Method and system for cleaning accumulated dust on photovoltaic panel

Abstract

The invention relates to a photovoltaic panel accumulated dust cleaning method and system, and the method comprises the steps: S1, obtaining a surface image of a photovoltaic panel, converting the surface image of the photovoltaic panel from an RGB color space into a CIE Lab color space, and carrying out the preprocessing; s2, calculating and obtaining the color difference of the photovoltaic panel through a color difference formula by using the preprocessed surface image of the photovoltaic panel and preset reference data, comparing the color difference with a preset color difference threshold value, if the color difference exceeds the threshold value, judging that the photovoltaic panel needs to be cleaned, otherwise, judging that the photovoltaic panel does not need to be cleaned, and returning to the step S1; s3, acquiring the dust deposition density of the surface of the current photovoltaic panel based on the trained dust deposition density-chromatic aberration relation model by utilizing the chromatic aberration of the photovoltaic panel; and S4, distributing resources required by accumulated dust cleaning based on the accumulated dust density, and performing accumulated dust cleaning on the photovoltaic panel. Compared with the prior art, the method has the advantages that resources required by accumulated dust cleaning are allocated according to the accumulated dust density, and cleaning resources can be saved to the greatest extent.

Description

Technical Field

[0001] The present invention relates to the technical field of solar photovoltaic panel cleaning, and in particular to a method and system for cleaning dust accumulated on photovoltaic panels. Background Art

[0002] The cleanliness of photovoltaic panels significantly impacts their power generation efficiency and lifespan. With the rapid development of photovoltaic power generation, the inspection and maintenance of panel cleanliness has become increasingly important. Traditionally, panel cleanliness inspection relies primarily on visual inspection or simple tool detection for surface contamination. However, with the increasing number and widespread distribution of photovoltaic panels, traditional inspection methods have become inefficient, labor-intensive, and inaccurate. Consequently, methods for inspecting photovoltaic panel cleanliness using machine vision and drone technology have emerged. Currently, cleaning robots use rotating brushes with water sprays and sun trackers to continuously clean solar panels, improving their efficiency. However, these robots have not yet integrated real-time monitoring of dust accumulation with cleaning methods, and automated cleaning often results in excessive and wasteful use of water resources.

[0003] Solar panel cleaning robots are currently widely used in various photovoltaic power station scenarios. These robots, equipped with intelligent sensing and autonomous navigation capabilities, follow a pre-defined, clear route, avoiding duplicate cleaning or missed areas. However, research has found that each photovoltaic panel has varying levels of dust accumulation, leading to wasted water and cleaning resources on many dust-free panels. Furthermore, in some areas, there is a lack of human oversight for cleaning robots. When it rains, scheduled cleaning robots begin operating, resulting in wasted resources. Furthermore, when dust levels do not coincide with the robot's scheduled cleaning schedule, the robot cannot clean promptly, impacting power generation efficiency.

[0004] Chinese patent CN117895899A addresses the issue that current cleaning technologies fail to account for the impact of environmental and weather changes, as well as panel surface reflectivity, on panel cleanliness inspection using machine vision. The patent proposes a photovoltaic panel cleanliness inspection method and system. By analyzing weather changes and panel surface reflectivity, the system further reduces the significant errors that can occur when using machine vision for panel cleanliness inspection. However, the patent still fails to address the issue of excessive water consumption during normal automated cleaning on sunny days.

[0005] Therefore, it is necessary to propose a method for cleaning dust accumulation on photovoltaic panels that can save water resources. Summary of the Invention

[0006] The purpose of the present invention is to provide a method and system for cleaning dust accumulation on photovoltaic panels in order to overcome the defects of the prior art.

[0007] The purpose of the present invention can be achieved by the following technical solutions:

[0008] A method for cleaning dust accumulation on a photovoltaic panel, the method comprising:

[0009] Step S1, obtaining a photovoltaic panel surface image, converting the photovoltaic panel surface image from an RGB color space to a CIE Lab color space, and performing preprocessing;

[0010] Step S2: Using the pre-processed photovoltaic panel surface image and preset reference data, the color difference of the photovoltaic panel is calculated using a color difference formula and compared with a preset color difference threshold. If the color difference exceeds the threshold, it is determined that the photovoltaic panel needs to be cleaned; otherwise, it is determined that cleaning is not required, and the process returns to step S1.

