Method, system and device for detecting dirt layer on surface of photovoltaic module based on spectral unmixing

Abstract

The invention discloses a photovoltaic module surface pollution layer detection method, system and device based on spectral unmixing, and the method comprises the steps: obtaining end member spectrums of existing and potential surface pollution layers and corresponding abundance matrixes through employing the spatial-spectral information provided by a hyperspectral image and employing the fixed initialization spectral unmixing and robust spectral unmixing of part of end members; the spatial distribution condition of the surface dirt layer is accurately extracted, and the state of the surface dirt layer of the photovoltaic module is effectively detected. According to the invention, power generation efficiency reduction and potential hot spot risk caused by pollution layer change can be continuously monitored, tracked and warned, and a quantitative basis is provided for fine decision and optimization of a photovoltaic module cleaning strategy.

Description

Technical Field

[0001] The present invention relates to the technical field of optical image processing, and more particularly to a method, system and device for detecting contamination on the surface of a photovoltaic module based on spectral unmixing. Background Art

[0002] The rapid expansion of photovoltaic power generation is posing a severe challenge to the safe operation and maintenance of power plants, as well as optimizing their operating efficiency. Among these challenges, the accumulation of dirt on the surfaces of photovoltaic modules has become a key factor limiting system stability and economic viability. During operation, dirt, sand, bird droppings, fallen leaves, and other contaminants gradually accumulate on the module surfaces, forming a shading layer. This layer not only impairs the module's light transmission and reduces solar radiation absorption, thereby affecting photovoltaic conversion efficiency and power output, but also causes localized shading (such as spot or strip coverage) due to uneven distribution, triggering hot spot effects and posing a significant threat to the safety and long-term durability of power plants.

[0003] Currently, monitoring and maintenance of photovoltaic module surface contamination still relies primarily on manual judgment and qualitative observation, making it difficult to achieve intelligent, high-precision, and real-time operation and maintenance. While cleaning technologies (such as pipe flushing, robotic cleaning, or vehicle-based cleaning) are already in use, data support for cleaning schedules, route planning, and resource management is still lacking. Adopting a fixed-cycle cleaning strategy alone not only makes it difficult to balance power generation efficiency and operation and maintenance costs, but also results in a waste of manpower and resources.

[0004] In recent years, some photovoltaic power plants have begun implementing drone-based intelligent inspection technology. These methods use visible light and infrared sensors to capture module images and leverage computer vision technology to detect and analyze the status of contamination. While these methods can detect hot spots or localized obscuration, they lack the ability to analyze material composition and accurately quantify the composition and coverage of contamination. Given the diverse composition of contamination and complex operating environments, their detection performance falls short of meeting the needs of photovoltaic power plants for intelligent monitoring and maintenance of contamination on PV module surfaces. The extremely high spectral resolution of hyperspectral imagery enables it to diagnose fine spectral features and identify and analyze feature types, materials, and composition. UAVs and other aerial platforms equipped with hyperspectral imagers for inspection imaging and real-time analysis provide a new and effective means for intelligent monitoring of contamination on PV module surfaces. In actual operating environments, contamination on PV module surfaces is formed by the adhesion of various liquid and solid particulate contaminants to the module surface. Hyperspectral images captured are a mixture of the module material spectrum and the spectra of various contamination components. Spectral unmixing technology aims to decipher the spectra and their proportions from mixed pixels, providing a new approach for analyzing the status of contamination on PV module surfaces. However, the existing spectral unmixing technology faces problems such as spectral variability, severe noise interference, and poor robustness when applied to hyperspectral images of photovoltaic modules. This limits the accuracy of quantitatively solving the spectrum of surface pollution components and their proportions, and the application effect is poor.

