A method and system for detecting the cleanliness of photovoltaic panels

Abstract

This invention discloses a method and system for detecting the cleanliness of photovoltaic panels, relating to the field of photovoltaic panel detection technology. The method includes first collecting benchmark data of a standard clean photovoltaic panel, dividing the effective light-receiving surface of the panel under test into several independent detection zones, calculating the cleanliness of each zone, extracting the spatial distribution characteristics of cleanliness, calculating the uniformity of light absorption in combination with local light absorption intensity, and obtaining an initial judgment result of overall cleanliness through correlation determination; then, based on the cleanliness threshold of each zone and the natural contamination pattern of the array, selecting contamination characteristic areas and clean benchmark areas to obtain on-site verification data, calibrating the initial judgment result, and obtaining the final detection result. The system includes modules corresponding to the steps of the method. This method can provide specific information on the contamination distribution in each area for the clean operation and maintenance of photovoltaic panels, improve the pertinence and actual implementation effect of photovoltaic panel clean operation and maintenance work, and provide effective data support for the daily operation and maintenance of photovoltaic power plants.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic panel testing technology, and in particular to a method and system for testing the cleanliness of photovoltaic panels. Background Technology

[0002] In the daily operation and maintenance of photovoltaic power plants, the cleanliness of photovoltaic panels is one of the core maintenance aspects. Photovoltaic panels are exposed to the outdoor environment for extended periods, and their effective light-receiving surfaces are prone to accumulating dust, sand, oil, and other contaminants. These contaminants increase the scattering effect of light on the panel surface and reduce the panel's absorption efficiency of sunlight. This not only directly affects the photoelectric conversion efficiency of the photovoltaic panel, but long-term contaminant accumulation can also accelerate the aging of the panel surface material, shortening the lifespan of the photovoltaic panel. Therefore, accurate detection of the cleanliness of photovoltaic panels is of significant practical importance for improving the power generation efficiency and reducing the operation and maintenance costs of photovoltaic power plants.

[0003] Currently, the mainstream methods for testing the cleanliness of photovoltaic panels in the industry mostly collect optical and physical characteristic data of the photovoltaic panel surface (such as surface scattered light intensity, surface roughness, multispectral reflectivity, etc.), directly summarize and calculate the data of the entire photovoltaic panel, and finally obtain a single overall cleanliness value, which is used as the sole basis for judging the cleanliness status of the photovoltaic panel. Although some testing methods may divide the photovoltaic panel into simple areas, they only use the test data of each area to calculate the average of the overall cleanliness value, without further in-depth analysis and processing of the cleanliness data of each area.

[0004] The existing detection methods described above can only output a single numerical value that characterizes the overall cleanliness of the photovoltaic panel. They cannot effectively extract and reflect the spatial distribution characteristics of the surface cleanliness of the photovoltaic panel, nor can they accurately reflect the differences in the degree of pollution in different areas of the photovoltaic panel.

[0005] In actual operation and maintenance of photovoltaic power plants, the contamination of photovoltaic panels often exhibits a non-uniform distribution. A single overall cleanliness value can only roughly determine whether the panel is clean or contaminated, and cannot specify key information such as the specific area and distribution pattern of contamination. Consequently, it cannot provide operation and maintenance personnel with accurate and differentiated cleaning and maintenance basis. Summary of the Invention

[0006] To address the technical problem that existing photovoltaic panel cleanliness detection methods cannot effectively extract and reflect the spatial distribution characteristics of photovoltaic panel surface cleanliness, and cannot accurately reflect the differences in contamination levels in different areas of the photovoltaic panel, this invention provides a photovoltaic panel cleanliness detection method and system.

[0007] The technical solution adopted in this invention is:

[0008] The first aspect of this application provides a method for detecting the cleanliness of a photovoltaic panel, comprising the following steps:

[0009] Step 1: Collect the baseline values ​​of surface scattered light intensity and surface roughness of the photovoltaic panel under standard clean conditions; divide the effective light-receiving surface of the photovoltaic panel under test into several independent detection zones according to the preset grid.

[0010] Step 2: Collect real-time scattered light intensity data and real-time surface roughness data for each independent detection zone. By calculating the increase rate of real-time scattered light intensity data relative to the scattered light intensity benchmark value and the increase rate of real-time surface roughness data relative to the surface roughness benchmark value, and combining the two-dimensional collaborative weight calculation, the cleanliness of each independent detection zone is obtained.

[0011] Step 3: Based on the cleanliness of all independently detected zones, extract the spatial distribution characteristics of the photovoltaic panel's cleanliness.

[0012] Step 4: Collect measured data of local light absorption intensity of each independent detection zone of the photovoltaic panel, and calculate the light absorption uniformity of the photovoltaic panel based on the measured data of local light absorption intensity.

[0013] Step 5: Correlate the spatial distribution characteristics of cleanliness with the uniformity of light absorption to obtain the preliminary judgment result of the overall cleanliness of the photovoltaic panel;

[0014] Step 6: Select the pollution characteristic area and the clean reference area of ​​the photovoltaic panel, and collect the on-site cleanliness status verification data of the pollution characteristic area and the clean reference area;

[0015] Step 7: Combine the on-site cleaning status verification data to calibrate the preliminary overall cleanliness judgment result and obtain the overall cleanliness test result of the photovoltaic panel.

[0016] Preferably, the partition cleanliness of each independent detection partition obtained by combining the two-dimensional collaborative weight calculation in step 2 includes the following:

[0017] The weighting coefficient for the amplification rate of scattered light intensity is set as follows: The weighting coefficient for the surface roughness increase rate is ,and The cleanliness of a zone is calculated using the following formula:

[0018]

[0019] In the formula, The cleanliness level of the zone ranges from 0 to 1. The amplification rate of the scattered light intensity; This represents the rate of increase in surface roughness.

[0020] Preferably, the process of extracting the spatial distribution features of cleanliness in step 3 is as follows: the gradient distribution features of cleanliness of the partition are extracted by gradient analysis algorithm, the clustering distribution features of the high-pollution partition are extracted by density clustering algorithm, and the connectivity distribution features of the clean partition are extracted by neighborhood connectivity algorithm. The gradient distribution features, clustering distribution features, and connectivity distribution features constitute the spatial distribution features of cleanliness.