[0011] Step S3, using the photovoltaic panel color difference and based on the trained dust density-color difference relationship model, obtaining the dust density on the current photovoltaic panel surface;

[0012] Step S4: Allocate resources required for dust cleaning based on the dust density to clean the photovoltaic panels.

[0013] Furthermore, the pre-processing process includes: noise reduction, image enhancement, dynamic illumination compensation, perspective distortion correction and solar reflection suppression.

[0014] Furthermore, the color difference formula is:

[0015]

[0016] Among them, ΔL′, ΔC′ ab and ΔH′ ab Represents the difference of brightness, chroma and hue respectively, k L 、k C and k H is the preset coefficient, S L 、S C and S H is the weighting function, R T is the hue rotation function.

[0017] Furthermore, in step S2, if the current weather is rainy, it is determined that cleaning is not required.

[0018] Furthermore, the training process of the dust density-color difference relationship model includes:

[0019] Acquire images of photovoltaic panel surfaces with different degrees of dust accumulation;

[0020] The collected images of the dusty photovoltaic panel surfaces are converted into the CIE Lab color space, and the color difference between the dusty photovoltaic panel surface images and the clean photovoltaic panel images is calculated using the color difference formula;

[0021] Based on the surface images of photovoltaic panels with different dust accumulation levels, the brightness information of photovoltaic panels with different dust accumulation levels is extracted through image analysis algorithms, and the extinction coefficient of photovoltaic panels with different dust accumulation levels is calculated in combination with optical principles;

[0022] Based on the extinction coefficients of the photovoltaic panels with different dust accumulation levels, the dust accumulation density of the photovoltaic panels with different dust accumulation levels is calculated using the extinction coefficient formula;

[0023] Establishing a mapping based on the color difference between the surface images of each dusty photovoltaic panel and the clean photovoltaic panel image and the dust density quality inspection of photovoltaic panels with different dust accumulation levels, and obtaining a training data set after sorting;

[0024] Establishing a dust accumulation density-color difference relationship model, and using a training data set to train the dust accumulation density-color difference relationship model based on a mechanical algorithm;

[0025] After the training is completed, the dust accumulation density-color difference relationship model is evaluated. If the evaluation is qualified, the trained dust accumulation density-color difference relationship model is output; otherwise, the training steps are repeated.

[0026] Furthermore, the extinction coefficient formula is:

[0027]

[0028] Where K is the extinction coefficient, ρ is the dust particle density, q is the extinction efficiency of a single dust particle, d p is the diameter of a single dust particle.

[0029] Furthermore, the process of allocating resources required for dust cleaning based on the dust density includes:

[0030] The dust accumulation density is compared with the preset light, medium and heavy dust accumulation density thresholds. If the dust accumulation density is less than the light dust accumulation density threshold, it is determined that cleaning is not required and the cleaning is ended. If the dust accumulation density is greater than the light dust accumulation density threshold but less than the medium dust accumulation density threshold, it is determined to use a brush for cleaning. If the dust accumulation density is greater than the medium dust accumulation density threshold but less than the heavy dust accumulation density threshold, it is determined to use water for cleaning. If the dust accumulation density is greater than the heavy dust accumulation density threshold, it is determined to use a water-brush combination cleaning.

[0031] Furthermore, after the dust on the photovoltaic panel is cleaned, the surface image of the photovoltaic panel is obtained again, the color difference of the photovoltaic panel is obtained and compared with the preset color difference threshold. If it exceeds the threshold, it is determined that it still needs to be cleaned, and the process returns to step S3. Otherwise, it is determined that the cleaning is completed, and the inspection and cleaning process of the next photovoltaic panel is entered.