[0005] Therefore, it is necessary to propose a new solution to solve the above problems. Summary of the Invention

[0006] The purpose of the present invention is to overcome the above-mentioned deficiencies of the prior art and to provide a method, system and device for detecting contamination on the surface of photovoltaic modules based on spectral unmixing.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] A method for detecting contamination on photovoltaic modules based on spectral unmixing, the method comprising:

[0009] Step S1, selecting a single photovoltaic module for cleaning and collecting a hyperspectral image of the cleaned photovoltaic module sample;

[0010] Step S2, selecting a single photovoltaic module with a surface contamination layer and collecting a hyperspectral image of the photovoltaic module sample with the surface contamination layer;

[0011] Step S3, using a drone equipped with a hyperspectral imager to collect hyperspectral images of all photovoltaic modules to be inspected in the photovoltaic power station;

[0012] Step S4, preprocessing the collected hyperspectral image;

[0013] Step S5, segmenting and extracting the photovoltaic module area through the photovoltaic module segmentation network;

[0014] Step S6, extracting photovoltaic module endmember spectra from the hyperspectral image of the clean photovoltaic module sample;

[0015] Step S7, extracting the existing surface contamination layer endmember spectrum from the hyperspectral image of the surface contamination layer photovoltaic module sample;

[0016] Step S8, performing initial spectral unmixing with fixed endmembers on the hyperspectral images of all photovoltaic modules to be inspected in the photovoltaic power station to obtain initial endmember spectra and abundance matrices;

[0017] Step S9, performing robust spectral unmixing on the hyperspectral image of the photovoltaic module area with the surface contamination layer to obtain the final endmember spectrum and abundance matrix;

[0018] Step S10: determining a spatial distribution density map of the contamination layer on the surface of the photovoltaic module and a warning area.

[0019] Furthermore, in step S5, the photovoltaic component segmentation network adopts one of UNet and Mask R-CNN.

[0020] Furthermore, in step S6, the endmember spectrum extraction method adopts one of the clean pixel index, simplex growth, and vertex component analysis, and the extracted photovoltaic module endmember spectrum is , where L is the number of bands, is the number of PV module terminals, is the set of real numbers.

[0021] Furthermore, in step S7, the endmember spectrum extraction method adopts one of the clean pixel index, simplex growth, and vertex component analysis, and the extracted existing surface pollution layer endmember spectrum is , where L is the number of bands, is the number of end members of the pollution layer.

[0022] Furthermore, in step S8, the initialization spectrum unmixing with some end members fixed includes the following steps:

[0023] Step S81, set the number of abnormal surface pollution layer end members , the abnormal surface contamination layer end member is the end member of the potential contamination layer that is not included in the surface contamination layer photovoltaic module sample in step S7;

[0024] Step S82: Set the initial end member set , calculate the initial simplex center

[0025] , ;

[0026] Step S83, calculate the orthogonal projection matrix

[0027]

[0028] ,

[0029] Where U is Centralized calculation result; I is the identity matrix; superscript T is the matrix transpose operator; is the regularization parameter;

[0030] Step S84, for , iteratively extract the endmember spectrum of the abnormal surface pollution layer ;

[0031] Step S85, based on the photovoltaic module end member spectrum , surface pollution layer end-member spectrum and abnormal surface contamination endmember spectra To obtain the initial endmember spectrum results , and based on the initial endmember spectral results , the initial abundance matrix is ​​obtained by estimating the fully constrained least squares method ,in is the photovoltaic module end member spectrum The corresponding abundance matrix, The existing surface contamination end-member spectrum The corresponding abundance matrix, is the abnormal surface contamination endmember spectrum The corresponding abundance matrix.

[0032] Further, in step S84, the abnormal surface contamination layer end member spectrum The iterative extraction process is:

[0033] Step S841: for all pixels in the hyperspectral image , calculate the projection distance , and select the pixel with the largest distance As a new abnormal surface pollution layer end member :

[0034] ,

[0035]

[0036]

[0037] Step S842: Update the end member set , simplex center

[0038]

[0039]

[0040] Step S843: Update the orthogonal projection matrix

[0041]

[0042]

[0043]

[0044] Where v is the residual vector; w is the projection vector;

[0045] Step S844, for Repeat the iterative steps S841 to S843 to add all the new abnormal pollution layer end members Combined together to obtain the abnormal surface pollution layer end-member spectrum, combined photovoltaic module end-member spectrum , surface pollution layer end-member spectrum and abnormal surface contamination endmember spectra , and obtain the initial endmember spectrum results .

