A method and system for defect detection of photovoltaic cells

Abstract

A defect detection method and system for photovoltaic cells, an initial point pair set corresponding to each line laser profiler is obtained, a first correction slope corresponding to the line laser profiler is calculated, and then the initial point pair set corresponding to the line laser profiler is profile corrected to obtain a real-time point pair set corresponding to the line laser profiler; a first type of defect detection result is obtained according to the real-time point pair set corresponding to each line laser profiler; and / or a second type of defect detection result is obtained according to the real-time point pair set corresponding to each line laser profiler and the standard profile data of the three-dimensional vision sensor corresponding thereto. The error caused by mechanism jitter, installation non-parallelism and the like is suppressed or removed through profile correction and one-dimensional signal filtering and the like, so as to improve the repeatability and stability of measurement, so that the measurement data is close to the true value, and then the accuracy of the defect detection result is ensured, and on this basis, integrated measurement of various defects is realized.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic cell defect detection, and specifically to a defect detection method and system for photovoltaic cells. Background Technology

[0002] In the production process of photovoltaic cells, the photovoltaic cell is the core component of the photovoltaic power generation device. Its quality determines the photoelectric conversion efficiency of photovoltaic power generation and directly affects the processing quality of subsequent processes and the performance of the final product. However, photovoltaic cells have many types of defects, and some defects (such as line marks) often have a precision unit of micrometers, making traditional manual identification difficult. Therefore, achieving intelligent detection in the quality inspection of photovoltaic cells and modules is particularly important.

[0003] Existing methods often employ point lasers for defect detection in photovoltaic cells. However, point lasers have a long setup time and require high installation precision. Furthermore, their main drawbacks include:

[0004] (1) During the movement of the mechanism, due to factors such as mechanical structure, belt vibration, transmission device or external environment, vibration and jumping phenomena are inevitable, resulting in poor static repeatability of the overall measurement and deviation from the true value; and the existing method of defect detection by point laser cannot suppress the vibration generated by the movement of the mechanism.

[0005] (2) Non-parallel installation between the point laser rangefinder and the photovoltaic cell under test will produce non-linear errors, resulting in differences between the measured value and the true value, which in turn makes it impossible to achieve stable and accurate measurement; and small deviations in installation accuracy will be amplified through non-linear effects, further exacerbating the unreliability of the measurement results.

[0006] (3) In traditional methods, the defect detection method using point lasers can only perform spot checks on some points, which cannot cover the entire surface of the photovoltaic cell and cannot fully reflect the overall quality of the photovoltaic cell. It is easy to miss detection, which can lead to potential quality problems and make it difficult to meet the production requirements of high precision and high quality.

[0007] Therefore, it is necessary to design a method that can effectively suppress the jitter interference generated during the movement of the mechanism, remove the nonlinear error caused by non-parallel installation, and on this basis, realize an integrated measurement and defect detection method for multiple defect types on photovoltaic cells. Summary of the Invention

[0008] The main technical problem solved by this invention is how to effectively suppress the jitter interference generated during the movement of the mechanism, remove the nonlinear error caused by non-parallel installation, and on this basis realize the integrated measurement and detection of multiple defect types on photovoltaic cells.

[0009] According to a first aspect, one embodiment provides a defect detection method for photovoltaic cells, applied to a defect detection system, the defect detection system including multiple sets of line laser profilometers, each set of line laser profilometers including multiple line laser profilometers, each of the line laser profilometers including a three-dimensional vision sensor, the defect detection method including:

[0010] For any of the line laser profilometers: based on the initial profile data of the photovoltaic cell under test collected by the line laser profilometer, an initial set of point pairs corresponding to the line laser profilometer is obtained; based on the initial set of point pairs corresponding to the line laser profilometer, a first correction slope corresponding to the line laser profilometer is calculated; based on the first correction slope corresponding to the line laser profilometer, profile correction is performed on the initial set of point pairs corresponding to the line laser profilometer to obtain the real-time set of point pairs corresponding to the line laser profilometer.

[0011] A first type of defect detection result is obtained based on the real-time point pair set corresponding to each of the line laser profilometers, the first type of defect detection result including line mark detection result and warpage detection result; and / or, a second type of defect detection result is obtained based on the real-time point pair set corresponding to each of the line laser profilometers and the standard profile data of the three-dimensional vision sensor corresponding to each of the line laser profilometers, the second type of defect detection result including thickness detection result.

[0012] According to a second aspect, one embodiment provides a defect detection system for photovoltaic cells, comprising:

[0013] Multiple sets of line laser profilometers, each set including multiple line laser profilometers; each line laser profilometer is used to acquire initial profile data of the photovoltaic cell under test; each line laser profilometer includes a three-dimensional vision sensor;

[0014] And a processor, used for:

[0015] For any of the line laser profilometers: based on the initial profile data of the photovoltaic cell under test collected by the line laser profilometer, an initial set of point pairs corresponding to the line laser profilometer is obtained; based on the initial set of point pairs corresponding to the line laser profilometer, a first correction slope corresponding to the line laser profilometer is calculated; based on the first correction slope corresponding to the line laser profilometer, profile correction is performed on the initial set of point pairs corresponding to the line laser profilometer to obtain the real-time set of point pairs corresponding to the line laser profilometer.

[0016] A first type of defect detection result is obtained based on the real-time point pair set corresponding to each of the line laser profilometers, the first type of defect detection result including line mark detection result and warpage detection result; and / or, a second type of defect detection result is obtained based on the real-time point pair set corresponding to each of the line laser profilometers and the standard profile data of the three-dimensional vision sensor corresponding to each of the line laser profilometers, the second type of defect detection result including thickness detection result.