[0021] Preferably, the calculation process for the uniformity of light absorption in step 4 is as follows: the coefficient of variation of the measured local light absorption intensity data is calculated using the following formula:

[0022]

[0023] In the formula, The coefficient of variation is 1. is the standard deviation of the measured data, and Mean is the average value of the measured data; the coefficient of variation is normalized to obtain the light absorption uniformity index with a value range of 0-1.

[0024] Preferably, the correlation determination process in step 5 is as follows:

[0025] Establish a rule base for the correlation between the spatial distribution characteristics of cleanliness and the uniformity of light absorption. The rule base is pre-set with different levels of feature thresholds and uniformity thresholds.

[0026] The extracted cleanliness spatial distribution features and light absorption uniformity index are matched with thresholds in the rule base, and the matching confidence is calculated. When the confidence is higher than the preset confidence threshold, the corresponding overall cleanliness preliminary judgment result is output. When it is lower than the preset confidence threshold, the matching threshold is re-optimized and the judgment is made again.

[0027] Preferably, the selection process for the pollution characteristic area and the clean baseline area in step 6 is as follows:

[0028] Preset first threshold Second threshold ,and The cleanliness of the partition is lower than The continuous and independent detection zones form a pollution characteristic area, and the cleanliness of the zone is higher than that of the other zones. The continuous independent detection zones constitute the clean reference area; the first threshold Second threshold It can be dynamically adjusted according to the environmental parameters of the photovoltaic panel.

[0029] Preferably, the calibration process in step 7 is as follows:

[0030] Outliers are removed from the on-site cleaning status verification data, and valid data within a preset reasonable range are retained.

[0031] Calculate the deviation between the valid data and the previous detection data of the corresponding independent detection zone. ;

[0032] Based on deviation value The overall deviation correction factor K is calculated using the following formula:

[0033]

[0034] In the formula, The overall deviation correction factor is m, where m is the index of the valid data, and the value is 1, 2, ..., n; For the first The deviation value of each valid data point. The number of valid data;

[0035] Based on the overall deviation correction coefficient The quantitative value of the initial overall cleanliness assessment obtained in step 5 is corrected for error to obtain the quantitative value of overall cleanliness. The calibration formula is as follows:

[0036]

[0037] in, This is the initial assessment of the overall cleanliness.

[0038] The second aspect of this application provides a photovoltaic panel cleanliness detection system, which applies the above-mentioned photovoltaic panel cleanliness detection method, including:

[0039] The benchmark acquisition and partitioning module is used to acquire the benchmark values ​​of surface scattered light intensity and surface roughness of the photovoltaic panel under standard clean conditions; and to divide the effective light-receiving surface of the photovoltaic panel under test into several independent detection zones according to a preset grid.

[0040] The partition detection and cleanliness calculation module is used to collect real-time scattered light intensity data and real-time surface roughness data of each independent detection partition. By calculating the increase rate of real-time scattered light intensity data relative to the scattered light intensity benchmark value and the increase rate of real-time surface roughness data relative to the surface roughness benchmark value, and combining the two-dimensional collaborative weight calculation, the partition cleanliness of each independent detection partition is obtained.

[0041] A cleanliness spatial feature extraction module is used to extract the spatial distribution features of the cleanliness of photovoltaic panels based on the cleanliness of all independent detection zones.

[0042] The light absorption detection and uniformity calculation module is used to collect measured local light absorption intensity data of each independent detection zone of the photovoltaic panel, and calculate the light absorption uniformity of the photovoltaic panel based on the measured local light absorption intensity data.

[0043] The cleanliness correlation determination and preliminary judgment module is used to determine the correlation between the spatial distribution characteristics of cleanliness and the uniformity of light absorption to obtain the preliminary judgment result of the overall cleanliness of the photovoltaic panel.

[0044] The on-site verification area selection and data acquisition module is used to select the pollution characteristic area and the clean benchmark area of ​​the photovoltaic panel, and to collect on-site cleanliness status verification data of the pollution characteristic area and the clean benchmark area.

[0045] The cleanliness result calibration and detection module is used to calibrate the overall cleanliness preliminary judgment result by combining the on-site cleanliness status verification data, and obtain the overall cleanliness detection result of the photovoltaic panel.

[0046] The beneficial effects of the present invention are at least one of the following:

[0047] This method first divides the effective light-receiving surface of the photovoltaic panel under test into several independent detection zones. After calculating the cleanliness of each independent detection zone, it further extracts the spatial distribution characteristics of the cleanliness of the photovoltaic panel, which can clearly reflect the differences in the degree of pollution in different areas of the photovoltaic panel, making the cleanliness test results of the photovoltaic panel more consistent with its actual pollution distribution.

[0048] Based on the extraction of spatial distribution characteristics of cleanliness, this method combines steps such as correlation determination of light absorption uniformity and calibration of on-site cleanliness status verification data to obtain the overall cleanliness test results of photovoltaic panels. The results are a comprehensive judgment based on the actual pollution status of each area. Compared with the existing test results obtained by simply summarizing and calculating the overall data, the accuracy of the test results is effectively improved.

[0049] The spatial distribution characteristics of photovoltaic panel cleanliness extracted by this method can provide specific information on the pollution distribution in each area for the clean operation and maintenance of photovoltaic panels. Based on this, targeted and differentiated clean operation and maintenance work can be carried out, which helps to reduce the unreasonable consumption of clean resources, improve the pertinence and actual implementation effect of photovoltaic panel clean operation and maintenance work, and provide effective data support for the daily operation and maintenance of photovoltaic power plants. Attached Figure Description

[0050] Figure 1 This is a schematic diagram of the method flow of Embodiment 1 of the present invention;

[0051] Figure 2 This is a system structure block diagram of Embodiment 2 of the present invention. Detailed Implementation

[0052] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0053] As the core component of photovoltaic power plants, the cleanliness of photovoltaic panels directly determines the photoelectric conversion efficiency, thus affecting the power generation benefits and operation and maintenance costs of the photovoltaic power plant. In actual outdoor operation and maintenance scenarios, photovoltaic panels are exposed to the natural environment for a long time, and their surfaces are prone to accumulating pollutants such as dust, sand, and oil. Furthermore, the distribution of these pollutants is often non-uniform. For example, areas of the photovoltaic panel near the ground tend to accumulate dust, while the windward side is prone to accumulating pollutants, and other areas are relatively clean. This non-uniform contamination characteristic places high demands on the refinement and precision of cleanliness testing methods.