[0046] Furthermore, in step S9, the robust spectrum unmixing is calculated by repeatedly iteratively updating the initial endmember spectrum result through the following calculation formula: and the initial abundance matrix

[0047]

[0048]

[0049] Where, Represents matrix dot product; is the sparse regularization constraint parameter; 、 is a diagonal matrix, and the lth element on its diagonal is calculated as

[0050]

[0051] .

[0052] Furthermore, in step S10, determining the spatial distribution density map and warning area of ​​the contamination layer on the surface of the photovoltaic module includes the following steps:

[0053] Step S101, from the initial abundance matrix Decompose to obtain the abundance matrix corresponding to the end member of the surface pollution layer , ;

[0054] Step S102: For each pixel j, calculate the total abundance of the surface pollution layer

[0055] ,

[0056] Step S103: The total abundance of the surface pollution layer of each pixel Map pixel coordinates to a two-dimensional matrix ,in and are the height and width of the image respectively;

[0057] Step S104: When the total abundance of the surface pollution layer of the pixel Greater than the dynamic threshold When the pixel is divided into the warning area, the dynamic threshold The calculation method is

[0058]

[0059] Where, is the basic threshold; is the adjustment coefficient; PM is the air index of the day, and Historical air index mean and standard deviation.

[0060] The present invention also provides a photovoltaic module surface contamination detection system based on spectral unmixing, the system comprising:

[0061] A hyperspectral image acquisition unit, used for collecting data of the photovoltaic components to be inspected to obtain a hyperspectral image;

[0062] A preprocessing unit, used for preprocessing the hyperspectral image by performing reflectance correction and outlier removal;

[0063] Component segmentation unit, used to segment and extract the photovoltaic component area;

[0064] Endmember extraction unit, used to extract the endmember spectrum of photovoltaic modules and the endmember spectrum of existing surface pollution layers;

[0065] Spectral unmixing unit, used for initialization spectral unmixing and robust spectral unmixing with some endmembers fixed, to obtain endmember spectra and abundance matrices;

[0066] The surface contamination layer detection unit is used to determine the spatial distribution density map and warning area of ​​the contamination layer on the surface of the photovoltaic module.

[0067] The present invention also provides a photovoltaic module surface contamination detection device based on spectral unmixing, comprising:

[0068] Memory, used to store computer programs and data;

[0069] The processor is configured to implement the steps of the above-mentioned method for detecting contamination on the surface of photovoltaic modules based on spectral unmixing when executing the computer program.

[0070] The beneficial effects of the present invention are:

[0071] This invention uses spectral unmixing technology to quantitatively analyze the spatial distribution of contamination on photovoltaic module surfaces, enabling robust and accurate detection of contamination status. This technology can continuously monitor, track, and provide early warning of power generation efficiency decline and potential hot spot risks caused by changes in the contamination layer, providing a quantitative basis for refined decision-making and optimization of photovoltaic module cleaning strategies. This not only helps ensure the safe operation of power plants and conserve cleaning resources, but also has application value in improving the operational and maintenance efficiency of photovoltaic power plants. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 4 is a flow chart of a method for detecting contamination on the surface of a photovoltaic module based on spectral unmixing in this embodiment;

[0073] Figure 2 1 is a structural framework diagram of a photovoltaic module surface contamination detection system based on spectral unmixing in this embodiment;

[0074] Figure 3 This is a structural framework diagram of a photovoltaic module surface contamination detection device based on spectral unmixing in this embodiment.

[0075] Reference numerals: hyperspectral image acquisition unit 1 , preprocessing unit 2 , component segmentation unit 3 , endmember extraction unit 4 , spectrum unmixing unit 5 , surface contamination detection unit 6 , memory 7 , processor 8 . DETAILED DESCRIPTION

[0076] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0077] Example: Detection method of photovoltaic module surface contamination based on spectral unmixing, such as Figure 1 As shown, the method includes:

[0078] Step S1: Select a single photovoltaic module for cleaning and collect a hyperspectral image of the cleaned photovoltaic module sample.

[0079] Step S2: Select a single photovoltaic module with a surface contamination layer and collect a hyperspectral image of the photovoltaic module sample with the surface contamination layer.