[0017] According to the above embodiment, a defect detection method and system for photovoltaic cells involves acquiring contour data of the photovoltaic cell using multiple sets of line laser profilometers to cover the entire surface of the photovoltaic cell, ensuring that the acquired contour data comprehensively reflects the overall quality of the photovoltaic cell. Based on the contour data acquired by each line laser profilometer, a first correction slope is calculated, and the initial contour data of that line laser profilometer is then corrected to obtain a real-time point pair set corresponding to that line laser profilometer. This contour correction method removes nonlinear tilt errors caused by non-parallel installation. A first type of defect detection (line mark detection and warpage detection) is then performed based on the real-time point pair set corresponding to each line laser profilometer, resulting in a first type of defect detection result. During this process, frequency... One-dimensional signal filtering is performed using domain or spatial domain filtering to remove spikes and jitter in the real-time contour data corresponding to the line laser profilometer, thereby effectively suppressing jitter and ensuring the reliability of the detection results. Second-type defect detection (thickness detection) is performed based on the real-time point pair set corresponding to each line laser profilometer and the standard contour data of the corresponding 3D vision sensor, yielding the second-type defect detection results. Errors caused by mechanism jitter and non-parallel installation are suppressed or removed through tilt correction and one-dimensional signal filtering, improving the repeatability and stability of the measurement, ensuring that the measured data is close to the true value, and thus ensuring the accuracy of the defect detection results. Based on this, integrated detection of various defect types in photovoltaic cells, including thickness, line marks, and warpage defects, is achieved. Attached Figure Description

[0018] Figure 1 This is a flowchart of a defect detection method for photovoltaic cells;

[0019] Figure 2 This is a system block diagram of a defect detection system for photovoltaic cells;

[0020] Figure 3 This is a schematic diagram showing real-time contour data before and after filtering. Detailed Implementation

[0021] The present invention will now be described in further detail with reference to specific embodiments and accompanying drawings. Similar elements in different embodiments are referred to by associated similar element reference numerals. In the following embodiments, many details are described to facilitate a better understanding of this application. However, those skilled in the art will readily recognize that some features may be omitted in different situations, or may be replaced by other elements, materials, or methods. In some cases, certain operations related to this application are not shown or described in the specification. This is to avoid obscuring the core parts of this application with excessive description. For those skilled in the art, detailed description of these related operations is not necessary; they can fully understand the related operations based on the description in the specification and general technical knowledge in the art.

[0022] Furthermore, the features, operations, or characteristics described in the specification can be combined in any suitable manner to form various embodiments. At the same time, the steps or actions in the method description can be rearranged or adjusted in a manner obvious to those skilled in the art. Therefore, the various orders in the specification and drawings are only for the clear description of a particular embodiment and do not imply a necessary order, unless otherwise stated that a particular order must be followed.

[0023] The serial numbers assigned to components in this document, such as "first" and "second," are used only to distinguish the described objects and have no sequential or technical meaning. The terms "connection" and "linkage" used in this application, unless otherwise specified, include both direct and indirect connections (linkages).

[0024] A line laser profilometer emits a laser line onto the surface of the object being measured, while simultaneously receiving the reflected light from the object's surface using a 3D vision sensor. Based on the principle of triangulation, the reflection information formed by the laser line on the object's surface is calculated to obtain the height distribution of points within the laser line's scanning area, thus generating the corresponding contour data. In the final generated contour data, each contour point corresponds to a specific location on the object's surface, and the numerical value corresponding to that contour point represents the actual height at that location.

[0025] In this embodiment of the invention, firstly, multiple sets of line laser profilometers are used to collect contour data of the photovoltaic cell to cover the entire surface of the photovoltaic cell. Then, contour correction is performed on the initial contour data and standard contour data corresponding to each line laser profilometer to remove nonlinear errors caused by non-parallel installation. Next, based on the initial contour data collected by each line laser profilometer, a real-time point pair set corresponding to that line laser profilometer is obtained for the first type of defect detection (including line mark detection and warpage detection). During this process, one-dimensional signal filtering is performed by frequency domain filtering or spatial domain low-pass filtering to remove spike jitter in the real-time contour data corresponding to that line laser profilometer, thereby effectively suppressing jitter and ensuring the reliability of the detection results. Finally, based on the real-time point pair set corresponding to each line laser profilometer and the standard contour data of its corresponding three-dimensional vision sensor, a second type of defect detection (thickness detection) is performed. By using tilt correction (contour correction) and one-dimensional signal filtering, the effects of mechanism vibration and non-parallel installation are effectively suppressed or removed, ensuring the stability of online detection and the accuracy of measurement results. On this basis, it is possible to achieve integrated measurement of various defect types of photovoltaic cells, including thickness, line marks, warpage, etc.

[0026] Some embodiments provide a defect detection method for photovoltaic cells, which is applied to a defect detection system, wherein the defect detection system includes multiple sets of line laser profilometers, each set of line laser profilometers includes multiple line laser profilometers, and each line laser profilometer includes a three-dimensional vision sensor.

[0027] In this embodiment, for a photovoltaic cell on the production line, multiple line laser profilometers are respectively set on the upper and lower sides of the photovoltaic cell. Taking the upper side of the photovoltaic cell as an example, multiple line laser profilometers are set on the upper side of the photovoltaic cell. For example, three line laser profilometers can be set on the upper side of the photovoltaic cell; wherein the laser line direction of each line laser profilometer is parallel to the movement direction of the photovoltaic cell.

[0028] Assuming that the side corresponding to the length direction of the photovoltaic cell is parallel to the direction of movement of the photovoltaic cell, then the side corresponding to the width direction of the photovoltaic cell is perpendicular to the direction of movement of the photovoltaic cell. In this embodiment, the line laser emitted by the three line laser profilometers can cover the entire width of the photovoltaic cell. For ease of description, this embodiment divides the width of the photovoltaic cell into three equal parts, which are denoted as the left, middle and right parts of the photovoltaic cell, respectively. That is, each part corresponds to one line laser profilometer.

[0029] As the photovoltaic cell moves, each line laser profilometer will collect data multiple times. To ensure that the final collected profile data covers the entire upper surface of the photovoltaic cell, in this embodiment, the profile data obtained by each line laser profilometer after each acquisition will have a region of positional overlap with the profile data obtained in the previous acquisition. This ensures that the corresponding profile data at each position on the upper surface of the photovoltaic cell is collected, avoiding omissions. During this process, the overlapping region in the profile data obtained from two adjacent acquisitions can be determined by controlling the movement speed of the photovoltaic cell and the acquisition frequency of the line laser profilometer.