[0054] Currently, the mainstream photovoltaic panel cleanliness testing methods in the industry are mainly divided into two categories: one is the overall testing method, which directly collects relevant data of the entire photovoltaic array, summarizes and calculates to obtain a single overall cleanliness value, and uses this as the sole basis for judging the cleanliness status of the array; the other is the simple zoning testing method, which divides the array into regions, but only averages the test data of each region, and finally outputs a single overall cleanliness value without performing in-depth analysis and feature extraction of the cleanliness data of each region.

[0055] Both of the above-mentioned existing detection methods have obvious defects. Since they only output a single overall cleanliness value, they cannot reflect the differences in the degree of pollution in different areas of the photovoltaic array, nor can they reflect the spatial distribution pattern of pollutants, and thus cannot meet the actual needs of refined operation and maintenance of photovoltaic power plants.

[0056] To address the problems that existing methods cannot adapt to large-area photovoltaic arrays, cannot reflect the spatial distribution characteristics of cleanliness, and cannot accurately reflect the differences in pollution levels in different locations, such as... Figure 1 As shown, it includes the following steps:

[0057] Step 1: Collect the baseline values ​​of surface scattered light intensity and surface roughness of the photovoltaic panel under standard clean conditions; divide the effective light-receiving surface of the photovoltaic panel under test into several independent detection zones according to the preset grid.

[0058] It should be noted that a standard clean photovoltaic panel refers to a photovoltaic panel with no contaminants attached to its surface, and whose surface material and optical properties are in the factory standard condition. Its surface light scattering intensity and surface roughness can be used as a benchmark reference for subsequent testing. In addition, the model and specifications of this panel are completely consistent with the photovoltaic panel under test to ensure the comparability of the benchmark data.

[0059] The surface scattered light intensity benchmark value refers to the measured value of the scattered light intensity of a photovoltaic panel surface under standard clean conditions. It is the core benchmark for measuring the change in scattered light intensity on the surface of the photovoltaic panel under test and judging the degree of contamination. The surface roughness benchmark value refers to the measured value of the surface roughness of a photovoltaic panel under standard clean conditions. It is the core benchmark for measuring the change in surface roughness of the photovoltaic panel under test and judging the degree of contamination.

[0060] The effective light-receiving surface refers to the area of ​​the photovoltaic panel that can receive sunlight and perform photoelectric conversion, excluding the edges of the panel, such as the frame and junction box, which do not have photoelectric conversion functions. The preset grid refers to a pre-defined grid used to divide the effective light-receiving surface of the photovoltaic panel into regions. It is the basis for forming independent detection zones, and its division rules are predetermined according to the panel size and detection accuracy requirements.

[0061] Independent detection zones refer to several independent, non-overlapping, and fully covered smallest detection units formed on the effective light-receiving surface of a photovoltaic panel after being divided by a preset grid. Each independent detection zone corresponds to a unique set of detection data, which can accurately reflect the cleanliness status of the area.

[0062] Because contamination of photovoltaic panels increases surface scattering light intensity and surface roughness, only by obtaining baseline data under standard clean conditions can the degree of contamination in each area of ​​the panel under test be accurately quantified through comparative calculations. Without baseline data, it is impossible to determine whether changes in real-time detection data are caused by contamination or differences in the panel's inherent characteristics, leading to distorted test results. Dividing the photovoltaic panel under test into independent testing zones is fundamental to achieving refined testing, avoiding the problem of regional differences masking by overall testing or simple averaging of zones.

[0063] In the specific implementation process, a standard clean photovoltaic panel with the same model and specifications as the photovoltaic panel to be tested is first selected. Under the same environmental conditions as the photovoltaic panel to be tested in subsequent testing, the surface scattered light intensity benchmark value and surface roughness benchmark value are collected. During the collection process, the environmental conditions are kept consistent. Each benchmark value is collected multiple times, and the average value is taken as the final benchmark data to ensure the accuracy and stability of the benchmark data.

[0064] Then, based on the size of the effective light-receiving surface of the photovoltaic panel under test, a preset grid division rule is set. The division rule strictly follows the principle of no overlap and full coverage of the effective light-receiving surface. After division, the size of each independent detection zone is determined according to the preset precision to ensure that the detection data of each independent detection zone can truly reflect the cleanliness status of the area. After the division is completed, a unique location identifier is assigned to each independent detection zone, and the physical coordinates of each independent detection zone on the effective light-receiving surface are recorded to facilitate subsequent data association and location positioning.

[0065] Step 2: Collect real-time scattered light intensity data and real-time surface roughness data for each independent detection zone. By calculating the increase rate of real-time scattered light intensity data relative to the scattered light intensity benchmark value and the increase rate of real-time surface roughness data relative to the surface roughness benchmark value, and combining the two-dimensional collaborative weight calculation, the cleanliness of each independent detection zone is obtained.

[0066] It should be noted that real-time scattered light intensity data refers to the value of the scattered light intensity of the surface of each independent testing zone of the photovoltaic panel under test, obtained from the surface of that zone. Real-time surface roughness data refers to the surface roughness value of each independent testing zone of the photovoltaic panel under test, obtained from the surface of that zone.