[0080] Step S3: using a drone equipped with a hyperspectral imager to collect hyperspectral images of all photovoltaic modules to be inspected in the photovoltaic power station.

[0081] Step S4, preprocessing the collected hyperspectral image, including reflectance correction and outlier removal;

[0082] The reflectivity correction calculation method is:

[0083]

[0084] Where, For the wavelength Reflectivity at For the wavelength The radiance at (available from the sensor); is the dark current response obtained by measuring the black calibration plate; The radiance obtained by measuring the white reference plate; is the known reflectivity value of a white reference plate, usually 1 or a known high reflectivity value.

[0085] The outlier removal method is:

[0086]

[0087] Where, For the wavelength The reflectivity at .

[0088] Step S5: Segment and extract the photovoltaic component area through a photovoltaic component segmentation network; the photovoltaic component segmentation network adopts one of UNet and Mask R-CNN.

[0089] Step S6, extracting photovoltaic module endmember spectra from the hyperspectral image of the clean photovoltaic module sample;

[0090] Among them, the end-member spectrum extraction method adopts one of the following methods: pure pixel index, single body growth, and vertex component analysis. The extracted photovoltaic module end-member spectrum is , where L is the number of bands, is the number of PV module terminals, is the set of real numbers.

[0091] Step S7, extracting the existing surface contamination layer endmember spectrum from the hyperspectral image of the surface contamination layer photovoltaic module sample;

[0092] Among them, the endmember spectrum extraction method adopts one of the following methods: pure pixel index, simplex growth, and vertex component analysis. The extracted endmember spectrum of the existing surface pollution layer is , where L is the number of bands, is the number of end members of the pollution layer.

[0093] Step S8, performing initial spectral unmixing with fixed endmembers on the hyperspectral images of all photovoltaic modules to be inspected in the photovoltaic power station to obtain initial endmember spectra and abundance matrices;

[0094] Specifically, the following steps are included:

[0095] Step S81, set the number of abnormal surface pollution layer end members , the abnormal surface contamination layer end member is the end member of the potential contamination layer that is not included in the surface contamination layer photovoltaic module sample in step S7; Usually set to 1-3;

[0096] Step S82: Set the initial end member set , calculate the initial simplex center

[0097] , ;

[0098] Step S83, calculate the orthogonal projection matrix

[0099]

[0100] ,

[0101] Where U is Centralized calculation result; I is the identity matrix; superscript T is the matrix transpose operator; is the regularization parameter;

[0102] Step S84, for , iteratively extract the endmember spectrum of the abnormal surface pollution layer Specific:

[0103] Step S841: for all pixels in the hyperspectral image , calculate the projection distance , and select the pixel with the largest distance As a new abnormal surface pollution layer end member :

[0104] ,

[0105]

[0106]

[0107] Step S842: Update the end member set , simplex center

[0108]

[0109]

[0110] Step S843: Update the orthogonal projection matrix

[0111]

[0112]

[0113]

[0114] Where v is the residual vector; w is the projection vector;

[0115] Step S844, for Repeat the iterative steps S841 to S843 to add all the new abnormal pollution layer end members Combined together to obtain the abnormal surface pollution layer end-member spectrum, combined photovoltaic module end-member spectrum , surface pollution layer end-member spectrum and abnormal surface contamination endmember spectra , and obtain the initial endmember spectrum results ;

[0116] Step S85, in the initial end member spectrum result Based on the fully constrained least squares method, the initial abundance matrix is ​​obtained ,in is the photovoltaic module end member spectrum The corresponding abundance matrix, The existing surface contamination end-member spectrum The corresponding abundance matrix, is the abnormal surface contamination endmember spectrum The corresponding abundance matrix.

[0117] Step S9, performing robust spectral unmixing on the hyperspectral image of the photovoltaic module area with the surface contamination layer to obtain the final endmember spectrum and abundance matrix;

[0118] The robust spectrum unmixing is calculated by repeatedly iteratively updating the initial endmember spectrum results using the following calculation formula: and the initial abundance matrix

[0119]

[0120]

[0121] Where, Represents matrix dot product; is the sparse regularization constraint parameter; 、 is a diagonal matrix, and the lth element on its diagonal is calculated as

[0122]

[0123] .