[0030] Three line laser profilometers are also set on the lower side of the photovoltaic cell, corresponding to the left, middle and right parts of the corresponding side in the width direction of the photovoltaic cell, respectively. Then, two line laser profilometers on the upper and lower sides of the photovoltaic cell and in the same part of the corresponding side in the width direction of the photovoltaic cell constitute a set of line laser profilometers. That is to say, in this embodiment, there are a total of three sets of line laser profilometers, which correspond to the left, middle and right parts in the width direction of the photovoltaic cell, respectively, and each set of line laser profilometers contains two line laser profilometers.

[0031] Please refer to Figure 1 The defect detection method in this embodiment includes the following steps:

[0032] Step S100: For any line laser profilometer: Based on the initial profile data of the photovoltaic cell under test collected by the line laser profilometer, obtain the initial set of point pairs corresponding to the line laser profilometer.

[0033] Each line laser profilometer collects data from the photovoltaic cell under test, obtaining multiple initial profile data of the photovoltaic cell under test. For any line laser profilometer, each profile point in the initial profile data obtained by the line laser profilometer corresponds to a position on the photovoltaic cell, and the value corresponding to each profile point represents the height of the position corresponding to that profile point.

[0034] Since the initial contour data obtained by each line laser profilometer has a certain overlapping area, it is necessary to remove the overlapping area from the obtained contour data and renumber the index of each contour point in the initial contour data after removing the overlapping area, so that each contour point uniquely corresponds to a point index.

[0035] Then, based on the point index of each contour point and its corresponding height, the initial contour data after removing overlapping areas is transformed into a set of point pairs. The resulting set of point pairs is the initial set of point pairs corresponding to the line laser profilometer. For example, for the i-th contour point, the ordinate of its corresponding point pair is the height z of that contour point. iThe x-coordinate of the point pair corresponding to the contour point i The point index i of the contour point is equal to the distance r between the adjacent contour points along the contour line direction. x The product between them, i.e., x i =i×r x Then the point pair corresponding to the i-th contour point is (x i , z i ); and so on, all the corresponding point pairs of the contour points constitute the initial set of point pairs corresponding to the line laser profilometer.

[0036] It should be noted that, for ease of description, in this embodiment, the initial contour data after removing the overlapping area corresponding to each line laser profilometer is still referred to as the initial contour data of that line laser profilometer. That is, the initial contour data processed below is actually the initial contour data after removing the overlapping area.

[0037] Step S110: Calculate the first correction slope corresponding to the line laser profiler based on the initial set of point pairs corresponding to the line laser profiler.

[0038] Because the initial contour data collected by the line laser profilometer is not parallel to the photovoltaic cell under test during installation, it is generally tilted. This causes the measurement results of thickness and other parameters to have an additional height deviation due to the tilt. Therefore, this embodiment uses a stable contour correction method to correct the tilt of the initial contour data corresponding to each line laser profilometer, thereby eliminating the height deviation caused by factors such as non-parallel installation.

[0039] Assume the relationship between the measured height value and the corresponding location point is expressed as z. i =kx i +t, based on the calculated slope k, rotate and correct each point pair in the initial point pair set to obtain the point pair set after contour correction. At this time, the slope k is the first correction slope corresponding to the line laser profilometer, and the point pair set obtained after contour correction is the real-time point pair set of the line laser profilometer.

[0040] In this embodiment, the first correction slope can be calculated either directly or iteratively. Specifically, the first correction slope can be directly obtained from the initial set of point pairs corresponding to the line laser profilometer. The method for obtaining the first correction slope is as follows:

[0041]

[0042] Where k represents the first correction slope; n represents the total number of point pairs in the initial point pair set corresponding to the line laser profilometer; x i This represents the x-coordinate of the i-th point pair in the initial set of point pairs corresponding to the line laser profilometer; zi This represents the ordinate of the i-th point pair in the initial set of point pairs corresponding to the line laser profilometer.

[0043] Alternatively, the first correction slope can be obtained iteratively based on the initial set of point pairs corresponding to the line laser profilometer. In any iteration:

[0044] First, the basic correction slope is obtained based on the fitted line in the current iteration round. If the current iteration round is the first iteration, the fitted line in the current iteration round is obtained by fitting a straight line to the initial set of point pairs corresponding to the line laser profilometer. For example, the fitted line can be obtained by fitting a straight line to each point pair in the initial set of point pairs corresponding to the line laser profilometer using the least squares method. If the current iteration round is not the first iteration, the fitted line in the current iteration round is the fitted line updated in the previous iteration round.

[0045] The reference weight of each point pair is determined based on the distance between each point pair in the initial set of point pairs and the fitted line. For example, for the distance between each point pair in the initial set of point pairs and the fitted line, the maximum and minimum distance values ​​among all the obtained distances are obtained. Then, for the distance corresponding to any point pair in the initial set of point pairs, the maximum and minimum distance values ​​are normalized. The normalized result is the reference weight of the point pair, that is, the reference weight of each point pair is in the range of 0 to 1. Among them, the smaller the distance between a point pair and the fitted line, the larger the reference weight of the point pair; conversely, the larger the distance, the smaller the reference weight of the point pair.

[0046] Taking the i-th pair of points as an example, let d be the distance between this pair of points and the fitted line. i The maximum distance is d max The minimum distance is d min Then the reference weight w of that point pair i It can be represented as: w i =(d max -d i ) / (d max -d min ).

[0047] The basic correction slope is updated based on each point pair and its reference weight in the initial set of point pairs to obtain the updated basic correction slope in the current iteration; whereby the updated basic correction slope in the current iteration can be expressed as:

[0048]

[0049] Where k represents the updated baseline correction slope in the current iteration; n represents the total number of point pairs in the initial point pair set corresponding to the line laser profilometer; w i This represents the reference weight of the i-th point pair in the initial set of point pairs corresponding to the line laser profilometer; x i This represents the x-coordinate of the i-th point pair in the initial set of point pairs corresponding to the line laser profilometer; z i This represents the ordinate of the i-th point pair in the initial set of point pairs corresponding to the line laser profilometer.

[0050] If the current iteration is not the preset iteration, the fitted line is updated based on the updated basic correction slope to obtain the updated fitted line in the current iteration. That is, the updated first correction slope is used as the slope of the fitted line, thus obtaining the updated fitted line; and then proceed to the next iteration.

[0051] If the current iteration is the preset iteration, then the iteration ends, and the updated basic correction slope in the current iteration is used as the first correction slope.