[0032] A photovoltaic panel dust cleaning system, the system comprising:

[0033] Perception module: obtains the current photovoltaic panel surface image, converts the photovoltaic panel surface image from RGB color space to CIE Lab color space and performs preprocessing;

[0034] Control module: Utilizes the pre-processed photovoltaic panel surface image and preset reference data, calculates the photovoltaic panel color difference through a color difference formula, and compares it with a preset color difference threshold. If the threshold is exceeded, the photovoltaic panel is determined to need cleaning; otherwise, it is determined not to need cleaning. If cleaning is determined to be necessary, the photovoltaic panel color difference is used to obtain the current dust density on the photovoltaic panel surface based on a trained dust density-color difference relationship model, and the resources required for dust cleaning are allocated based on the dust density.

[0035] Cleaning module: performs dust cleaning on the photovoltaic panels according to the resources required for dust cleaning allocated by the control module.

[0036] Furthermore, the system further includes a moving module. When the control module determines that the current photovoltaic panel does not need to be cleaned, the moving module automatically moves the photovoltaic panel dust cleaning system to the next photovoltaic panel.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] 1. This method converts the photovoltaic panel surface image into the CIE Lab color space, combines it with a preprocessing step, and then uses a color difference formula to calculate the color difference and compare it with a preset threshold. This method can scientifically and accurately determine whether the photovoltaic panel needs cleaning. This avoids the errors and arbitrariness of subjective judgment, improves the accuracy and reliability of the judgment, and allocates the required dust cleaning resources according to the dust density, thus saving cleaning resources to the greatest extent.

[0039] 2. The present invention takes into account the special situation of rainy days and directly determines that cleaning is not necessary, making the judgment logic more reasonable and complete;

[0040] 3. The proposed dust density-color difference relationship model is rigorously developed. By acquiring images of photovoltaic panels with varying degrees of dust accumulation, calculating color difference, extracting brightness information, and applying optical principles to calculate the extinction coefficient, the dust density is derived. A mapping between color difference and dust density is then established to form a training dataset for model training. This model can accurately determine the dust density on the photovoltaic panel surface based on color difference, providing a quantitative basis for subsequent cleaning work.

[0041] 4. This invention compares dust accumulation density with preset light, medium, and heavy dust density thresholds to allocate different cleaning resources, such as brush cleaning, water cleaning, and water-brush combination cleaning. This makes cleaning more reasonable and efficient, avoids waste of resources, and can also adopt appropriate cleaning methods for different dust accumulation levels, improving cleaning results.

[0042] 5. After cleaning the photovoltaic panels, the present invention re-acquires the image, calculates the color difference, and compares it with a preset threshold to determine whether further cleaning is required. This feedback mechanism can ensure the cleaning quality of the photovoltaic panels and keep them in good working condition at all times.

[0043] 6. The photovoltaic panel dust cleaning system of the present invention has clear divisions of labor among its sensing module, control module, cleaning module, and movement module. The sensing module is responsible for acquiring and preprocessing images; the control module performs cleaning determinations, calculates dust density, and allocates resources; the cleaning module performs cleaning tasks; and the movement module automatically moves to the next photovoltaic panel when cleaning is no longer necessary. The coordinated operation of these modules enhances the automation and intelligence of the system and ensures the orderly execution of cleaning tasks.

[0044] 7. The method and system of the present invention comprehensively consider multiple factors, such as light and weather, and can accurately determine the dust accumulation of photovoltaic panels and perform appropriate cleaning under different environmental conditions. It has strong environmental adaptability and can play a role in different application scenarios, ensuring the normal operation and power generation efficiency of photovoltaic panels.

[0045] 8. The modular design of the present invention supports flexible adjustment of cleaning strategies. The model can be continuously iterated and upgraded as data accumulates, which has good technical scalability.

