Self-cleaning method and device for a fishery-solar complementary photovoltaic module

Abstract

The present invention discloses a self-cleaning method and device for a fishery-solar complementary photovoltaic module, relating to the technical field of cleaning. The method includes: obtaining ambient light intensity information and ambient light illumination angle through a light sensor. Obtaining the image acquisition control parameters of an image acquisition device for the photovoltaic module. Performing image analysis according to the image acquisition control parameters, the ambient light intensity information, and the ambient light illumination angle to generate a reference image of the solar panel surface. Obtaining a real-time image of the solar panel surface of the image acquisition device. Analyzing the real-time image of the solar panel surface and the reference image of the solar panel surface through a twin comparison network to generate a target cleaning area. Based on the target cleaning area, controlling a self-cleaning nozzle to perform cleaning. The technical problem that the cleaning intelligence of the photovoltaic module cleaning method in the prior art is low, resulting in difficult improvement of the cleaning efficiency and cleaning effect, is solved.

Description

Technical Field

[0001] This application relates to the technical field of cleaning, and particularly to a self-cleaning method and device for a photovoltaic module with complementary fishing and photovoltaic power generation. Background Art

[0002] With the development of photovoltaic power generation technology, photovoltaic modules are increasingly widely used. Among them, photovoltaic fishery, as a new model of the combination of agriculture and energy, is gradually attracting attention. Photovoltaic fishery can not only effectively utilize water surface resources for photovoltaic power generation, but also provide a suitable growth environment for aquaculture. However, the surface of photovoltaic modules is easily affected by pollutants such as dust, bird droppings, and aquatic organisms, resulting in a decrease in power generation efficiency. Most of the existing cleaning methods rely on manual labor or simple mechanical equipment, with low efficiency and difficulty in achieving precise cleaning.

[0003] Therefore, in the prior art, there are technical problems in the cleaning method of photovoltaic modules, such as low cleaning intelligence, which makes it difficult to improve the cleaning efficiency and cleaning effect. Summary of the Invention

[0004] This application provides a self-cleaning method and device for a photovoltaic module with complementary fishing and photovoltaic power generation, which solves the technical problems in the prior art that the cleaning method of photovoltaic modules has low cleaning intelligence, resulting in difficult improvement of cleaning efficiency and cleaning effect. It achieves the technical effects of greatly improving the cleaning efficiency and effect of photovoltaic modules in fishery applications, reducing labor costs, extending the service life of photovoltaic modules, improving the overall power generation efficiency, and ensuring the stability of the aquaculture environment.

[0005] This application provides a self-cleaning method for a photovoltaic module with complementary fishing and photovoltaic power generation. The method includes: obtaining ambient light intensity information and ambient light illumination angle through a light sensor, where the ambient light illumination angle refers to the included angle between the line connecting the sun position and the image acquisition device and the solar panel. Obtaining the image acquisition control parameters of the image acquisition device for the photovoltaic module. Performing image analysis based on the image acquisition control parameters, the ambient light intensity information, and the ambient light illumination angle to generate a reference image of the solar panel surface. Obtaining a real-time image of the solar panel surface by the image acquisition device. Analyzing the real-time image of the solar panel surface and the reference image of the solar panel surface through a twin comparison network to generate a target cleaning area. Controlling a self-cleaning nozzle to perform cleaning based on the target cleaning area.

[0006] In an implementation manner, image analysis is performed according to the image acquisition control parameters and the ambient light intensity information to generate a reference image of the solar panel surface, including: configuring reference acquisition control parameters, reference ambient light intensity, reference illumination angle, and reference solar panel surface image, where the reference solar panel surface image has a reference color eigenvalue distribution matrix; collecting an image acquisition control parameter record data set, an ambient light intensity record data set, an illumination angle record data set, and a solar panel surface image record data set, where the solar panel surface image record data set has a recorded color eigenvalue distribution matrix set; according to the reference acquisition control parameters, the reference ambient light intensity, the reference illumination angle, the reference color eigenvalue distribution matrix, combining the image acquisition control parameter record data set, the ambient light intensity record data set, the illumination angle record data set, and the recorded color eigenvalue distribution matrix set, training a color distribution deviation vector matrix analysis channel; according to the reference color eigenvalue distribution matrix, the recorded color eigenvalue distribution matrix set, the reference solar panel surface image, and the solar panel surface image record data set, training an image generation channel; merging the color distribution deviation vector matrix analysis channel and the image generation channel to generate an image analysis model, and performing image analysis according to the image acquisition control parameters and the ambient light intensity information to generate the reference image of the solar panel surface.