[0052] Step S120: Based on the first correction slope corresponding to the line laser profiler, perform profile correction on the initial set of point pairs corresponding to the line laser profiler to obtain the real-time set of point pairs corresponding to the line laser profiler.

[0053] First, the correction center is determined based on the initial set of point pairs corresponding to the line laser profilometer, where the x-coordinate of the correction center is x. c The ordinate z of the correction center is the average of the x-coordinates of all point pairs in the initial set of point pairs. c The average of the ordinates of all point pairs in the initial set of point pairs is given; the correction angle is determined based on the first correction slope, where the correction angle in this embodiment is θ = arctan(k).

[0054] Then, based on the obtained correction center and correction angle, each point pair in the initial point pair set corresponding to the line laser profilometer is corrected. The point pair obtained after contour correction is the contour correction result of that point pair. The contour correction method in this embodiment is as follows:

[0055]

[0056] Among them, X i Z represents the x-coordinate of the corresponding point pair after contour correction for the i-th point pair; i Represents the ordinate of the i-th point pair after contour correction; θ is the correction angle; (x c , z c ) indicates the calibration center.

[0057] Each point pair in the initial point pair set corresponding to the line laser profilometer is processed sequentially, and the set of all point pair profile correction results is the real-time point pair set corresponding to the line laser profilometer.

[0058] Step S130: Obtain the first type of defect detection result based on the real-time point pair set corresponding to each line laser profilometer, and / or obtain the second type of defect detection result based on the real-time point pair set corresponding to each line laser profilometer and the standard profile data of the three-dimensional vision sensor corresponding to each line laser profilometer.

[0059] "New contour data" is generated based on the ordinate of each point pair in the real-time point pair set corresponding to the line laser profilometer. Since the real-time point pair set is a set of point pairs after tilt correction, the "new contour data" obtained based on this real-time point pair set is also contour data that has removed nonlinear errors caused by factors such as non-parallel installation.

[0060] Considering that factors such as multiple reflections and overexposure of the line laser profilometer may lead to abnormal data in the acquired contour data, in order to ensure the accuracy of the detection results, the abnormal data in the "new contour data" can be removed by a statistical threshold method. In this embodiment, the contour data after removing the abnormal data in the "new contour data" is recorded as the real-time contour data corresponding to the line laser profilometer.

[0061] In this embodiment, the first type of defect detection results include line mark detection results and warpage detection results; wherein, during line mark detection, the real-time contour data corresponding to the line laser profilometer is filtered, and the specific filtering process is as follows:

[0062] Based on the real-time contour data corresponding to the line laser profilometer, jitter-suppressed contour data corresponding to the line laser profilometer is obtained; by performing jitter suppression processing on the real-time contour data corresponding to the line laser profilometer, jitter-suppressed contour data corresponding to the line laser profilometer is obtained; the jitter suppression processing in this embodiment includes spatial domain low-pass filtering or frequency domain filtering to remove peak jitter in the real-time contour data corresponding to the line laser profilometer.

[0063] In the process of jitter suppression through frequency domain filtering, the real-time contour data corresponding to the line laser profilometer is subjected to Fourier transform to obtain the frequency domain signal; then the obtained frequency domain signal is multiplied by the sinc function, where sinc(f) = sin(πf) / (πf), and f represents the frequency variable.

[0064] Then, the dot product result is inversely transformed, and the result of the inverse transformation is the jitter-suppressed contour data corresponding to the line laser profilometer in this embodiment.

[0065] Spatial domain low-pass filtering can directly smooth real-time contour data in the time or spatial domain, thereby retaining low-frequency components and suppressing high-frequency components through local neighborhood operations, thus removing spike jitter in real-time contour data and achieving jitter suppression.

[0066] Based on the jitter suppression profile data corresponding to the line laser profiler, stable profile data corresponding to the line laser profiler is obtained. In this embodiment, discrete point cleaning processing is further performed on the jitter suppression profile data corresponding to the line laser profiler to obtain stable profile data corresponding to the line laser profiler. The discrete point cleaning processing in this embodiment includes median filtering and mean filtering to remove discrete points in the jitter suppression profile data corresponding to the line laser profiler.

[0067] Furthermore, before performing discrete point cleaning on the jitter suppression profile data corresponding to the line laser profilometer, it can be determined whether discrete point cleaning is necessary based on the degree of jitter in the jitter suppression profile data, thereby reducing the computational load and improving detection efficiency.

[0068] For example, the jitter level of the jitter-suppressed contour data is characterized by the variance or deviation from the mean of the jitter-suppressed contour data corresponding to the line laser profilometer. When the jitter level is greater than or equal to a preset threshold, it indicates that the jitter level of the jitter-suppressed contour data is still relatively large, and further spatial filtering is required to obtain stable contour data. When the jitter level is less than the preset threshold, it indicates that the jitter level of the jitter-suppressed contour data is relatively small and stable, and spatial filtering is not required. In this case, the jitter-suppressed contour data corresponding to the line laser profilometer is the stable contour data corresponding to the line laser profilometer.

[0069] In this embodiment, the real-time contour data before and after filtering is as follows: Figure 3 As shown in the figure, the final result after filtering the real-time contour data is stable contour data. It can be seen from the figure that the obtained stable contour data has removed the spike jitter in the real-time contour data. Therefore, it can be seen that the filtering method in this embodiment can effectively suppress the jitter interference generated during the movement of the mechanism.

[0070] The line mark value corresponding to the line laser profilometer is obtained based on the stable profile data. In this embodiment, the first 100 and last 100 profile points in the stable profile data are removed to avoid the influence of the boundary effect generated by the edge of the photovoltaic cell under test on the detection result. Then, the peak and valley values ​​of the stable profile data are obtained. The difference between each peak value and its adjacent valley value is calculated, and the maximum value of the difference is taken as the line mark value corresponding to the peak value. The largest line mark value among all the line mark values ​​corresponding to the peak values ​​is taken as the line mark value corresponding to the line laser profilometer.