[0046] 9. When judging whether cleaning is needed, the present invention first determines whether cleaning is needed by comparing the color difference with the threshold. When entering resource allocation, it again determines whether cleaning is needed by comparing the dust density with the threshold. When the dust density is lower than the threshold, cleaning is stopped even if the color difference determines that cleaning is needed. The nesting of double judgments further prevents possible misjudgments of single color difference judgments and prevents waste of cleaning resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 Flow chart of the method of the present invention;

[0048] Figure 2 This is a schematic block diagram of the system of the present invention. DETAILED DESCRIPTION

[0049] 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 part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0050] Example 1

[0051] This embodiment discloses a method for cleaning dust accumulation on photovoltaic panels. Figure 1 As shown, including:

[0052] S1, obtaining a photovoltaic panel surface image, converting the photovoltaic panel surface image from RGB color space to CIE Lab color space and performing preprocessing;

[0053] S2, using the pre-processed photovoltaic panel surface image and preset reference data, calculate the photovoltaic panel color difference through a color difference formula, and compare it with a preset color difference threshold. If it exceeds the threshold, it is determined that the photovoltaic panel needs to be cleaned; otherwise, it is determined that it does not need to be cleaned, and the process returns to step S1;

[0054] S3, using the photovoltaic panel color difference and based on the trained dust density-color difference relationship model, obtains the dust density on the current photovoltaic panel surface;

[0055] S4, allocating resources required for dust cleaning based on dust density, and performing dust cleaning on the photovoltaic panels.

[0056] In step S1, a high-resolution camera is first used to photograph the surface of the photovoltaic panel to obtain an RGB image of the surface of the photovoltaic panel. During shooting, the camera resolution is set to 4K (3840×2160 pixels) to ensure that sufficient image details are captured. After shooting is completed, the acquired RGB image is converted into the CIE Lab color space through a color space conversion algorithm. The CIE Lab color space is a device-independent color space, in which L represents brightness, a represents the range from green to red, and b represents the range from blue to yellow. This color space is closer to the human eye's perception of color, which is conducive to subsequent color difference analysis.

[0057] If this method is used on a cleaning robot, the above operations are performed based on the RGB camera provided on the robot.

[0058] The preprocessing process includes: noise reduction, image enhancement, dynamic illumination compensation, perspective distortion correction and solar reflection suppression.

[0059] In this embodiment, the noise reduction process uses a Gaussian filter algorithm, the filter kernel size is set to 5×5, and the standard deviation σ=1.5, which effectively removes random noise in the image;

[0060] Image enhancement uses adaptive histogram equalization technology to divide the image into 8×8 grids, perform histogram equalization on each grid, and then combine the results through bilinear interpolation to improve the contrast and clarity of the image;

[0061] Dynamic illumination compensation analyzes the brightness distribution of an image and establishes an illumination model to compensate for uneven illumination conditions. This is achieved by dividing the image into multiple regions, calculating the average brightness value of each region, and then adjusting the brightness of each region based on the global brightness target value (set to 128, the middle value of an 8-bit image).

[0062] Perspective distortion correction detects the four corner points of the photovoltaic panel edge and establishes a perspective transformation matrix to convert the tilted photovoltaic panel image into a front view image. The transformation matrix is calculated based on the four-point correspondence relationship and the least squares method is used to solve the transformation parameters.

[0063] Solar reflection suppression detects high-brightness areas in the image (pixels with brightness values greater than 230) and applies a local brightness compression algorithm to map the brightness values of these areas to a reasonable range (180-220), reducing the interference of direct sunlight or reflection on image analysis.

[0064] In step S2, preset benchmark data is first retrieved from a database. This benchmark data is a CIE Lab color space image of a clean photovoltaic panel under standard lighting conditions. The preprocessed photovoltaic panel surface image is then compared with the preset benchmark data, and the color difference of the photovoltaic panel is calculated using a color difference formula.

[0065] The color difference formula is:

[0066]

[0067] Among them, ΔL′, ΔC′ ab and ΔH′ ab Represents the difference of brightness, chroma and hue respectively, k L 、k C and k H is the preset coefficient, S L 、S C and S H is the weighting function, R T is the hue rotation function.

[0068] In this embodiment, the calculated color difference value of the photovoltaic panel is compared with a preset color difference threshold, and the preset color difference threshold is set to 5.0. If the color difference value exceeds 5.0, it is determined that the photovoltaic panel needs to be cleaned; if the color difference value does not exceed 5.0, it is determined that the photovoltaic panel does not need to be cleaned, and the process returns to step S1 to continue monitoring the next photovoltaic panel or perform the next round of monitoring on the current photovoltaic panel.

