Solar cell panel invisible defect detection device based on compressed sensing spectral imaging

Abstract

The invention provides a solar cell panel invisible defect detection device based on compressed sensing spectral imaging, and the device comprises a photoexcitation light source which is used for applying near-infrared light excitation to the surface of a to-be-detected photovoltaic cell panel so as to induce the photovoltaic cell panel to generate a fluorescence signal, thereby enhancing a characteristic light signal of an invisible defect area and inhibiting the interference of ambient light; the imaging system is used for receiving a fluorescence signal emitted by the photovoltaic cell panel to be detected; and the image acquisition and processing system is used for receiving, storing and processing the acquired fluorescence image in real time. Interference of outdoor solar spectral distribution energy on a photoluminescence imaging or electroluminescence imaging system is avoided or inhibited, so that the detection level and detection efficiency of invisible defects of the solar cell panel are improved, meanwhile, the complexity and data acquisition amount of a spectral imaging device can be remarkably reduced by reducing the number of acquired wavelengths, and the detection efficiency of the solar cell panel is improved. The scanning time is reduced, so that the scanning efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of solar panel applications, and particularly relates to a solar panel invisible defect detection device based on compressed sensing spectral imaging. Background Art

[0002] Compressed sensing spectral imaging features snapshot spectral imaging, and thus is particularly suitable for building an online imaging detection device for invisible defects of solar panels, including airborne spectral imaging, handheld spectral imaging, etc. Thanks to the characteristics of multi-channel spectral imaging, it can avoid or suppress the interference of the spectral distribution energy of the outdoor sun on the photoluminescence imaging or electroluminescence imaging system as much as possible.

[0003] Silicon crystal solar panel components are key devices in solar power generation systems. Due to product manufacturing, transportation, installation, and operation reasons, various invisible defects that are difficult to identify by the human eye often exist on the surface of silicon crystal solar panels, including hidden cracks, fragments, broken grids, surface contamination, etc. These defects greatly affect the photoelectric conversion efficiency, service life, etc. of photovoltaic panel components, and even pose potential safety hazards to the operation of the entire system. Therefore, it is necessary to detect these invisible defects in the manufacturing and use links of solar panel components.

[0004] Currently, the online detection technologies for invisible defects of photovoltaic panel components mainly include infrared thermal imaging, electroluminescence imaging, photoluminescence imaging, volt-ampere (I-V) curve characteristic detection and analysis, etc. These technologies complement each other and can play roles in different application scenarios. However, the existing electroluminescence imaging and photoluminescence imaging need to operate in a relatively dark environment to avoid the interference of high-irradiance ambient light on the imaging system.

[0005] As an emerging spectral imaging technology, compressed sensing spectral imaging can reconstruct complete spectral data on the premise of reducing the sampling amount, and features snapshot spectral imaging. Therefore, this technology is particularly suitable for online detection of invisible defects of solar panels, including airborne spectral imaging and handheld spectral imaging in complex scenarios. After combining with compressed sensing technology, by reducing the data acquisition amount and data transmission amount, the real-time performance and efficiency of the spectral imaging device can be significantly improved.

[0006] Among the existing invention patents, Chinese invention patent CN118199517A, "Photoluminescence Imaging System and Method for Photovoltaic Panels Based on Linear Array InGaAs Camera (9)", CN201340393Y, "On-line Detection Device for Defects of Reflective Solar Cell Modules", world invention patent WO2011152445A1, "Electroluminescence Detection Device and Electroluminescence Detection Method for Solar Panels", WO2017172611A1, "Solar Panel Automatic Identification and Defect Detection System and Method Using Infrared Imaging", etc., all do not involve spectral imaging problems. Documents such as "Research on Photovoltaic Panel Defect Detection Technology Based on High Frame Rate InGaAs Camera (9)", "Research on Photovoltaic Panel Defect Detection Technology Based on InGaAs Camera (9) under High Illuminance", "Application of Photoluminescence Technology in Defect Detection of Si-based Solar Cells", "Photovoltaic Module Defect Detection System Based on Electroluminescence Imaging", etc., also do not involve compressive sensing spectral imaging problems. Summary of the Invention

[0007] The object of the present invention is to provide a solar panel hidden defect detection device based on compressive sensing spectral imaging, aiming to avoid or suppress the interference of the spectral distribution energy of the outdoor sun on the photoluminescence imaging or electroluminescence imaging system, thereby improving the detection level and detection efficiency of solar panel hidden defects. At the same time, by reducing the number of acquisition wavelengths, the complexity of the spectral imaging device and the data acquisition volume can be significantly reduced, and the scanning time can be reduced, so as to improve the scanning efficiency.