[0071] The line mark detection results of the photovoltaic cell under test are obtained based on the line mark values ​​corresponding to all line laser profilometers. In this embodiment, the maximum value among the line mark values ​​corresponding to all line laser profilometers is taken as the line mark value of the photovoltaic cell under test. By comparing the obtained line mark value with the preset line mark threshold, it is determined whether the photovoltaic cell under test meets the production requirements. For example, if the line mark value of the photovoltaic cell under test is greater than the preset line mark threshold, it is considered that the photovoltaic cell under test has a defect. Otherwise, it is considered that the photovoltaic cell under test does not have a defect. In the subsequent production process, the photovoltaic cells with defects can be sorted, thereby completing the line mark defect detection of the photovoltaic cell under test.

[0072] When detecting warpage, the real-time contour data corresponding to each line laser profilometer is obtained based on the set of real-time point pairs corresponding to each line laser profilometer. Each contour point in the real-time contour data corresponds to a height value.

[0073] Obtain the maximum and minimum height values ​​of all contour points in the real-time contour data corresponding to the line laser profilometer, and then obtain the maximum and minimum height values ​​based on the real-time contour data corresponding to the line laser profilometer.

[0074] Then, based on the obtained maximum and minimum height values, the warp degree corresponding to the line laser profilometer is obtained. In this embodiment, the warp degree corresponding to the line laser profilometer is calculated based on the sign of the maximum and minimum height values. Specifically, if the maximum and minimum height values ​​have the same sign, the larger absolute value of the absolute values ​​of the maximum and minimum height values ​​is taken as the warp degree corresponding to the line laser profilometer. If the maximum and minimum height values ​​have opposite signs, the absolute value of the difference between the maximum and minimum height values ​​is taken as the warp degree corresponding to the line laser profilometer.

[0075] Finally, the warpage detection result of the photovoltaic cell under test is obtained based on the warpage corresponding to all line laser profilometers. For example, the average value of the warpage corresponding to all line laser profilometers can be used as the warpage of the photovoltaic cell under test, or the maximum or median value of the warpage corresponding to all line laser profilometers can be used as the warpage of the photovoltaic cell under test. Subsequently, by comparing the obtained warpage with the preset warpage threshold, it can be determined whether the photovoltaic cell under test meets the production requirements. For example, if the warpage of the photovoltaic cell under test is greater than the preset warpage threshold, it is considered that the photovoltaic cell under test has a defect; otherwise, it is considered that the photovoltaic cell under test does not have a defect, thus completing the warpage defect detection of the photovoltaic cell under test.

[0076] The second type of defect detection result in this embodiment includes the thickness detection result. The second type of defect detection result is obtained based on the real-time point pair set corresponding to each line laser profilometer and the standard contour data of the three-dimensional vision sensor corresponding to each line laser profilometer. The standard contour data of each three-dimensional vision sensor is obtained by calibrating the three-dimensional vision sensors corresponding to the two line laser profilometers in each group of line laser profilometers using a standard block. That is, for the two line laser profilometers in the same group of line laser profilometers, the contour data collected by the three-dimensional vision sensors corresponding to the two line laser profilometers are aligned to the same coordinate system, thereby obtaining the standard contour data of each three-dimensional vision sensor, and the thickness of the standard block is used as the standard height h.

[0077] The standard contour data of the 3D vision sensor obtained using the standard block may still introduce nonlinear errors due to the non-parallel installation of the line laser profilometer and the standard block. This error will cause the standard contour data itself to deviate from the ideal value. Therefore, directly comparing the actual acquired contour data with the standard contour data cannot eliminate the influence of non-parallel installation. The difference between the two cannot accurately reflect the true situation of the photovoltaic cell.

[0078] Therefore, in this embodiment, when detecting the thickness of the photovoltaic cell under test, the standard point pair set corresponding to the line laser profilometer is first obtained based on the standard contour data of the three-dimensional vision sensor corresponding to the line laser profilometer; the second correction slope corresponding to the line laser profilometer is calculated based on the standard point pair set; and contour correction is performed on the standard point pair set corresponding to the line laser profilometer based on the second correction slope to obtain the reference point pair set corresponding to the line laser profilometer. The method for obtaining the standard point pair set based on the standard contour data is the same as the method for obtaining the initial point pair set based on the initial contour data described above. The method for contour correction of the standard point pair set based on the second correction slope is the same as the method for contour correction of the initial point pair set based on the first correction slope described above, and will not be repeated here.

[0079] Based on the set of reference points corresponding to the line laser profilometer, the reference line corresponding to the line laser profilometer is obtained; the obtained reference line can be represented as aY. i +bH i +c=0, and satisfy a 2 +b 2 =1; where Y i H represents the x-coordinate of the i-th point pair in the set of reference point pairs; i This represents the ordinate of the i-th point pair in the set of reference point pairs.

[0080] In this embodiment, the parameters a and b of the reference line are obtained as follows: The average values ​​of the x-coordinate and y-coordinate of each pair of points in the reference point pair set are calculated respectively. The average value of the obtained x-coordinate is recorded as the first mean, and the average value of the obtained y-coordinate is recorded as the second mean. For any pair of points in the reference point pair set: the difference between the x-coordinate of the pair and the first mean is recorded as the first difference of the pair; the difference between the y-coordinate of the pair and the second mean is recorded as the second difference of the pair; a matrix is ​​then constructed based on the first and second differences of each pair of points in the reference point pair set; and then eigenvalue decomposition is performed on the obtained matrix, i.e.:

[0081]

[0082] Where a and b are two parameters of the reference line; Y i Represents the x-coordinate of the i-th point pair in the set of reference point pairs; H represents the average of the x-coordinates of all point pairs in the set of reference points, i.e., the first mean; i Represents the ordinate of the i-th point pair in the set of reference point pairs; The second mean is the average of the ordinates of all point pairs in the set of reference points. This represents the first difference of the i-th point pair in the set of reference point pairs; λ represents the second difference of the i-th point pair in the reference point pair set; n is the total number of point pairs in the reference point pair set; λ represents the eigenvalue.

[0083] After decomposing the eigenvalues, the elements in the eigenvector corresponding to the smallest eigenvalue are used as the values ​​of parameters a and b of the reference line, thereby obtaining the reference line corresponding to the laser profilometer.