[0124] Step S10, determining a spatial distribution density map of the contamination layer on the surface of the photovoltaic module and a warning area;

[0125] Specifically, the following steps are included:

[0126] Step S101, from the initial abundance matrix Decompose to obtain the abundance matrix corresponding to the end member of the surface pollution layer , ;

[0127] Step S102: For each pixel j, calculate the total abundance of the surface pollution layer

[0128] ,

[0129] Step S103: The total abundance of the surface pollution layer of each pixel Map pixel coordinates to a two-dimensional matrix ,in and are the height and width of the image respectively;

[0130] Step S104: When the total abundance of the surface pollution layer of the pixel Greater than the dynamic threshold When the pixel is divided into the warning area, the dynamic threshold The calculation method is

[0131]

[0132] Where, is the basic threshold, which can generally be set to [0.1, 0.3]; is the adjustment coefficient; PM is the air index of the day, which can generally be PM10 index. and Historical air index mean and standard deviation.

[0133] In the technical solution provided in this embodiment, first, a hyperspectral image of a clean photovoltaic module sample is acquired; then, a hyperspectral image of a photovoltaic module sample with a surface contamination layer is acquired; then, hyperspectral images of all photovoltaic modules to be inspected are acquired; the hyperspectral images are preprocessed, including reflectance correction and outlier removal; endmember spectra of photovoltaic modules are extracted from the hyperspectral images of the clean photovoltaic module samples; endmember spectra of existing surface contamination layers are extracted from the hyperspectral images of the photovoltaic module sample with a surface contamination layer; then, the hyperspectral images of all photovoltaic modules to be inspected in the photovoltaic power station are initialized with fixed endmembers to obtain initial endmember spectra and abundance matrices; then, robust spectral unmixing is performed on the hyperspectral images of the photovoltaic module area with a surface contamination layer to obtain final endmember spectra and abundance matrices; and the spatial distribution density map and warning area of ​​the surface contamination layer of the photovoltaic modules are determined.

[0134] In the technical solution provided in this embodiment, the spatial-spectral information provided by the hyperspectral image is utilized, and the initialization spectral unmixing with fixed partial end members and robust spectral unmixing are adopted to obtain the end member spectra and corresponding abundance matrices of existing and potential surface contamination layers, accurately extract the spatial distribution of the surface contamination layer, and effectively realize the detection of the surface contamination status of the photovoltaic module.

[0135] See Figure 2 , which is a structural schematic diagram of a photovoltaic module surface contamination detection system based on spectral unmixing provided in this embodiment, the system includes a hyperspectral image acquisition unit 1, a preprocessing unit 2, a module segmentation end member, an end member extraction unit 4, a spectral unmixing unit 5, and a surface contamination detection unit 6.

[0136] The hyperspectral image acquisition unit 1 is used to acquire data of the photovoltaic module to be inspected and obtain a hyperspectral image;

[0137] Preprocessing unit 2, used for preprocessing the hyperspectral image by correcting the reflectivity and removing abnormal values;

[0138] Component segmentation unit 3, used for segmenting and extracting the photovoltaic component area;

[0139] Endmember extraction unit 4, used to extract the endmember spectrum of the photovoltaic module and the endmember spectrum of the existing surface pollution layer;

[0140] Spectral unmixing unit 5, used for initialization spectral unmixing and robust spectral unmixing with some endmembers fixed, to obtain endmember spectra and abundance matrices;

[0141] The surface contamination layer detection unit 6 is used to determine the spatial distribution density map and warning area of ​​the contamination layer on the surface of the photovoltaic module.