[0008] To achieve the above object, the present invention provides the following solution:

[0009] A solar panel hidden defect detection device based on compressive sensing spectral imaging, comprising a photoexcitation light source, a photovoltaic panel to be measured, an imaging system, and an image acquisition and processing system;

[0010] The photoexcitation light source is used to apply near-infrared light excitation to the surface of the photovoltaic panel to be measured, so as to induce the photovoltaic panel to generate a fluorescence signal, thereby enhancing the characteristic light signal in the hidden defect area and suppressing the interference of ambient light;

[0011] The imaging system is used to receive the fluorescence signal emitted by the photovoltaic panel to be measured;

[0012] The image acquisition and processing system is used to receive, store and process the collected fluorescence images in real time.

[0013] The photo-excitation light source is used to apply near-infrared light excitation to the surface of the photovoltaic panel to be measured, so as to induce the photovoltaic panel to generate a fluorescence signal, thereby enhancing the characteristic light signal in the invisible defect area and suppressing the interference of ambient light. The photo-excitation light source is a narrow-band near-infrared surface-emitting light source, and its wavelength design matches the material characteristics of the photovoltaic panel, and can efficiently excite the fluorescence signal of the photovoltaic panel. The spectral energy of the photoluminescence is mainly concentrated in the short-wave infrared range with a wavelength of 1050nm to 1250nm. The narrow-band light source reduces the interference of redundant spectral components and improves the directivity and spectral purity of the excitation light.

[0014] The photovoltaic panel to be measured is a monocrystalline silicon photovoltaic panel or a polycrystalline silicon photovoltaic panel; under the irradiation of the near-infrared excitation light source, it generates photoluminescence, and the spectral energy of the photoluminescence is mainly concentrated in the short-wave infrared range with a wavelength of 1050nm to 1250nm.

[0015] The imaging system includes an objective lens, a coding element, a relay lens, a dispersion element, an imaging lens, and a camera. The imaging system is used to receive the fluorescence signal emitted by the photovoltaic panel and image it on the coding element through the objective lens; the coding element can be implemented using a spatial light modulator or a quartz chrome mask plate; the image formed on the coding element reaches the post-optical path through spatial coding; after passing through the relay lens, parallel light enters the dispersion element; the dispersion element can be a dispersion prism or a diffraction grating, so that the parallel light of different wavelengths entering the dispersion element is separated in the spatial direction according to a certain rule and propagates at different angles; finally, the parallel light propagating at different angles is imaged again through the imaging lens and received by the camera. The camera is a near-infrared or short-wave infrared camera, including a short-wave infrared InGaAs camera and other types of short-wave infrared cameras, which have high spectral sensitivity in the wavelength range of 1050nm - 1250nm and are dedicated to capturing the fluorescence image emitted by the photovoltaic panel; the spectral response range of the camera matches the excitation wavelength of the photo-excitation light source and the fluorescence emission wavelength of the photovoltaic panel to ensure high-sensitivity image acquisition effect.

[0016] The image acquisition and processing system is connected to the camera and is used to receive, store, and process the collected fluorescence images in real time. The system uses a convex optimization algorithm based on the theory of compressive sensing to further enhance the characteristics of the invisible defect area of the photovoltaic panel in the image, and removes the residual ambient light interference through a denoising algorithm to achieve high-precision defect recognition and positioning.

[0017] The system uses a convex optimization algorithm based on the theory of compressive sensing to efficiently process the collected spectral data, including the following steps:

[0018] Step 1, Sparse Representation: Model the fluorescence signal of the photovoltaic panel as a sparse signal. Assume that it is sparse under a certain basis (such as wavelet basis, discrete cosine transform basis), that is, most of the coefficients are zero or close to zero. Select a suitable sparse transform basis or dictionary Ψ, and represent the fluorescence signal x as:

[0019] x = Ψs

[0020] where s is the sparse representation of the signal under the transform basis.