[0069] Furthermore, during step S2, the system checks the current weather conditions. If it is raining (real-time weather information is obtained through the meteorological data interface), cleaning is determined not to be necessary, regardless of the color difference. This is because cleaning on rainy days is not only inefficient but may also damage cleaning equipment. Rainwater itself also has a certain cleaning effect.

[0070] In step S3, the training process of the dust density-color difference relationship model includes:

[0071] Acquire images of photovoltaic panel surfaces with different degrees of dust accumulation;

[0072] The collected images of the dusty photovoltaic panel surfaces are converted into the CIE Lab color space, and the color difference between the dusty photovoltaic panel surface images and the clean photovoltaic panel images is calculated using the color difference formula;

[0073] Based on the surface images of photovoltaic panels with different dust accumulation levels, the brightness information of photovoltaic panels with different dust accumulation levels is extracted through image analysis algorithms, and the extinction coefficient of photovoltaic panels with different dust accumulation levels is calculated in combination with optical principles;

[0074] Based on the extinction coefficient of photovoltaic panels with different dust accumulation levels, the dust accumulation density of photovoltaic panels with different dust accumulation levels is calculated using the extinction coefficient formula;

[0075] A mapping is established based on the color difference between the surface images of each dusty photovoltaic panel and the clean photovoltaic panel image, as well as the dust density quality inspection of photovoltaic panels with different dust accumulation levels, and the training data set is obtained after sorting.

[0076] Establish a dust density-color difference relationship model, and use the training data set to train the dust density-color difference relationship model based on the mechanical algorithm;

[0077] After the training is completed, the dust accumulation density-color difference relationship model is evaluated. If the evaluation is qualified, the trained dust accumulation density-color difference relationship model is output, otherwise the training steps are repeated.

[0078] The extinction coefficient formula is:

[0079]

[0080] Where K is the extinction coefficient, ρ is the dust particle density, q is the extinction efficiency of a single dust particle, d p is the diameter of a single dust particle.

[0081] In this embodiment, surface images of photovoltaic panels with different degrees of dust accumulation are obtained, including images of completely clean photovoltaic panels, images of photovoltaic panels with light dust accumulation, images of photovoltaic panels with moderate dust accumulation, and images of photovoltaic panels with heavy dust accumulation, totaling 100 sets of sample data, covering different lighting conditions, different dust distributions, and different dust accumulation types.

[0082] In this example, a support vector regression (SVR) algorithm was used to train the dust density-color difference relationship model based on a machine learning algorithm. The radial basis function (RBF) kernel function was selected, the penalty parameter C was set to 10, and the kernel parameter γ was set to 0.1. During the training process, the dataset was divided into a training set and a validation set in a ratio of 8:2.

[0083] In step S4, the process of allocating resources required for dust cleaning based on dust density includes:

[0084] The dust accumulation density is compared with the preset light, medium and heavy dust accumulation density thresholds. If the dust accumulation density is less than the light dust accumulation density threshold, it is determined that cleaning is not required and the cleaning is ended. If the dust accumulation density is greater than the light dust accumulation density threshold but less than the medium dust accumulation density threshold, it is determined to use a brush for cleaning. If the dust accumulation density is greater than the medium dust accumulation density threshold but less than the heavy dust accumulation density threshold, it is determined to use water for cleaning. If the dust accumulation density is greater than the heavy dust accumulation density threshold, it is determined to use a water-brush combination cleaning.

[0085] In this embodiment, the preset light dust density threshold is 2.0g / m 2 The moderate dust density threshold is 5.0g / m 2 The threshold value of heavy dust accumulation density is 8.0g / m 2 .

[0086] If the dust accumulation density is less than the light dust accumulation density threshold, it is determined that cleaning is not required and the cleaning process ends;

[0087] If the dust density is greater than the light dust density threshold but less than the moderate dust density threshold, a brush cleaning is performed. Use a soft nylon brush with a speed of 60-80 rpm and a pressure of 100-150 Pa for one cleaning.