[0067] The rate of increase in scattered light intensity refers to the proportion of increase in real-time scattered light intensity data relative to the baseline value of surface scattered light intensity. It is used to quantitatively characterize the degree of change in the scattered light intensity of the tested area relative to a standard clean state. The larger the rate of increase, the more severe the contamination of the area. The rate of increase in surface roughness refers to the proportion of increase in real-time surface roughness data relative to the baseline value of surface roughness. It is used to quantitatively characterize the degree of change in the surface roughness of the tested area relative to a standard clean state. The larger the rate of increase, the more severe the contamination of the area.

[0068] The dual-dimensional collaborative weighted calculation method combines two dimensions: the amplification rate of scattered light intensity and the amplification rate of surface roughness. A corresponding weight coefficient is assigned to each dimension, and the sum of the weight coefficients for both dimensions is 1. This weighted calculation yields the cleanliness of each zone, used to comprehensively quantify the cleanliness level of each independent detection zone. Zone cleanliness refers to a quantitative index of the cleanliness level of each independent detection zone, with a value ranging from 0 to 1. A value closer to 1 indicates a cleaner zone, while a value closer to 0 indicates a more severely contaminated zone.

[0069] Because contaminants adhering to the surface of photovoltaic panels disrupt their smoothness, leading to enhanced scattering of incident light and increased surface roughness, both of which are positively correlated with the degree of contamination, a single indicator cannot comprehensively and accurately quantify the cleanliness of a zone. Employing a dual-dimensional synergistic weighting calculation comprehensively considers the impact of both key indicators, overcoming the limitations of single-indicator detection and making the calculation of zone cleanliness more comprehensive and closer to actual contamination conditions.

[0070] In the specific implementation process, according to the location identifier of each independent detection zone, the real-time scattered light intensity data and real-time surface roughness data of each independent detection zone are acquired one by one. During the acquisition process, the environmental conditions for data acquisition are kept consistent with those for the baseline data acquisition. Each type of data for each independent detection zone is acquired multiple times, and the average value is taken as the final data for that zone.

[0071] The amplification rate of scattered light intensity and the amplification rate of surface roughness for each independent detection zone were calculated separately. The amplification rate was obtained by the relative amplification calculation method.

[0072] The weighting coefficient for the amplification rate of scattered light intensity is set as follows: The weighting coefficient for the surface roughness increase rate is ,and The cleanliness of a zone is calculated using the following formula:

[0073]

[0074] In the formula, The cleanliness level of the zone ranges from 0 to 1. The amplification rate of the scattered light intensity; This represents the rate of increase in surface roughness.

[0075] After the calculation is completed, the cleanliness of each independent detection partition is bound and stored with the corresponding location identifier for easy retrieval and analysis in subsequent steps.

[0076] Step 3: Based on the cleanliness of all independent detection zones, extract the spatial distribution characteristics of the cleanliness of the photovoltaic panels.

[0077] It should be noted that the spatial distribution characteristics of cleanliness refer to the spatial distribution patterns and characteristics of the cleanliness of each independent detection zone on the effective light-receiving surface of the photovoltaic panel. It mainly includes the gradient change characteristics of the cleanliness of the zones, the clustering distribution characteristics of the high-pollution zones, and the connectivity distribution characteristics of the clean zones. These three together constitute the complete spatial distribution characteristics of cleanliness, which can comprehensively reflect the spatial distribution patterns of pollutants.

[0078] The spatial distribution characteristics of cleanliness are key to overcoming the shortcomings of existing technologies. Current methods fail to reflect regional pollution differences because they do not extract this characteristic. Based on the cleanliness of all independently monitored zones and combined with the location information of each zone, spatial analysis algorithms can extract the spatial distribution patterns of cleanliness, clearly showing which areas are severely polluted, which areas are clean, and how pollutants accumulate.

[0079] Specifically, in one possible implementation, gradient distribution features of cleanliness in a zone are extracted using a gradient analysis algorithm, clustering distribution features of highly polluted zones are extracted using a density clustering algorithm, and connectivity distribution features of clean zones are extracted using a neighborhood connectivity algorithm. The gradient distribution features, clustering distribution features, and connectivity distribution features constitute the spatial distribution features of cleanliness.

[0080] In this step, the gradient analysis algorithm uses the surface gradient calculation method, which quantifies the spatial rate and direction of change of cleanliness by calculating the ratio of the cleanliness difference between adjacent partitions to the spatial distance.

[0081] The density clustering algorithm used in this step is the existing conventional DBSCAN algorithm (density-based noisy spatial clustering algorithm), which does not require a preset number of clusters and can adaptively identify the clustering status of highly polluted partitions.

[0082] The neighborhood connectivity algorithm in this step uses the existing conventional 8-neighborhood connectivity analysis algorithm (a general algorithm for spatial pixel / partition connectivity determination). It extracts the connectivity state of clean partitions by determining whether the cleanliness of adjacent partitions meets the threshold condition.

[0083] In the specific implementation process, firstly, the cleanliness data of all independent detection zones and the physical coordinates corresponding to each zone are retrieved (a Cartesian coordinate system is established with the lower left corner of the effective light-receiving surface of the photovoltaic panel as the origin, and the physical coordinates of each zone are represented by the coordinates (x, y) of its geometric center). The cleanliness data C of each zone is then associated and bound with the physical coordinates (x, y) to form {(x1, y1, C1), (x2, y2, C2), ..., (x N ,y N C N The dataset is N, where N is the total number of independent detection partitions.

[0084] In the gradient analysis algorithm (surface gradient calculation method), the cleanliness gradient value is the ratio of the cleanliness difference between adjacent partitions to the spatial Euclidean distance, and the calculation formula is as follows:

[0085]

[0086] In the formula: This represents the cleanliness gradient value between partition i and its adjacent partition j. , These represent the partition cleanliness of partitions i and j, respectively.

[0087] , These are the geometric center coordinates of partitions i and j, respectively.

[0088] In practical applications, for each independent detection partition in the dataset, all its neighboring partitions (within an 8-neighborhood range) are traversed; the cleanliness gradient value between the partition and each neighboring partition is calculated according to the above formula; the average gradient value and the direction of maximum gradient for each partition are statistically analyzed, and a gradient heat map of the effective light-receiving surface of the photovoltaic panel is drawn; the overall trend of cleanliness change is extracted from the heat map, such as "gradient value gradually decreases from bottom to top" and "gradient value is higher in the west than in the east", forming gradient distribution characteristics.