[0021] Step 2, Compressive Sampling: Use a random measurement matrix Φ to perform non-uniform sampling on the sparse signal to obtain the sampled data y; the sampling process is:

[0022] y = Φx = ΦΨs

[0023] where y is the low-dimensional measurement data, and Φ is a measurement matrix that satisfies the RIP (Restricted Isometry Property). Gaussian random matrix, Bernoulli matrix or partial Fourier matrix, etc. can be selected.

[0024] Step 3, Signal Reconstruction: Use a convex optimization method to reconstruct the original signal x from the compressive sampling data y. The goal of signal reconstruction is to solve the following optimization problem:

[0025] min||s|| 1 ||y - ΦΨs|| 2 ≤ε

[0026] where ||s|| 1 is the L1 norm of the sparse representation coefficients, and ε is the allowable error range.

[0027] Common convex optimization algorithms include: Basis Pursuit (BP), which solves the L1 norm minimization problem through standard linear programming methods. Lasso (Least Absolute Shrinkage and Selection Operator), which introduces a regularization parameter to balance sparsity and fitting error. Gradient projection algorithm, which iteratively optimizes the sparse signal while satisfying the constraint conditions. Here, the Basis Pursuit (BP) algorithm is used for signal reconstruction.

[0028] Step 4, Transform the optimization problem into a standard linear programming form:

[0029]

[0030] Through this transformation, the problem can be efficiently solved using existing linear programming solvers (such as interior point method or simplex method), and the reconstructed signal x is returned after solving.

[0031] Step 5, Feature Enhancement and Defect Location: The reconstructed spectral signal is processed by a polarization imaging analysis algorithm to further enhance the characteristic signal of the invisible defect area of the photovoltaic panel; polarization analysis can suppress background noise and highlight the polarization characteristics of the fluorescence signal. Apply the differential method or edge detection algorithm to the enhanced image to accurately identify the boundary position of the invisible defect and achieve precise defect location.

[0032] Step 6, Denoising Processing: Apply a denoising algorithm (such as wavelet denoising, non-local means denoising) to the reconstructed image to eliminate the interference of residual ambient light and improve the signal quality and contrast of the image. Brief Description of the Drawings

[0033] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the embodiments.

[0034] Figure 1 It is a schematic diagram of the composition principle of the invisible defect detection device for solar panels based on compressive sensing spectral imaging of the present invention;

[0035] Figure 2 It is a schematic diagram of the principle of the invisible defect detection device for solar panels based on diffraction grating and compressive sensing spectral imaging provided by the embodiment of the present invention;

[0036] Figure 3 It is a schematic diagram of the principle of the invisible defect detection device for solar panels based on DMD and compressive sensing spectral imaging provided by the embodiment of the present invention. Detailed Embodiment

[0037] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0038] To make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below with reference to the drawings and specific embodiments.

[0039] Embodiment 1

[0040] As Figure 1As shown in the figure, the present invention provides a solar panel invisible defect detection device based on compressed sensing spectral imaging, which includes a photoinduced excitation light source 1, a photovoltaic panel to be measured 2, an imaging system, an image acquisition and processing system 3, etc. The photoinduced excitation light source 1 is a narrow-band near-infrared surface-emitting light source, and its wavelength design matches the material characteristics of the photovoltaic panel, and is used to apply near-infrared light excitation to the surface of the photovoltaic panel to be measured 2 to induce the photovoltaic panel to generate a fluorescence signal, thereby enhancing the characteristic light signal in the invisible defect area and suppressing the interference of ambient light. The spectral energy of this photoluminescence is mainly concentrated in the short-wave infrared range with a wavelength of 1050nm to 1250nm. The narrow-band light source reduces the interference of redundant spectral components and improves the directivity and spectral purity of the excitation light.

[0041] The imaging system includes an objective lens 4, a coding element 5, a relay lens 6, a dispersion element 7, an imaging lens 8, and a camera 9. The coding element 5 is a quartz mask coding element. The dispersion element 7 is a dispersion prism.

[0042] The imaging system is used to receive the fluorescence signal emitted by the photovoltaic panel, image it on the coding element 5 through the objective lens 4; the image formed on the coding element 5 reaches the post-optical path through spatial coding; the parallel light is formed through the relay lens 6 and enters the dispersion prism; the parallel light of different wavelengths entering the dispersion prism, that is, the dispersion element 7, is separated in the spatial direction according to a certain rule and propagates along different angles; finally, the parallel light propagating along different angles is imaged again through the imaging lens 8 and received by the camera 9.