[0007] In an implementation manner, according to the reference acquisition control parameters, the reference ambient light intensity, the reference illumination angle, the reference color eigenvalue distribution matrix, combining the image acquisition control parameter record data set, the ambient light intensity record data set, the illumination angle record data set, and the recorded color eigenvalue distribution matrix set, training a color distribution deviation vector matrix analysis channel includes: based on the reference acquisition control parameters, traversing the acquisition control parameter record data set for deviation analysis to generate a control parameter deviation vector array set; based on the reference ambient light intensity, traversing the ambient light intensity record data set for deviation analysis to generate an ambient light intensity deviation vector set; based on the reference illumination angle, traversing the illumination angle record data set for deviation analysis to generate an illumination angle deviation vector set; based on the reference color eigenvalue distribution matrix, traversing the recorded color eigenvalue distribution matrix set for deviation analysis to generate a color eigenvalue distribution deviation vector matrix set; according to the control parameter deviation vector array set, the ambient light intensity deviation vector set, the illumination angle deviation vector set, and the color eigenvalue distribution deviation vector matrix set, training a color distribution deviation vector matrix analysis channel.

[0008] In an implementation manner, according to the reference color feature value distribution matrix, the recorded color feature value distribution matrix set, the reference solar panel surface image, and the solar panel surface image recording data set, training the image generation channel includes: training the image generation channel according to the color feature value distribution deviation vector matrix set, the reference solar panel surface image, and the solar panel surface image recording data set.

[0009] In an implementation manner, analyzing the real-time image of the solar panel surface and the reference image of the solar panel surface through the siamese comparison network to generate a target cleaning area includes: extracting the first color feature distribution matrix of the real-time image of the solar panel surface through the first feature extraction channel of the siamese comparison network; extracting the second color feature distribution matrix of the reference image of the solar panel surface through the second feature extraction channel of the siamese comparison network, where the first color feature distribution matrix and the second color feature distribution matrix have coordinate consistency; parsing the first color feature distribution matrix and the second color feature distribution matrix through the feature comparison channel of the siamese comparison network to obtain the target cleaning area.

[0010] In an implementation manner, parsing the first color feature distribution matrix and the second color feature distribution matrix through the feature comparison channel of the siamese comparison network to obtain the target cleaning area includes: obtaining the first color feature and the second color feature of the first coordinate; when the color feature distance between the first color feature and the second color feature is greater than or equal to the color feature distance threshold, adding the first coordinate to the initial target cleaning area; after all coordinates are divided, clustering the target cleaning coordinates of the initial target cleaning area according to the distribution distance threshold to generate a set of aggregated areas, where the set of aggregated areas has a set of area labels; extracting the aggregated areas greater than or equal to the area threshold from the set of aggregated areas according to the set of area labels and setting them as the target cleaning area.

[0011] In an implementation manner, based on the target cleaning area, controlling the self-cleaning nozzle to perform cleaning includes: configuring the dust color feature; sorting from the target cleaning area based on the dust color feature to obtain the area ratio of the dust distribution area; when the area ratio of the dust distribution area is greater than or equal to the area ratio threshold, controlling the self-cleaning nozzle to perform cleaning based on the first impact force; when the area of the dust distribution area is less than the area ratio threshold, controlling the self-cleaning nozzle to perform cleaning based on the second impact force, where the second impact force is greater than the first impact force.