[0088] If the dust density is greater than the moderate dust density threshold but less than the heavy dust density threshold, water cleaning is used. Use deionized water with a pressure of 0.2-0.3 MPa and a distance of 20-30 cm between the nozzle and the photovoltaic panel surface for one cleaning.

[0089] If the dust density exceeds the heavy dust density threshold, a water-brush combination cleaning method is used. This method involves spraying the dust with deionized water (at a pressure of 0.3-0.4 MPa) to moisten the dust. After allowing the dust to sit for 1-2 minutes, the dust is cleaned with a soft nylon brush (at a speed of 80-100 rpm and a pressure of 150-200 Pa). Finally, the dust is rinsed with deionized water (at a pressure of 0.2-0.3 MPa).

[0090] In step S4, after the dust on the photovoltaic panel is cleaned, the surface image of the photovoltaic panel is obtained again, the color difference of the photovoltaic panel is obtained and compared with the preset color difference threshold. If it exceeds the threshold, it is determined that it still needs to be cleaned, and the process returns to step S3. If it does not exceed the threshold, it is determined that the cleaning is completed, and the inspection and cleaning process of the next photovoltaic panel is entered.

[0091] Example 2

[0092] This embodiment discloses a photovoltaic panel dust cleaning system. Figure 2 As shown, specifically including:

[0093] Perception module 1: obtains the current photovoltaic panel surface image, converts the photovoltaic panel surface image from RGB color space to CIE Lab color space and performs preprocessing;

[0094] Control Module 2: Utilizes the pre-processed PV panel surface image and preset reference data, calculates the PV panel color difference using a color difference formula, and compares it with a preset color difference threshold. If the threshold is exceeded, the PV panel is determined to require cleaning; otherwise, it is determined not to require cleaning. If cleaning is determined to be necessary, the PV panel color difference is used to obtain the current dust density on the PV panel surface based on a trained dust density-color difference relationship model, and resources required for dust cleaning are allocated based on the dust density.

[0095] Cleaning module 3: performs dust cleaning on the photovoltaic panels according to the resources required for dust cleaning allocated by the control module.

[0096] Moving module 4: When the control module 2 determines that the current photovoltaic panel does not need to be cleaned, the entire photovoltaic panel dust cleaning system is automatically moved to the next photovoltaic panel.

[0097] The specific details of the above modules can be understood by referring to the relevant descriptions and effects in Example 1.

[0098] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and such modifications or substitutions are intended to be within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. A method for cleaning dust accumulation on photovoltaic panels, characterized in that: The method comprises: Step S1, obtaining a photovoltaic panel surface image, and converting the photovoltaic panel surface image from an RGB color space to a CIELab color space and performing preprocessing; Step S2: Using the pre-processed photovoltaic panel surface image and preset reference data, the color difference of the photovoltaic panel is calculated using a color difference formula and compared with a preset color difference threshold. If the color difference exceeds the threshold, it is determined that the photovoltaic panel needs to be cleaned; otherwise, it is determined that cleaning is not required, and the process returns to step S1. Step S3, using the photovoltaic panel color difference and based on the trained dust density-color difference relationship model, obtaining the dust density on the current photovoltaic panel surface; Step S4: Allocate resources required for dust cleaning based on the dust density to clean the photovoltaic panels.

2. A photovoltaic panel dust cleaning method according to claim 1, characterized in that: The preprocessing process includes: noise reduction, image enhancement, dynamic illumination compensation, perspective distortion correction and sun reflection suppression.

3. A photovoltaic panel dust cleaning method according to claim 1, characterized in that: The color difference formula is: Among them, ΔL′, ΔC′ ab and ΔH′ ab Represents the difference of brightness, chroma and hue respectively, k L 、k C and k H is the preset coefficient, S L 、S C and S H is the weighting function, R T is the hue rotation function.

4. A photovoltaic panel dust cleaning method according to claim 1, characterized in that: In step S2, if the current weather is rainy, it is determined that cleaning is not required.