[0043] Among them, the camera 9 is a near-infrared or short-wave infrared camera, which is dedicated to capturing the fluorescence image emitted by the photovoltaic panel. The spectral response range of the camera 9 matches the excitation wavelength of the photoinduced excitation light source 1 and the fluorescence emission wavelength of the photovoltaic panel to ensure a high-sensitivity image acquisition effect. The image acquisition and processing system 3 is connected to the camera 9 and is used to receive, store, and process the captured fluorescence image in real time. The system uses a convex optimization algorithm based on the compressed sensing theory to further enhance the characteristics of the invisible defect area of the photovoltaic panel in the image, and removes the residual ambient light interference through a denoising algorithm to achieve high-precision defect recognition and positioning.

[0044] Embodiment 2

[0045] As Figure 2As shown in the figure, the present invention provides a solar panel invisible defect detection device based on a diffraction grating and compressive sensing spectral imaging. Different from Embodiment 1, the device uses a diffraction grating as the dispersion element 7, and the photoinduced excitation light source 1 uses a narrowband LED light source with a central wavelength of 850 nm or 808 nm. This light source has the characteristics of a narrow bandwidth (the typical bandwidth is less than 30 nm) and high spectral purity, and can effectively excite the invisible defect area on the surface of the photovoltaic panel. To ensure the reliability of the detection results, the light source stably outputs the required light intensity through a high-precision current control circuit and avoids the interference of ambient light.

[0046] The imaging system includes an objective lens 4, a coding element 5, a relay lens 6, a diffraction grating, an imaging lens 8, and a camera 9. The film coding element 5 is a quartz diaphragm coding element.

[0047] The imaging system is used to receive the fluorescence signal emitted by the photovoltaic panel, and the image is formed on the coding element 5 through the objective lens 4; the image formed on the coding element 5 undergoes spatial coding and reaches the post-optical path; after passing through the relay lens 6, parallel light enters the diffraction grating; the parallel light of different wavelengths entering the diffraction grating is separated in the spatial direction according to a certain rule and propagates at different angles; finally, the parallel light propagating at different angles is imaged again through the imaging lens 8 and received by the camera 9. The spectral response range of the camera 9 matches the excitation wavelength of the photoinduced excitation light source 1 and the fluorescence emission wavelength of the photovoltaic panel to ensure a high-sensitivity image acquisition effect. The image acquisition and processing system 3 is connected to the camera 9 and is used to receive, store, and process the collected fluorescence images in real time. The system uses a convex optimization algorithm based on the compressive sensing theory to further enhance the characteristics of the invisible defect area of the photovoltaic panel in the image and removes the residual ambient light interference through a denoising algorithm to achieve high-precision defect identification and positioning.

[0048] Embodiment 3

[0049] As Figure 3 shown in the figure, the present invention provides a solar panel invisible defect detection device based on DMD and compressive sensing spectral imaging. Different from Embodiment 2, the device uses a DMD digital micromirror as the coding element 5. The DMD has the advantages of high resolution, high contrast, high brightness, fast response time, and high reliability.

[0050] The imaging system is used to receive the fluorescence signal emitted by the photovoltaic panel, and after being imaged on the DMD by the objective lens 4, it reaches the rear optical path after being reflected by the DMD; after passing through the relay lens 6, parallel light is formed and enters the diffraction grating; the parallel light entering the diffraction grating is separated in the spatial direction according to certain rules for different wavelengths of light and propagates at different angles; finally, the parallel light propagating at different angles is imaged again by the imaging lens 8 and received by the camera 9. The spectral response range of the camera 9 matches the excitation wavelength of the photoinduced excitation light source 1 and the fluorescence emission wavelength of the photovoltaic panel to ensure a high-sensitivity image acquisition effect. The image acquisition and processing system 3 is connected to the camera 9 and is used to receive, store, and process the acquired fluorescence images in real time. The system uses a convex optimization algorithm based on the compressive sensing theory to further enhance the characteristics of the invisible defect area of the photovoltaic panel in the image, and removes the residual ambient light interference through a denoising algorithm to achieve high-precision defect identification and positioning.