[0089] The core parameters and decision rules of the density clustering algorithm (DBSCAN algorithm) are as follows:

[0090] The neighborhood radius ε is set to the side length of two independent detection zones to fit the size of the photovoltaic panel zones and can be adjusted according to the actual zone accuracy; the minimum number of points MinPts is set to 3, meaning that at least 3 consecutive zones must meet the conditions to be considered as a clustered region; the clustering threshold is preset to 0.4, meaning that when the zone cleanliness C < 0.4, it is considered a candidate high-pollution zone.

[0091] In practical applications, all candidate high-pollution partitions (C<0.4) are selected from the dataset to form a high-pollution partition subset. The high-pollution partition subset is input into the DBSCAN algorithm, with ε=2×partition side length and MinPts=3. The algorithm automatically identifies the core point that satisfies the condition of containing at least 3 high-pollution partitions in its neighborhood, as well as all high-pollution partitions connected to the core point, forming high-pollution cluster regions. The geometric range, cluster area ratio, and cluster center coordinates of each cluster region are extracted to form cluster distribution characteristics.

[0092] The neighborhood connectivity algorithm (8-neighborhood connectivity analysis algorithm) has the following decision rules:

[0093] An 8-neighborhood refers to the eight adjacent partitions in the directions of up, down, left, right, upper left, lower left, upper right, and lower right of a given partition. The connectivity threshold is preset to 0.8, meaning that a partition is considered a candidate clean partition when its cleanliness C > 0.8. Connectivity is determined by the presence of at least one other candidate clean partition within its 8-neighborhood. Continuously connected candidate clean partitions form a connected region.

[0094] In practical applications, all candidate clean partitions (C>0.8) are selected from the dataset to form a subset of clean partitions; for each candidate clean partition, its 8-neighborhood is traversed to determine whether it is connected to other candidate clean partitions; all connected clean partitions are marked as the same connected region, and the area, shape, and position of each connected region are statistically analyzed, such as the 5×4 partition connected region on the upper right of the panel; the number of connected regions and the proportion of the largest connected region are extracted to form connectivity distribution features.

[0095] The aforementioned gradient distribution features, clustering distribution features, and connectivity distribution features together constitute the spatial distribution features of the cleanliness of photovoltaic panels. After extraction, the features are organized to form feature data that can be used for subsequent correlation determination, such as gradient trend descriptions, cluster region coordinates, and connected region areas, which are then bound and stored with the location information of each partition.

[0096] Step 4: Collect measured data of local light absorption intensity of each independent detection zone of the photovoltaic panel, and calculate the light absorption uniformity of the photovoltaic panel based on the measured data of local light absorption intensity.

[0097] It should be noted that the measured local light absorption intensity data refers to the absorption intensity value of sunlight on the surface of each independent testing zone of the photovoltaic panel under test. It is positively correlated with the cleanliness of the zone; the more polluted the area, the lower the light absorption intensity. Light absorption uniformity refers to the uniformity of local light absorption intensity among the independent testing zones of the effective light-receiving surface of the photovoltaic panel. It is an important indicator characterizing the overall light absorption performance of the photovoltaic panel and can also indirectly reflect the overall cleanliness of the panel. The better the light absorption uniformity, the smaller the difference in cleanliness between different areas of the panel, and the better the overall cleanliness.

[0098] Since the core function of a photovoltaic panel is to absorb sunlight and perform photoelectric conversion, cleanliness directly affects its light absorption performance. Severely polluted areas will lead to a decrease in light absorption intensity in that area, thereby disrupting the uniformity of light absorption across the entire panel. Calculating the uniformity of light absorption can help determine the cleanliness status of a photovoltaic panel from the perspective of light absorption performance. This complements the spatial distribution characteristics of cleanliness and can improve the accuracy of subsequent overall cleanliness assessments.

[0099] In the specific implementation process, according to the location identifier of each independent detection zone, the measured local light absorption intensity data of each independent detection zone is obtained one by one. During the acquisition process, the detection environment is kept consistent with the previous data acquisition environment. The measured local light absorption intensity data of each independent detection zone is obtained multiple times, and the average value is taken as the final data of that zone to ensure the accuracy and consistency of the data.

[0100] This step uses the coefficient of variation method to calculate the uniformity of light absorption in photovoltaic panels. The coefficient of variation method is a conventional method for calculating uniformity, and its specific application process is as follows: First, calculate the mean and standard deviation of the measured local light absorption intensity data for all independently tested zones. The coefficient of variation is the ratio of the standard deviation to the mean, and the calculation formula is as follows:

[0101]

[0102] In the formula, The coefficient of variation is 1. denoted as the standard deviation of the measured data, and Mean as the average value of the measured data. The calculated coefficient of variation is normalized using a linear normalization method, mapping the coefficient of variation to the range of 0 to 1 to obtain the light absorption uniformity index. The larger the value of the light absorption uniformity index, the more uniform the light absorption intensity in each area of ​​the photovoltaic panel, and the better the overall cleanliness of the panel. The smaller the index value, the worse the uniformity of light absorption intensity, and the more obvious the regional pollution differences in the panel.

[0103] Step 5: Correlation determination is made between the spatial distribution characteristics of cleanliness and the uniformity of light absorption to obtain the preliminary judgment result of the overall cleanliness of the photovoltaic panel.

[0104] It should be noted that correlation determination refers to establishing the intrinsic relationship between the spatial distribution characteristics of cleanliness and the uniformity of light absorption, and combining the index data of both to make a comprehensive judgment on the overall cleanliness status of the photovoltaic panel.

[0105] The preliminary overall cleanliness assessment result refers to the preliminary assessment result of the overall cleanliness of the photovoltaic panel based on the correlation between the spatial distribution characteristics of cleanliness and the uniformity of light absorption. It includes the quantitative value of the overall cleanliness and the corresponding cleanliness level, providing a basis for subsequent calibration steps.