[0142] In the technical solution provided in this embodiment,

[0143] First, a single photovoltaic module is selected for cleaning, and a hyperspectral image of the clean photovoltaic module sample is collected; then, a single photovoltaic module with a surface contamination layer is selected, and a hyperspectral image of the surface contaminated photovoltaic module sample is collected; then, a hyperspectral imager mounted on an unmanned aerial vehicle is used to collect hyperspectral images of all photovoltaic modules to be inspected in the photovoltaic power station; then, the collected hyperspectral images are preprocessed, including reflectance correction and outlier removal; then, the photovoltaic module area is segmented and extracted through a photovoltaic module segmentation network; then, the photovoltaic module endmember spectra are extracted from the hyperspectral images of the clean photovoltaic module sample; then, the existing surface contamination endmember spectra are extracted from the hyperspectral images of the surface contaminated photovoltaic module sample; then, the hyperspectral images of all photovoltaic modules to be inspected in the photovoltaic power station are initialized with fixed endmembers to obtain the initial endmember spectra and abundance matrix; then, the hyperspectral images of the photovoltaic module area with surface contamination are robustly unmixed to obtain the final endmember spectra and abundance matrix; finally, the spatial distribution density map of the photovoltaic module surface contamination layer and the warning area are determined.

[0144] See Figure 3 , is a schematic structural diagram of a photovoltaic module surface contamination detection device based on spectral unmixing provided in this embodiment. The device includes a memory 7 and a processor 8 connected to the memory 7.

[0145] The memory 7 is used to store computer programs and data; the processor 8 is used to implement the steps of the above-mentioned photovoltaic module surface contamination detection method based on spectral unmixing when executing the computer program.

[0146] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiment. All technical solutions based on the concept of the present invention are within the scope of protection of the present invention. It should be noted that for those skilled in the art, various improvements and modifications that do not depart from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for detecting contamination on photovoltaic modules based on spectral unmixing, characterized in that: The method comprises: Step S1, selecting a single photovoltaic module for cleaning and collecting a hyperspectral image of the cleaned photovoltaic module sample; Step S2, selecting a single photovoltaic module with a surface contamination layer and collecting a hyperspectral image of the photovoltaic module sample with the surface contamination layer; Step S3, using a drone equipped with a hyperspectral imager to collect hyperspectral images of all photovoltaic modules to be inspected in the photovoltaic power station; Step S4, preprocessing the collected hyperspectral image; Step S5, segmenting and extracting the photovoltaic module area through the photovoltaic module segmentation network; Step S6, extracting photovoltaic module endmember spectra from the hyperspectral image of the clean photovoltaic module sample; Step S7, extracting the existing surface contamination layer endmember spectrum from the hyperspectral image of the surface contamination layer photovoltaic module sample; Step S8, performing initial spectral unmixing with fixed endmembers on the hyperspectral images of all photovoltaic modules to be inspected in the photovoltaic power station to obtain initial endmember spectra and abundance matrices; Step S9, performing robust spectral unmixing on the hyperspectral image of the photovoltaic module area with the surface contamination layer to obtain the final endmember spectrum and abundance matrix; Step S10: determining a spatial distribution density map of the contamination layer on the surface of the photovoltaic module and a warning area.

2. The method for detecting contamination on photovoltaic modules based on spectral unmixing according to claim 1, characterized in that: In step S5, the photovoltaic module segmentation network adopts one of UNet and Mask R-CNN.

3. The method for detecting contamination on photovoltaic modules based on spectral unmixing according to claim 1, characterized in that: In step S6, the endmember spectrum extraction method adopts one of the following methods: pure pixel index, simplex growth, and vertex component analysis. The extracted photovoltaic module endmember spectrum is , where L is the number of bands, is the number of PV module terminals, is the set of real numbers.

4. The method for detecting contamination on the surface of photovoltaic modules based on spectral unmixing according to claim 1, characterized in that: In step S7, the endmember spectrum extraction method adopts one of the methods of pure pixel index, simplex growth and vertex component analysis, and the extracted endmember spectrum of the existing surface pollution layer is , where L is the number of bands, is the number of end members of the pollution layer.