[0084] The thickness component corresponding to the line laser profiler is obtained based on the real-time point pair set corresponding to the line laser profiler and the reference line corresponding to the line laser profiler. In this embodiment, the distance from each point pair in the real-time point pair set corresponding to the line laser profiler to the reference line is calculated, and the average value of the corresponding distances of all point pairs in the real-time point pair set is taken as the thickness component corresponding to the line laser profiler. The thickness component obtained at this time can also be called the distance between the real-time profile data corresponding to the line laser profiler and the reference profile data.

[0085] The thickness value of each group of line laser profilometers is obtained by considering the thickness components of all line laser profilometers in each group. Since the two line laser profilometers in the same group are located on opposite sides of the photovoltaic cell under test, the thickness value of the group of line laser profilometers can be obtained by considering the thickness components of these two line laser profilometers and the standard height obtained during the calibration process of the standard profile data of the 3D vision sensor.

[0086] It should be noted that in this embodiment, the direction closer to the 3D vision sensor is the negative direction, and the direction farther away from the 3D vision sensor is the positive direction; for example, for a line laser profilometer located on the upper side, the line laser profilometer has collected two sets of contour data A (equivalent to real-time contour data) and B (equivalent to reference contour data);

[0087] When contour data A is closer to the 3D vision sensor, it is considered to be in the negative direction of contour data B, and the distance of contour data A relative to contour data B is negative, meaning the corresponding thickness component is negative. Since contour data A is closer to the 3D vision sensor, the surface of the photovoltaic cell corresponding to contour data A is closer to the 3D vision sensor, so the thickness of this photovoltaic cell should be thicker than the standard block. Therefore, the thickness component should be subtracted from the standard height h. Conversely, when contour data A is farther from the 3D vision sensor, it is considered to be in the positive direction of contour data B, and the resulting thickness component is positive. Since contour data A is farther from the 3D vision sensor, the surface of the photovoltaic cell corresponding to contour data A is farther from the 3D vision sensor, so the photovoltaic cell should be thinner than the standard block. Therefore, the thickness component should also be subtracted from the standard height h. Thus, in this embodiment, the thickness value corresponding to one set of line laser profilometers is equal to the standard height h minus the thickness components corresponding to the two line laser profilometers in that set.

[0088] The second type of defect detection result of the photovoltaic cell under test is obtained based on the thickness value corresponding to each set of line laser profilometers. Since each set of line laser profilometers corresponds to the left, middle and right parts of the photovoltaic cell under test in the width direction, the average value of the thickness value corresponding to each set of line laser profilometers is used as the thickness value of the photovoltaic cell under test. Subsequently, by comparing the thickness value of the photovoltaic cell under test with the preset thickness range, it is determined whether the photovoltaic cell under test meets the production requirements. For example, if the thickness value of the photovoltaic cell under test exceeds the preset thickness range, the photovoltaic cell under test is considered to be too thick or too thin, and it is considered to have a defect. Otherwise, it is considered that the photovoltaic cell under test does not have a defect, thus completing the thickness defect detection of the photovoltaic cell under test.

[0089] This embodiment uses multiple sets of line laser profilometers to collect contour data of photovoltaic cells, covering the entire surface of the photovoltaic cells, so that the collected contour data can comprehensively reflect the overall quality of the photovoltaic cells. Based on the initial contour data and corresponding standard contour data collected by each line laser profilometer, a first correction slope and a second correction slope are calculated respectively. Contour correction is then performed on the initial contour data and standard contour data of each line laser profilometer, thereby removing nonlinear tilt errors caused by non-parallel installation. Then, the real-time point pair set corresponding to each line laser profilometer is obtained, and the first type of defect detection (line mark detection and warpage detection) is performed to obtain the first type of defect detection results. During this process, one-dimensional signal filtering is performed using frequency domain or spatial domain filtering to remove the line marks. The spike jitter in the real-time contour data corresponding to the laser profilometer is effectively suppressed, ensuring the reliability of the defect detection results. A set of reference point pairs is obtained based on the standard contour data of the three-dimensional vision sensor corresponding to each line laser profilometer after contour correction. Based on the real-time point pair set corresponding to each line laser profilometer and the set of reference point pairs, a second type of defect detection (thickness detection) is performed to obtain the second type of defect detection results. This embodiment uses contour correction and one-dimensional signal filtering to suppress or remove errors caused by mechanism jitter, non-parallel installation, etc., thereby improving the repeatability and stability of the measurement, ensuring that the measured data is close to the true value, and thus ensuring the accuracy of the defect detection results. On this basis, it further realizes the integrated measurement of various defect types of photovoltaic cells, including thickness, line marks, warpage, etc.

[0090] Please refer to Figure 2 Some embodiments provide a defect detection system for photovoltaic cells, including:

[0091] Multiple sets of line laser profilometers 20, each set of line laser profilometers includes multiple line laser profilometers; each line laser profilometer is used to acquire the initial profile data of the photovoltaic cell under test; each line laser profilometer includes a three-dimensional vision sensor 22.

[0092] And processor 21, used for:

[0093] For any line laser profilometer: Based on the initial profile data of the photovoltaic cell under test collected by the line laser profilometer, obtain the initial set of point pairs corresponding to the line laser profilometer; calculate the first correction slope corresponding to the line laser profilometer based on the initial set of point pairs corresponding to the line laser profilometer; perform profile correction on the initial set of point pairs corresponding to the line laser profilometer based on the first correction slope corresponding to the line laser profilometer to obtain the real-time set of point pairs corresponding to the line laser profilometer.

[0094] The first type of defect detection result is obtained based on the real-time point pair set corresponding to each line laser profilometer, wherein the first type of defect detection result includes line mark detection result and warpage detection result; and / or, the second type of defect detection result is obtained based on the real-time point pair set corresponding to each line laser profilometer and the standard profile data of the three-dimensional vision sensor corresponding to each line laser profilometer, wherein the second type of defect detection result includes thickness detection result.

[0095] It should be noted that the processing steps of the processor in this embodiment correspond to the method steps in the above-described defect detection method for photovoltaic cells. The specific implementation of the method has been described in detail in the above embodiments and will not be repeated here.