[0106] While both the spatial distribution characteristics of cleanliness and the uniformity of light absorption can reflect the cleanliness of a photovoltaic panel, each indicator has its limitations. By establishing the intrinsic relationship between these two indicators and integrating their information, a comprehensive assessment of the overall cleanliness of the panel can be achieved, avoiding the one-sidedness of relying on a single indicator.

[0107] In the specific implementation process, firstly, based on the historical data of photovoltaic panel cleanliness detection and the photoelectric conversion characteristics of photovoltaic panels, a rule library for the association between cleanliness spatial distribution characteristics and light absorption uniformity is established. The rule library presets cleanliness spatial distribution characteristic thresholds and light absorption uniformity index thresholds corresponding to different cleanliness levels. Each cleanliness level corresponds to a unique set of characteristic thresholds and uniformity thresholds.

[0108] The cleanliness spatial distribution features extracted in step 3 and the light absorption uniformity index calculated in step 4 are retrieved. The cleanliness spatial distribution features are matched one by one with the feature thresholds of each level in the rule base. The light absorption uniformity index is compared with the uniformity thresholds of each level in the rule base. Based on the matching and comparison results, the existing conventional confidence calculation method is used to calculate the matching confidence. The confidence is used to characterize the reliability of the matching results and the value ranges from 0 to 1.

[0109] When the calculated confidence level is higher than the preset confidence threshold, the corresponding preliminary overall cleanliness judgment result is directly output, including the overall cleanliness quantification value and cleanliness level; when the confidence level is lower than the preset confidence threshold, the matching threshold in the rule base is re-optimized, and the matching judgment is performed again until the preliminary overall cleanliness judgment result that meets the requirements is obtained.

[0110] Step 6: Select the pollution characteristic area and the clean reference area of ​​the photovoltaic panel, and collect on-site cleanliness status verification data of the pollution characteristic area and the clean reference area.

[0111] It should be noted that the pollution characteristic area refers to the lower part of all photovoltaic panels in a large-area photovoltaic array, close to the ground. This area is affected by sand and dust accumulation and mud splashing, and is the continuous area with the most severe pollution within the array. It does not require manual search and is directly delineated based on the natural pollution patterns of the array.

[0112] The clean baseline area refers to the uniform upper high area of ​​all photovoltaic panels in a large-area photovoltaic array. This area has good ventilation and low pollutant adhesion, and is the continuous area with the lightest pollution level in the array, but it is not absolutely free of pollutants.

[0113] The on-site cleaning status verification data refers to the detection data obtained again based on the natural pollution patterns of large-area photovoltaic arrays, targeting the most heavily polluted and the least polluted areas within the array. It is only used for error calibration of the detection results and does not involve any zonal cleaning operations.

[0114] Because the distribution of pollutants in large-area photovoltaic arrays follows certain patterns, it is unnecessary to search for contaminated areas on each panel individually. Instead, the contaminated characteristic areas and clean baseline areas can be directly delineated based on the upper and lower sections of the array. During the initial data acquisition process, environmental fluctuations may have affected the data, leading to overall errors. By verifying the data from these two representative locations across the entire area, the systematic errors of the initial detection can be effectively eliminated, ensuring that the final detection results accurately reflect the actual cleanliness of the large-area photovoltaic array.

[0115] In the specific implementation process, the selection process for the pollution characteristic area and the clean baseline area mentioned in step 6 is as follows:

[0116] Preset first threshold Second threshold ,and The cleanliness of the partition is lower than The continuous and independent detection zones form a pollution characteristic area, and the cleanliness of the zone is higher than that of the other zones. The continuous independent detection zones constitute the clean reference area; the first threshold Second threshold It can be dynamically adjusted according to the environmental parameters of the photovoltaic panel.

[0117] Based on the overall contamination patterns of the array, the lower part of all photovoltaic panels within the array, near the ground and with a cleanliness level lower than [specific level], was selected. Continuous independent detection zones were ultimately identified as areas with contamination characteristics; the upper high-level areas of all photovoltaic panels within the array, and areas with a cleanliness level higher than [missing information], were also identified. The continuous independent testing zones were ultimately determined as the clean baseline area.

[0118] In the field environment, the scattered light intensity data, surface roughness data, and light absorption intensity data of each independent detection zone in the pollution characteristic area and the clean baseline area are acquired. Each data is acquired multiple times, and the average value is taken as the final field clean status verification data. After acquisition, the verification data is associated with the previous data of the corresponding location and the corresponding independent detection zone for easy calculation of deviation values ​​in the future.

[0119] Step 7: Combine the on-site cleaning status verification data to calibrate the preliminary overall cleanliness judgment result and obtain the overall cleanliness test result of the photovoltaic panel.

[0120] It should be noted that calibration refers to the process of calculating the deviation between the previous data and the on-site data based on the on-site cleanliness verification data, and uniformly correcting the initial cleanliness judgment result through the deviation correction coefficient, thereby eliminating the overall error of the initial test and making the test result more consistent with the actual cleanliness status of the large-area photovoltaic array.

[0121] The overall cleanliness test result refers to the final judgment result of the overall cleanliness of the photovoltaic array after calibration, including the calibrated overall cleanliness quantification value and the corresponding cleanliness level, which can directly guide the overall cleanliness operation and maintenance of the photovoltaic array.

[0122] The deviation value refers to the difference between the on-site cleaning status verification data and the previous data of the corresponding independent detection zone, which is used to characterize the overall error of the previous data.

[0123] The overall deviation correction coefficient refers to the average value of all effective deviation values. It is used to uniformly correct the quantitative value of the initial judgment result of overall cleanliness, and adapts to the global error calibration requirements of large-area arrays.

[0124] Because the initial inspection of large-area photovoltaic arrays is easily affected by overall environmental fluctuations, resulting in global inspection errors, which in turn affect the accuracy of the initial cleanliness assessment. By calculating the deviation value and the overall deviation correction coefficient using array-level calibration data, the initial assessment results can be calibrated globally, effectively eliminating systematic errors and ensuring that the final inspection results accurately match the actual cleanliness status of the large-area photovoltaic array.