5. A photovoltaic panel dust cleaning method according to claim 1, characterized in that: The training process of the dust density-color difference relationship model includes: Acquire images of photovoltaic panel surfaces with different degrees of dust accumulation; The collected images of the dusty photovoltaic panel surfaces are converted into the CIE Lab color space, and the color difference between the dusty photovoltaic panel surface images and the clean photovoltaic panel images is calculated using the color difference formula; Based on the surface images of photovoltaic panels with different dust accumulation levels, the brightness information of photovoltaic panels with different dust accumulation levels is extracted through image analysis algorithms, and the extinction coefficient of photovoltaic panels with different dust accumulation levels is calculated in combination with optical principles; Based on the extinction coefficients of the photovoltaic panels with different dust accumulation levels, the dust accumulation density of the photovoltaic panels with different dust accumulation levels is calculated using the extinction coefficient formula; Establishing a mapping based on the color difference between the surface images of each dusty photovoltaic panel and the clean photovoltaic panel image and the dust density quality inspection of photovoltaic panels with different dust accumulation levels, and obtaining a training data set after sorting; Establishing a dust accumulation density-color difference relationship model, and using a training data set to train the dust accumulation density-color difference relationship model based on a mechanical algorithm; After the training is completed, the dust accumulation density-color difference relationship model is evaluated. If the evaluation is qualified, the trained dust accumulation density-color difference relationship model is output; otherwise, the training steps are repeated.

6. A photovoltaic panel dust cleaning method according to claim 5, characterized in that: The extinction coefficient formula is: Where K is the extinction coefficient, ρ is the dust particle density, q is the extinction efficiency of a single dust particle, d p is the diameter of a single dust particle.

7. A photovoltaic panel dust cleaning method according to claim 1, characterized in that: The process of allocating resources required for dust cleaning based on the dust density includes: The dust accumulation density is compared with the preset light, medium and heavy dust accumulation density thresholds. If the dust accumulation density is less than the light dust accumulation density threshold, it is determined that cleaning is not required and the cleaning is ended. If the dust accumulation density is greater than the light dust accumulation density threshold but less than the medium dust accumulation density threshold, it is determined to use a brush for cleaning. If the dust accumulation density is greater than the medium dust accumulation density threshold but less than the heavy dust accumulation density threshold, it is determined to use water for cleaning. If the dust accumulation density is greater than the heavy dust accumulation density threshold, it is determined to use a water-brush combination cleaning.

8. A photovoltaic panel dust cleaning method according to claim 1, characterized in that: After the dust accumulation on the photovoltaic panel is cleaned, the surface image of the photovoltaic panel is obtained again, the color difference of the photovoltaic panel is obtained and compared with the preset color difference threshold. If it exceeds the threshold, it is determined that it still needs to be cleaned and the process returns to step S3. Otherwise, the cleaning is determined to be completed and the inspection and cleaning process of the next photovoltaic panel is entered.

9. A photovoltaic panel dust cleaning system, characterized in that: The system comprises: Perception module: obtains the current photovoltaic panel surface image, converts the photovoltaic panel surface image from RGB color space to CIE Lab color space and performs preprocessing; Control module: Utilizes the pre-processed photovoltaic panel surface image and preset reference data, calculates the photovoltaic panel color difference through a color difference formula, and compares it with a preset color difference threshold. If the threshold is exceeded, the photovoltaic panel is determined to need cleaning; otherwise, it is determined not to need cleaning. If cleaning is determined to be necessary, the photovoltaic panel color difference is used to obtain the current dust density on the photovoltaic panel surface based on a trained dust density-color difference relationship model, and the resources required for dust cleaning are allocated based on the dust density. Cleaning module: performs dust cleaning on the photovoltaic panels according to the resources required for dust cleaning allocated by the control module.

10. A photovoltaic panel dust cleaning system according to claim 9, characterized in that: The system further includes a moving module, which automatically moves the photovoltaic panel dust cleaning system to the next photovoltaic panel when the control module determines that the current photovoltaic panel does not need to be cleaned.

Images