted to the focus detection area for re-detection processing.

7. An online optical inspection system for broken laser-etched lines in photovoltaic modules, characterized in that, The steps for implementing the optical online detection method for laser etching line breakage in photovoltaic modules according to any one of claims 1 to 6 include: The scanning and acquisition module is used to sequentially activate the visible light image acquisition module and the spectral detection module at the detection station, respectively to perform scanning and acquisition of the target photovoltaic module, and to establish morphological texture images and spectral datasets; The first detection result establishment module is used to perform spatial feature extraction and contour enhancement processing on the morphological texture image, establish an etching line connectivity matrix, calculate the explicit line breakage criterion using the pixel connectivity, geometric continuity and local gray-level gradient in the etching line connectivity matrix, and establish the first detection result. The second detection result establishment module is used to perform multi-segment energy ratio analysis and absorption peak position shift detection of the spectral dataset, and generate a spectral-spatial coupling vector after spatial constraint through the etching line connectivity matrix, and establish the second detection result based on the spectral-spatial coupling vector; An electrical excitation test module is used to perform electrical excitation tests on photovoltaic modules and record an excitation response set, which includes frequency domain phase, amplitude response, and photoelectric coupling response. The detection result output module is used to establish a third detection result based on the stimulus response set, perform interactive verification on the first detection result, the second detection result, and the third detection result, and output the online detection result.

Images