[0125] In the specific implementation process, outliers are first removed from the on-site cleaning status verification data, and valid data within the preset reasonable range are retained;

[0126] Calculate the deviation between the valid data and the previous detection data of the corresponding independent detection zone. ;

[0127] Based on deviation value The overall deviation correction factor K is calculated using the following formula:

[0128]

[0129] In the formula, The overall deviation correction factor is m, where m is the index of the valid data, and the value is 1, 2, ..., n; For the first The deviation value of each valid data point. The number of valid data;

[0130] Based on the overall deviation correction coefficient The quantitative value of the initial overall cleanliness assessment obtained in step 5 is corrected for error to obtain the quantitative value of overall cleanliness. The calibration formula is as follows:

[0131]

[0132] in, This is the initial assessment of the overall cleanliness.

[0133] Based on the calibrated overall cleanliness quantification By comparing the cleanliness level thresholds with those in the association rule base, the cleanliness level is updated accordingly, and the overall cleanliness detection result of the photovoltaic array is obtained. The detection result can directly guide the overall cleaning and maintenance operations.

[0134] Example 2

[0135] This embodiment provides a photovoltaic panel cleanliness detection system, which applies the photovoltaic panel cleanliness detection method described above, such as... Figure 2 As shown, it includes:

[0136] The benchmark acquisition and partitioning module is used to acquire the benchmark values ​​of surface scattered light intensity and surface roughness of the photovoltaic panel under standard clean conditions; and to divide the effective light-receiving surface of the photovoltaic panel under test into several independent detection zones according to a preset grid.

[0137] The partition detection and cleanliness calculation module is used to collect real-time scattered light intensity data and real-time surface roughness data of each independent detection partition. By calculating the increase rate of real-time scattered light intensity data relative to the scattered light intensity benchmark value and the increase rate of real-time surface roughness data relative to the surface roughness benchmark value, and combining the two-dimensional collaborative weight calculation, the partition cleanliness of each independent detection partition is obtained.

[0138] A cleanliness spatial feature extraction module is used to extract the spatial distribution features of the cleanliness of photovoltaic panels based on the cleanliness of all independent detection zones.

[0139] The light absorption detection and uniformity calculation module is used to collect measured local light absorption intensity data of each independent detection zone of the photovoltaic panel, and calculate the light absorption uniformity of the photovoltaic panel based on the measured local light absorption intensity data.

[0140] The cleanliness correlation determination and preliminary judgment module is used to determine the correlation between the spatial distribution characteristics of cleanliness and the uniformity of light absorption to obtain the preliminary judgment result of the overall cleanliness of the photovoltaic panel.

[0141] The on-site verification area selection and data acquisition module is used to select the pollution characteristic area and the clean benchmark area of ​​the photovoltaic panel, and to collect on-site cleanliness status verification data of the pollution characteristic area and the clean benchmark area.

[0142] The cleanliness result calibration and detection module is used to calibrate the overall cleanliness preliminary judgment result by combining the on-site cleanliness status verification data, and obtain the overall cleanliness detection result of the photovoltaic panel.

[0143] The embodiments described above are merely illustrative of specific implementations of the present invention, and while the descriptions are detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention.

Claims

1. A photovoltaic panel cleanliness detection method, characterized by, Includes the following steps: Step 1: Collect the surface scattered light intensity reference value and surface roughness reference value of the photovoltaic panel in standard clean condition, and divide the effective light-receiving surface of the photovoltaic panel under test into several independent detection zones according to the preset grid. Step 2: Collect real-time scattered light intensity data and real-time surface roughness data for each independent detection zone. By calculating the increase rate of real-time scattered light intensity data relative to the scattered light intensity benchmark value and the increase rate of real-time surface roughness data relative to the surface roughness benchmark value, and combining the two-dimensional collaborative weight calculation, the cleanliness of each independent detection zone is obtained. Step 3: Based on the cleanliness of all independent detection zones, extract the spatial distribution features of the cleanliness of the photovoltaic panel. Extract the gradient distribution features of the cleanliness of the zones through gradient analysis algorithm, extract the cluster distribution features of the high-pollution zones through density clustering algorithm, and extract the connectivity distribution features of the clean zones through neighborhood connectivity algorithm. The gradient distribution features, cluster distribution features, and connectivity distribution features constitute the spatial distribution features of cleanliness. Step 4: Collect measured data of local light absorption intensity of each independent detection zone of the photovoltaic panel, and calculate the light absorption uniformity of the photovoltaic panel based on the measured data of local light absorption intensity. Step 5: The spatial distribution characteristics of cleanliness are correlated with the uniformity of light absorption to obtain the preliminary judgment result of the overall cleanliness of the photovoltaic panel. First, based on the historical data of the cleanliness detection of the photovoltaic panel and the photoelectric conversion characteristics of the photovoltaic panel, a rule library for the correlation between the spatial distribution characteristics of cleanliness and the uniformity of light absorption is established. The rule library is preset with thresholds for the spatial distribution characteristics of cleanliness and thresholds for the uniformity of light absorption corresponding to different cleanliness levels. Each cleanliness level corresponds to a unique set of characteristic thresholds and uniformity thresholds. Retrieve the cleanliness spatial distribution features extracted in step 3 and the light absorption uniformity index calculated in step 4. Match the cleanliness spatial distribution features with the feature thresholds of each level in the rule base one by one. Compare the light absorption uniformity index with the uniformity thresholds of each level in the rule base. Calculate the matching confidence based on the matching and comparison results. When the calculated confidence level is higher than the preset confidence threshold, the corresponding preliminary judgment result of overall cleanliness is directly output, including the overall cleanliness quantification value and cleanliness level. When the confidence level is lower than the preset confidence threshold, the matching threshold in the rule base is re-optimized, and the matching judgment is performed again until the preliminary judgment result of overall cleanliness that meets the requirements is obtained. Step 6: Select the pollution characteristic area and the clean reference area of ​​the photovoltaic panel, and collect the on-site cleanliness status verification data of the pollution characteristic area and the clean reference area; Preset first threshold Second threshold ,and The cleanliness of the partition is lower than The continuous and independent detection zones form a pollution characteristic area, and the cleanliness of the zone is higher than that of the other zones. The continuous independent detection zones constitute the clean reference area, and the first threshold Second threshold It can be dynamically adjusted according to the environmental parameters of the photovoltaic panel. Step 7: Combine the on-site cleanliness status verification data to calibrate the initial overall cleanliness judgment result and obtain the overall cleanliness test result of the photovoltaic panel.