Claims

1. Photovoltaic panel invisible defect detection device based on compressed sensing spectral imaging, characterized in that: include: Photoexcitation light source (1), photovoltaic panel to be tested (2), imaging system and image acquisition and processing system (3); The photoexcitation light source (1) is used to apply near-infrared light excitation to the surface of the photovoltaic cell panel (2) to be tested, so as to induce the photovoltaic cell panel to generate a fluorescent signal, thereby enhancing the characteristic light signal of the invisible defect area and suppressing the interference of ambient light; The imaging system is used to receive the fluorescent signal emitted by the photovoltaic panel (2) to be tested; The image acquisition and processing system (3) is used to receive, store and process the acquired fluorescence images in real time.

2. The photovoltaic panel invisible defect detection device based on compressed sensing spectral imaging according to claim 1 is characterized in that: The photovoltaic cell panel (2) to be tested is a single crystal silicon photovoltaic cell panel or a polycrystalline silicon photovoltaic cell panel; it generates photoluminescence under the irradiation of a near-infrared excitation light source.

3. The photovoltaic panel invisible defect detection device based on compressed sensing spectral imaging according to claim 1 is characterized in that: The photoexcitation light source (1) is a narrow-band light source in the near-infrared band.

4. The photovoltaic panel invisible defect detection device based on compressed sensing spectral imaging according to claim 1 is characterized in that: The imaging system comprises an objective lens (4), a coding element (5), a relay lens (6), a dispersion element (7), an imaging lens (8), and a camera (9) in sequence; the fluorescent signal emitted by the photovoltaic panel (2) to be tested is imaged on the coding element (5) via the objective lens (4); the image imaged on the coding element (5) reaches the rear optical path after spatial encoding; parallel light is formed through the relay lens (6) and enters the dispersion element (7); the parallel light of different wavelengths entering the dispersion element (7) is separated in a spatial direction according to a certain rule and propagates along different angles; finally, the equal light propagated along different angles is imaged again through the imaging lens (8) and received by the camera (9).

5. The photovoltaic panel invisible defect detection device based on compressed sensing spectral imaging according to claim 4 is characterized in that: The coding element (5) is realized by using a spatial light modulator or a quartz mask; The dispersion element (7) is a dispersion prism or a diffraction grating; The camera (9) is a near infrared or short wave infrared camera.

6. The photovoltaic panel invisible defect detection device based on compressed sensing spectral imaging according to claim 4 is characterized in that: The image acquisition and processing system (3) is connected to the camera (9) and is used to receive, store and process the collected fluorescent images in real time; the image acquisition and processing system (3) uses a convex optimization algorithm based on compressed sensing theory to further enhance the invisible defect area features of the photovoltaic panel in the image, and removes residual ambient light interference through a denoising algorithm to achieve high-precision defect recognition and positioning; The convex optimization algorithm based on compressed sensing theory uses compressed sensing theory to process the collected fluorescence spectrum data, and includes the following steps: Step 1: Sparse representation: Model the fluorescence signal of the photovoltaic panel as a sparse signal, make it sparse under a certain basis, select a suitable sparse transformation basis or dictionary Ψ, and represent the fluorescence signal x as: x=Ψs Among them, s is the sparse representation of the signal under the transformation basis; Step 2, compressed sampling: Use the random measurement matrix Φ to perform non-uniform sampling on the sparse signal to obtain the sampled data y; the sampling process is: y=Φx=ΦΨs Among them, y is the low-dimensional measurement data, Φ is the measurement matrix that satisfies the RIP restricted isometry property; Step 3: Signal reconstruction: Reconstruct the original signal x from the compressed sampled data y using a convex optimization method; the goal of signal reconstruction is to solve the following optimization problem: min||s||1 ||y-ΦΨs||2≤ε Among them, ||s||1 is the L1 norm of the sparse representation coefficient, and ε is the allowable error range; Step 4: Convert the optimization problem into a standard linear programming form: Through this transformation, the problem is efficiently solved using an existing linear programming solver, and the reconstructed signal x is returned after the solution; Step 5: Feature enhancement and defect location: The reconstructed spectral signal is processed by the polarization imaging analysis algorithm to enhance the characteristic signal of the invisible defect area of ​​the photovoltaic panel; polarization analysis can suppress background noise and highlight the polarization characteristics of the fluorescence signal; the difference method or edge detection algorithm is applied to the enhanced image to accurately identify the boundary position of the invisible defect and achieve precise location of the defect; Step 6: De-noising: Apply a denoising algorithm to the reconstructed image to eliminate the interference of residual ambient light and improve the signal quality and contrast of the image.

Images