[0012] This application also provides a self-cleaning device for a fishery-solar complementary photovoltaic module, including:

[0013] A data acquisition module, configured to obtain ambient light intensity information and ambient light illumination angle through an optical sensor, where the ambient light illumination angle refers to the angle between the line connecting the sun position and the image acquisition device and the solar panel;

[0014] An acquisition control parameter obtaining module, configured to obtain the image acquisition control parameters of the image acquisition device of the photovoltaic module;

[0015] A reference image obtaining module, configured to perform image analysis according to the image acquisition control parameters, the ambient light intensity information, and the ambient light illumination angle to generate a reference image of the solar panel surface;

[0016] An image acquisition module, configured to obtain a real-time image of the solar panel surface of the image acquisition device;

[0017] A cleaning area obtaining module, configured to analyze the real-time image of the solar panel surface and the reference image of the solar panel surface through a twin comparison network to generate a target cleaning area;

[0018] A cleaning module, configured to control a self-cleaning nozzle to perform cleaning based on the target cleaning area.

[0019] It is intended to propose a self-cleaning method and device for a fishery-light complementary photovoltaic module through this application. By using an optical sensor, ambient light intensity information and ambient light illumination angle are obtained, where the ambient light illumination angle refers to the angle between the line connecting the sun position and the image acquisition device and the solar panel. The image acquisition control parameters of the image acquisition device of the photovoltaic module are obtained. Image analysis is performed according to the image acquisition control parameters, the ambient light intensity information, and the ambient light illumination angle to generate a reference image of the solar panel surface. A real-time image of the solar panel surface of the image acquisition device is obtained. The real-time image of the solar panel surface and the reference image of the solar panel surface are analyzed through a twin comparison network to generate a target cleaning area. Based on the target cleaning area, a self-cleaning nozzle is controlled to perform cleaning. The technical problem that the cleaning intelligence of the existing photovoltaic module cleaning method is relatively low, resulting in difficult improvement of cleaning efficiency and cleaning effect, is solved. The technical effects of greatly improving the cleaning efficiency and effect of the photovoltaic module in fishery applications, reducing labor costs, extending the service life of the photovoltaic module, improving the overall power generation efficiency, and ensuring the stability of the aquaculture environment are achieved. Description of the Drawings

[0020] To more clearly illustrate the technical solutions of the embodiments of the present disclosure, the accompanying drawings of the embodiments of the present disclosure will be briefly introduced below. Flowcharts are used in this application to illustrate the operations performed by the devices according to the embodiments of the present application. It should be understood that the operations above or below do not necessarily need to be executed precisely in order. On the contrary, according to needs, various steps can be executed in reverse order or simultaneously. At the same time, other operations can also be added to these processes, or one or several steps can be removed from these processes.

[0021] Figure 1 Schematic flow chart of a self-cleaning method for a fishery-solar complementary photovoltaic module provided by an embodiment of the present application;

[0022] Figure 2 Schematic structural diagram of a self-cleaning device for a fishery-solar complementary photovoltaic module provided by an embodiment of the present application.

[0023] Explanation of reference numerals: data acquisition module 11, acquisition control parameter acquisition module 12, reference image acquisition module 13, image acquisition module 14, cleaning area acquisition module 15, cleaning module 16. Detailed implementation manners

[0024] The above description is only an overview of the technical solutions of the present application. In order to be able to understand the technical means of the present application more clearly, it can be implemented according to the content of the description. And in order to make the above and other purposes, features and advantages of the present application more obvious and understandable, the following specifically gives the detailed implementation manners of the present application.

[0025] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limitations of the present application. All other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present application.

[0026] In the following description, "some embodiments" are involved, which describe a subset of all possible embodiments. However, it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict. The terms "first / second" involved are only used to distinguish similar objects and do not represent a specific order for the objects. The terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, device, product or server that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application.

[0027] An embodiment of the present application provides a self-cleaning method and device for a fishing-light complementary photovoltaic module, as Figure 1 shown, the method includes:

[0028] Obtain ambient light intensity information and ambient light illumination angle through a light sensor, where the ambient light illumination angle refers to the included angle between the line connecting the sun position and the image acquisition device and the solar panel;

[0029] Obtain the image acquisition control parameters of the image acquisition device of the photovoltaic module;

[0030] Perform image analysis according to the image acquisition control parameters, the ambient light intensity information and the ambient light illumination angle to generate a reference image of the solar panel surface;