2. The method for detecting the cleanliness of a photovoltaic panel according to claim 1, characterized in that, The partition cleanliness of each independent detection partition, calculated by combining two-dimensional collaborative weights in step 2, includes the following: The weighting coefficient for the amplification rate of scattered light intensity is set as follows: The weighting coefficient for the surface roughness increase rate is ,and The cleanliness of a zone is calculated using the following formula: ； In the formula, The cleanliness level of the zone ranges from 0 to 1. The amplification rate of the scattered light intensity. This represents the rate of increase in surface roughness.

3. The method for detecting the cleanliness of a photovoltaic panel according to claim 1, characterized in that, The calculation process for the uniformity of light absorption in step 4 is as follows: The coefficient of variation of the measured local light absorption intensity data is calculated using the following formula: ； In the formula, The coefficient of variation is 1. The standard deviation of the measured data is . The coefficient of variation is normalized to the average value of the measured data to obtain a light absorption uniformity index with a value range of 0-1.

4. The method for detecting the cleanliness of a photovoltaic panel according to claim 1, characterized in that, The calibration process described in step 7 is as follows: Outliers are removed from the on-site cleaning status verification data, and valid data within a preset reasonable range are retained. Calculate the deviation between the valid data and the previous detection data of the corresponding independent detection zone. ; Based on deviation value The overall deviation correction factor K is calculated using the following formula: ； In the formula, This is the overall deviation correction factor. This is the sequence number of the valid data, and its value is... , For the first The deviation value of each valid data point. The number of valid data; Based on the overall deviation correction coefficient The quantitative value of the initial overall cleanliness assessment obtained in step 5 is corrected for error to obtain the quantitative value of overall cleanliness. The calibration formula is as follows: ； in, This is the initial assessment of the overall cleanliness.

5. A photovoltaic panel cleanliness detection system, characterized in that, The photovoltaic panel cleanliness testing method according to any one of claims 1-4 includes: The benchmark acquisition and partitioning module is used to acquire the surface scattered light intensity benchmark value and surface roughness benchmark value of the photovoltaic panel under standard clean condition, and divide the effective light-receiving surface of the photovoltaic panel under test into several independent detection partitions according to a preset grid. The partition detection and cleanliness calculation module is used to collect real-time scattered light intensity data and real-time surface roughness data of each independent detection partition. By calculating the increase rate of real-time scattered light intensity data relative to the scattered light intensity benchmark value and the increase rate of real-time surface roughness data relative to the surface roughness benchmark value, and combining the two-dimensional collaborative weight calculation, the partition cleanliness of each independent detection partition is obtained. The cleanliness spatial feature extraction module is used to extract the spatial distribution features of the cleanliness of the photovoltaic panel based on the cleanliness of all independent detection zones. It extracts the gradient distribution features of the cleanliness of the zones through a gradient analysis algorithm, extracts the cluster distribution features of the high-pollution zones through a density clustering algorithm, and extracts the connectivity distribution features of the clean zones through a neighborhood connectivity algorithm. The gradient distribution features, cluster distribution features, and connectivity distribution features constitute the cleanliness spatial distribution features. The light absorption detection and uniformity calculation module is used to collect measured local light absorption intensity data of each independent detection zone of the photovoltaic panel, and calculate the light absorption uniformity of the photovoltaic panel based on the measured local light absorption intensity data. The cleanliness correlation determination and preliminary judgment module is used to determine the correlation between the spatial distribution characteristics of cleanliness and the uniformity of light absorption to obtain the preliminary judgment result of the overall cleanliness of the photovoltaic panel. First, based on the historical data of the cleanliness detection of the photovoltaic panel and the photoelectric conversion characteristics of the photovoltaic panel, a rule library for the correlation between the spatial distribution characteristics of cleanliness and the uniformity of light absorption is established. The rule library presets the threshold values ​​of the spatial distribution characteristics of cleanliness and the threshold values ​​of the uniformity of light absorption index corresponding to different cleanliness levels. Each cleanliness level corresponds to a unique set of characteristic threshold values ​​and uniformity threshold values. Retrieve the cleanliness spatial distribution features extracted in step 3 and the light absorption uniformity index calculated in step 4. Match the cleanliness spatial distribution features with the feature thresholds of each level in the rule base one by one. Compare the light absorption uniformity index with the uniformity thresholds of each level in the rule base. Calculate the matching confidence based on the matching and comparison results. When the calculated confidence level is higher than the preset confidence threshold, the corresponding preliminary judgment result of overall cleanliness is directly output, including the overall cleanliness quantification value and cleanliness level. When the confidence level is lower than the preset confidence threshold, the matching threshold in the rule base is re-optimized, and the matching judgment is performed again until the preliminary judgment result of overall cleanliness that meets the requirements is obtained. The on-site verification area selection and data acquisition module is used to select the pollution characteristic area and the cleanliness benchmark area of ​​the photovoltaic panel, collect on-site cleanliness status verification data of the pollution characteristic area and the cleanliness benchmark area, and preset a first threshold. Second threshold ,and The cleanliness of the partition is lower than The continuous and independent detection zones form a pollution characteristic area, and the cleanliness of the zone is higher than that of the other zones. The continuous independent detection zones constitute the clean reference area, and the first threshold Second threshold It can be dynamically adjusted according to the environmental parameters of the photovoltaic panel. The cleanliness result calibration and detection module is used to calibrate the overall cleanliness preliminary judgment result by combining the on-site cleanliness status verification data, and obtain the overall cleanliness detection result of the photovoltaic panel.

Images