[0096] Those skilled in the art will understand that all or part of the functions of the various methods in the above embodiments can be implemented by hardware or by computer programs. When all or part of the functions in the above embodiments are implemented by computer programs, the program can be stored in a computer-readable storage medium, which may include: read-only memory, random access memory, disk, optical disk, hard disk, etc., and the program is executed by a computer to achieve the above functions. For example, the program can be stored in the memory of a device, and when the program in the memory is executed by the processor, all or part of the above functions can be achieved. In addition, when all or part of the functions in the above embodiments are implemented by computer programs, the program can also be stored in a server, another computer, disk, optical disk, flash drive, or external hard drive, etc., and can be downloaded or copied to the memory of a local device, or the system of the local device can be updated. When the program in the memory is executed by the processor, all or part of the functions in the above embodiments can be achieved.

[0097] The above examples illustrate the present invention only to aid in understanding it and are not intended to limit the scope of the invention. Those skilled in the art can make various simple deductions, modifications, or substitutions based on the principles of this invention.

Claims

1. A method for defect detection of photovoltaic cells, applied to a defect detection system, the defect detection system comprising a plurality of groups of line laser profilers, each group of line laser profilers comprising a plurality of line laser profilers, each line laser profiler comprising a three-dimensional vision sensor, characterized in that, The defect detection method includes: For any of the line laser profilometers: based on the initial profile data of the photovoltaic cell under test collected by the line laser profilometer, an initial set of point pairs corresponding to the line laser profilometer is obtained; based on the initial set of point pairs corresponding to the line laser profilometer, a first correction slope corresponding to the line laser profilometer is calculated; based on the first correction slope corresponding to the line laser profilometer, profile correction is performed on the initial set of point pairs corresponding to the line laser profilometer to obtain the real-time set of point pairs corresponding to the line laser profilometer. The first type of defect detection results are obtained based on the real-time point pair set corresponding to each of the line laser profilometers. The first type of defect detection results includes line mark detection results and warpage detection results, including: The real-time contour data corresponding to the line laser profilometer is obtained from the set of real-time point pairs corresponding to the line laser profilometer. When the defect detection result of the first type corresponds to the line mark detection result, the jitter-suppressed contour data corresponding to the line laser profilometer is obtained based on the real-time contour data corresponding to the line laser profilometer. Specifically, jitter suppression processing is performed on the real-time contour data corresponding to the line laser profilometer to obtain the jitter-suppressed contour data corresponding to the line laser profilometer. The jitter suppression processing includes spatial domain low-pass filtering or frequency domain filtering to remove spike jitter in the real-time contour data corresponding to the line laser profilometer. Based on the jitter suppression profile data corresponding to the line laser profilometer, the stable profile data corresponding to the line laser profilometer is obtained; based on the stable profile data corresponding to the line laser profilometer, the line mark value corresponding to the line laser profilometer is obtained; based on the line mark values ​​corresponding to all line laser profilometers, the defect detection result of the first type is obtained; And / or, a second type of defect detection result is obtained based on the real-time point pair set corresponding to each of the line laser profilometers and the standard profile data of the three-dimensional vision sensor corresponding to each of the line laser profilometers, wherein the second type of defect detection result includes thickness detection result.

2. The method for defect detection of a photovoltaic cell as claimed in claim 1, wherein, The standard contour data for each of the three-dimensional vision sensors is obtained by calibrating the three-dimensional vision sensors corresponding to multiple line laser profilometers in each group of line laser profilometers using standard blocks.

3. The defect detection method for photovoltaic cells as described in claim 1, characterized in that, The step of calculating the first correction slope corresponding to the line laser profilometer based on the initial set of point pairs includes: The first correction slope is directly calculated based on the initial set of point pairs corresponding to the line laser profilometer. Alternatively, the first correction slope can be obtained iteratively based on the initial set of point pairs corresponding to the line laser profilometer.

4. The defect detection method for photovoltaic cells as described in claim 3, characterized in that, In the process of obtaining the first correction slope through iteration based on the initial set of point pairs corresponding to the line laser profilometer, for any iteration: The basic correction slope is obtained based on the fitted line in the current iteration round; where, if the current iteration round is the first iteration, the fitted line in the current iteration round is obtained by fitting a straight line to the initial set of point pairs corresponding to the laser profilometer; if the current iteration round is not the first iteration, the fitted line in the current iteration round is the fitted line updated in the previous iteration round. The reference weight of each point pair is determined based on the distance between each point pair in the initial set of point pairs and the fitted line; The basic correction slope is updated based on each point pair and its reference weight in the initial set of point pairs to obtain the updated basic correction slope in the current iteration. If the current iteration is not the preset iteration, the fitted line is updated according to the updated basic correction slope to obtain the updated fitted line in the current iteration, and then the next iteration is entered. If the current iteration is the preset iteration, then the iteration ends, and the updated basic correction slope in the current iteration is used as the first correction slope.

5. The defect detection method for photovoltaic cells as described in claim 1, characterized in that, The step of performing contour correction on the initial set of point pairs corresponding to the line laser profilometer based on the first correction slope to obtain the real-time set of point pairs corresponding to the line laser profilometer includes: The calibration center is determined based on the initial set of point pairs corresponding to the line laser profilometer. The correction angle is determined based on the first correction slope; Based on the correction center and the correction angle, each point pair in the initial set of point pairs is corrected to obtain the contour correction result of each point pair; The real-time point pair set corresponding to the line laser profilometer is obtained based on the contour correction result of each point pair.

6. The defect detection method for photovoltaic cells as described in claim 1, characterized in that, The step of obtaining the second type of defect detection result based on the real-time point pair set corresponding to each of the line laser profilometers and the standard contour data of the three-dimensional vision sensor corresponding to each of the line laser profilometers includes: Based on the standard contour data of the 3D vision sensor corresponding to the line laser profilometer, the set of standard point pairs corresponding to the line laser profilometer is obtained; based on the set of standard point pairs corresponding to the line laser profilometer, the second correction slope corresponding to the line laser profilometer is calculated; based on the second correction slope corresponding to the line laser profilometer, the standard point pair set corresponding to the line laser profilometer is contour corrected to obtain the reference point pair set corresponding to the line laser profilometer. The reference line corresponding to the line laser profilometer is obtained from the set of reference point pairs corresponding to the line laser profilometer. The thickness component of the line laser profiler is obtained based on the real-time point pair set corresponding to the line laser profiler and the reference line corresponding to the line laser profiler. The thickness value of the line laser profilometer in each group is obtained by the thickness component of all line laser profilometers in each group. The second type of defect detection result is obtained based on the thickness value corresponding to each group of line laser profilometers.