5. The method for detecting surface contamination of photovoltaic modules based on spectral unmixing according to claim 1, characterized in that: In step S8, the initialization spectrum unmixing with some end members fixed includes the following steps: Step S81, set the number of abnormal surface pollution layer end members , the abnormal surface contamination layer end member is the end member of the potential contamination layer that is not included in the surface contamination layer photovoltaic module sample in step S7; Step S82: Set the initial end member set , calculate the initial simplex center ， ； Step S83, calculate the orthogonal projection matrix ， Where U is Centralized calculation results; I is the unit matrix; The superscript T is the matrix transpose operator; is the regularization parameter; Step S84, for , iteratively extract the endmember spectrum of the abnormal surface pollution layer ; Step S85, based on the photovoltaic module end member spectrum , surface pollution layer end-member spectrum and abnormal surface contamination endmember spectra To obtain the initial endmember spectrum results , and based on the initial endmember spectral results , the initial abundance matrix is ​​obtained by estimating the fully constrained least squares method ,in is the photovoltaic module end member spectrum The corresponding abundance matrix, The existing surface contamination end-member spectrum The corresponding abundance matrix, is the abnormal surface contamination endmember spectrum The corresponding abundance matrix.

6. The method for detecting contamination on the surface of photovoltaic modules based on spectral unmixing according to claim 5, characterized in that: In step S84, the abnormal surface contamination endmember spectrum The iterative extraction process is: Step S841: for all pixels in the hyperspectral image , calculate the projection distance , and select the pixel with the largest distance As a new abnormal surface pollution layer end member : ， Step S842: Update the end member set , simplex center Step S843: Update the orthogonal projection matrix Where v is the residual vector; w is the projection vector; Step S844, for Repeat the iterative steps S841 to S843 to add all the new abnormal pollution layer end members Combined together to obtain the abnormal surface pollution layer end-member spectrum, combined photovoltaic module end-member spectrum , surface pollution layer end-member spectrum and abnormal surface contamination endmember spectra , and obtain the initial endmember spectrum results .

7. The method for detecting contamination on photovoltaic modules based on spectral unmixing according to claim 1, characterized in that: In step S9, the robust spectrum unmixing is calculated by repeatedly iteratively updating the initial endmember spectrum result using the following calculation formula: and the initial abundance matrix Where, Represents matrix dot product; is the sparse regularization constraint parameter; 、 is a diagonal matrix, and the lth element on its diagonal is calculated as 。 8. The method for detecting contamination on photovoltaic modules based on spectral unmixing according to claim 1, characterized in that: In step S10, determining the spatial distribution density map and warning area of ​​the contamination layer on the surface of the photovoltaic module includes the following steps: Step S101, from the initial abundance matrix Decompose to obtain the abundance matrix corresponding to the end member of the surface pollution layer , ; Step S102: For each pixel j, calculate the total abundance of the surface pollution layer ， Step S103: The total abundance of the surface pollution layer of each pixel Map pixel coordinates to a two-dimensional matrix ,in and are the height and width of the image respectively; Step S104: When the total abundance of the surface pollution layer of the pixel Greater than the dynamic threshold When the pixel is divided into the warning area, the dynamic threshold The calculation method is Where, is the basic threshold; is the adjustment coefficient; PM is the air index of the day, and Historical air index mean and standard deviation.

9. A photovoltaic module surface contamination detection system based on spectral unmixing for implementing the method of claim 1, characterized in that: The system comprises: A hyperspectral image acquisition unit (1) is used to collect data of the photovoltaic module to be inspected and acquire a hyperspectral image; A preprocessing unit (2) is used for preprocessing the hyperspectral image by performing reflectance correction and outlier removal; A component segmentation unit (3), used for segmenting and extracting the photovoltaic component area; An end member extraction unit (4) is used to extract the end member spectrum of the photovoltaic module and the end member spectrum of the existing surface pollution layer; A spectral unmixing unit (5) is used for initialization spectral unmixing and robust spectral unmixing of fixed end members to obtain end member spectra and abundance matrices; The surface contamination layer detection unit (6) is used to determine the spatial distribution density map and early warning area of ​​the contamination layer on the surface of the photovoltaic module.

10. A photovoltaic module surface contamination detection device based on spectral unmixing, characterized in that: include: Memory (7) for storing computer programs and data; A processor (8) is configured to implement the steps of the photovoltaic module surface contamination detection method based on spectral unmixing as claimed in any one of claims 1 to 8 when executing a computer program.

Images