7. The defect detection method for photovoltaic cells as described in claim 6, characterized in that, The step of obtaining the reference line corresponding to the line laser profilometer based on the set of reference point pairs includes: For each pair of points in the set of reference points, calculate the average value of the horizontal coordinate and the average value of the vertical coordinate to obtain the first mean and the second mean. For any pair of points in the set of reference points: the difference between the x-coordinate of the pair of points and the first mean is recorded as the first difference of the pair of points; the difference between the y-coordinate of the pair of points and the second mean is recorded as the second difference of the pair of points. A matrix is ​​constructed based on the first and second differences of each pair of points in the set of reference points; the matrix is ​​then decomposed into eigenvalues, and the reference line is obtained based on the eigenvector corresponding to the smallest eigenvalue.

8. The defect detection method for photovoltaic cells as described in claim 1, characterized in that, When performing jitter suppression processing through frequency domain filtering, Fourier transform is performed on the real-time contour data corresponding to the line laser profilometer to obtain the frequency domain signal; the frequency domain signal is multiplied by the sinc function, and the result of the multiplication is inversely transformed to obtain the jitter-suppressed contour data corresponding to the line laser profilometer.

9. The defect detection method for photovoltaic cells as described in claim 1, characterized in that, The step of obtaining stable profile data corresponding to the line laser profiler based on jitter-suppressed profile data includes: Discrete point cleaning processing is performed on the jitter suppression contour data corresponding to the line laser profilometer to obtain stable contour data corresponding to the line laser profilometer; the discrete point cleaning processing includes median filtering and mean filtering to remove discrete points in the jitter suppression contour data corresponding to the line laser profilometer.

10. The defect detection method for photovoltaic cells as described in claim 9, characterized in that, Before performing discrete point cleaning processing on the jitter-suppressed contour data corresponding to the line laser profilometer, the following steps are also included: The jitter level is calculated based on the jitter suppression profile data corresponding to the line laser profilometer. When the jitter level is greater than or equal to a preset threshold, spatial filtering is performed on the jitter suppression profile data corresponding to the line laser profilometer to obtain stable profile data. When the jitter level is less than the preset threshold, the jitter suppression profile data corresponding to the line laser profilometer is used as stable profile data.

11. The defect detection method for photovoltaic cells as described in claim 1, characterized in that, The step of obtaining the line mark value corresponding to the line laser profilometer based on the stable profile data corresponding to the line laser profilometer includes: Obtain the peak and valley values ​​of the stable contour data; calculate the difference between each peak value and its adjacent valley value, and take the maximum value of the difference as the line mark value corresponding to the peak value; take the largest line mark value among all the line mark values ​​corresponding to the peak values ​​as the line mark value corresponding to the laser profilometer.

12. The defect detection method for photovoltaic cells as described in claim 1, characterized in that, Also includes: When the defect detection result of the first type is the warpage detection result, real-time contour data corresponding to each of the line laser profilometers is obtained, and each contour point in the real-time contour data corresponds to a height value. The maximum and minimum height values ​​are obtained based on the real-time contour data corresponding to the line laser profilometer. The warpage corresponding to the line laser profilometer is obtained based on the maximum height value and the minimum height value. The defect detection results for the first type are obtained based on the warpage corresponding to all line laser profilometers.

13. The defect detection method for photovoltaic cells as described in claim 12, characterized in that, The step of obtaining the warpage corresponding to the line laser profilometer based on the maximum height value and the minimum height value includes: The warpage of the line laser profilometer is calculated based on the sign of the maximum and minimum height values. If the maximum and minimum height values ​​have the same sign, the larger absolute value of the absolute values ​​of the maximum and minimum height values ​​is taken as the warpage of the line laser profilometer. If the maximum and minimum height values ​​have opposite signs, the absolute value of the difference between the maximum and minimum height values ​​is taken as the warpage of the line laser profilometer.

14. A defect detection system for photovoltaic cells, characterized in that, include: Multiple sets of line laser profilometers, each set including multiple line laser profilometers; each line laser profilometer is used to acquire initial profile data of the photovoltaic cell under test; each line laser profilometer includes a three-dimensional vision sensor; And a processor, used for: For any of the line laser profilometers: based on the initial profile data of the photovoltaic cell under test collected by the line laser profilometer, an initial set of point pairs corresponding to the line laser profilometer is obtained; based on the initial set of point pairs corresponding to the line laser profilometer, a first correction slope corresponding to the line laser profilometer is calculated; based on the first correction slope corresponding to the line laser profilometer, profile correction is performed on the initial set of point pairs corresponding to the line laser profilometer to obtain the real-time set of point pairs corresponding to the line laser profilometer. The first type of defect detection result is obtained based on the real-time point pair set corresponding to each of the line laser profilometers. The first type of defect detection result includes line mark detection result and warpage detection result, including: obtaining the real-time profile data corresponding to the line laser profilometer based on the real-time point pair set corresponding to the line laser profilometer. When the defect detection result of the first type corresponds to the line mark detection result, the jitter-suppressed contour data corresponding to the line laser profilometer is obtained based on the real-time contour data corresponding to the line laser profilometer. Specifically, jitter suppression processing is performed on the real-time contour data corresponding to the line laser profilometer to obtain the jitter-suppressed contour data corresponding to the line laser profilometer. The jitter suppression processing includes spatial domain low-pass filtering or frequency domain filtering to remove spike jitter in the real-time contour data corresponding to the line laser profilometer. Based on the jitter-suppressed contour data corresponding to the line laser profilometer, stable contour data corresponding to the line laser profilometer is obtained; based on the stable contour data corresponding to the line laser profilometer, line mark value corresponding to the line laser profilometer is obtained; based on the line mark values ​​corresponding to all line laser profilometers, the first type of defect detection result is obtained; and / or, based on the real-time point pair set corresponding to each line laser profilometer and the standard contour data of the three-dimensional vision sensor corresponding to each line laser profilometer, the second type of defect detection result includes thickness detection result.

15. A computer-readable storage medium, characterized in that, The medium stores a computer program that can be executed by a processor to implement the defect detection method for photovoltaic cells as described in any one of claims 1-13.

Images