[0031] With the development of photovoltaic power generation technology, the application of photovoltaic modules is becoming more and more extensive. Among them, photovoltaic fishery, as a new model of combining agriculture and energy, is gradually attracting attention. Photovoltaic fishery can not only effectively utilize water surface resources for photovoltaic power generation, but also provide a suitable growth environment for aquaculture. However, the surface of photovoltaic modules is easily affected by pollutants such as dust, bird droppings, and aquatic organisms, resulting in a decrease in power generation efficiency. Most of the existing cleaning methods rely on manual labor or simple mechanical equipment, with low efficiency and difficulty in achieving precise cleaning. To solve the problems existing in the prior art, through a light sensor, the light sensor can be arranged at the edge of the photovoltaic module or other parts that do not affect daylighting to ensure that it can accurately capture changes in ambient light. Obtain ambient light intensity information and ambient light angle, where the ambient light angle refers to the angle between the line connecting the sun position and the image acquisition device and the solar panel. Subsequently, obtain the image acquisition control parameters of the image acquisition device for the photovoltaic module. The image acquisition device, such as a camera, is installed around the photovoltaic module and is used to capture images of the surface of the solar panel in real time. The image acquisition control parameters of this device include resolution, shooting angle, exposure time, etc. These parameters will be dynamically adjusted according to the ambient light intensity information and the ambient light angle to ensure image quality and analysis accuracy. Further, based on the image acquisition control parameters, the ambient light intensity information, and the ambient light angle, perform image analysis to generate a reference image of the surface of the solar panel. The reference image of the surface of the solar panel is the image of the surface of the solar panel without pollutants in the same environment when the image acquisition device is performing real-time image acquisition.

[0032] The method provided by the embodiment of the present application further includes:

[0033] Configure reference acquisition control parameters, reference ambient light intensity, reference light angle, and reference solar panel surface image, where the reference solar panel surface image has a reference color feature value distribution matrix;

[0034] Collect an image acquisition control parameter record data set, an ambient light intensity record data set, a light angle record data set, and a solar panel surface image record data set, where the solar panel surface image record data set has a recorded color feature value distribution matrix set;

[0035] According to the reference acquisition control parameters, the reference ambient light intensity, the reference light angle, the reference color feature value distribution matrix, combined with the set of image acquisition control parameter record data sets, the ambient light intensity record data set, the light angle record data set, and the recorded color feature value distribution matrix set, train the color distribution deviation vector matrix analysis channel;

[0036] Train an image generation channel according to the reference color eigenvalue distribution matrix, the recorded color eigenvalue distribution matrix set, the reference solar panel surface image, and the solar panel surface image recording data set.

[0037] Combine the color distribution deviation vector matrix analysis channel and the image generation channel to generate an image analysis model, and perform image analysis according to the image acquisition control parameters and the ambient light intensity information to generate the reference image of the solar panel surface.

[0038] Performing image analysis according to the image acquisition control parameters and the ambient light intensity information to generate a reference image of the solar panel surface, including: configuring reference acquisition control parameters, reference ambient light intensity, reference illumination angle, and reference solar panel surface image, wherein the reference solar panel surface image has a reference color eigenvalue distribution matrix, and the reference data is the calibrated standard data for deviation analysis of the acquired data. Collect an image acquisition control parameter record data set, an ambient light intensity record data set, an illumination angle record data set, and a solar panel surface image record data set, wherein the solar panel surface image record data set has a recorded color eigenvalue distribution matrix set, and the recorded data is the data generated during the historical image acquisition process. Further, based on the deviation analysis result, train a color distribution deviation vector matrix analysis channel according to the reference acquisition control parameters, the reference ambient light intensity, the reference illumination angle, the reference color eigenvalue distribution matrix, in combination with the image acquisition control parameter record data set, the ambient light intensity record data set, the illumination angle record data set, and the recorded color eigenvalue distribution matrix set. The color distribution deviation vector matrix analysis channel is used to perform deviation analysis on the control parameters, ambient light intensity, illumination angle, solar panel surface image, and reference image obtained in real time, and finally obtain a color eigenvalue distribution deviation vector matrix. Train an image generation channel according to the reference color eigenvalue distribution matrix, the recorded color eigenvalue distribution matrix set, the reference solar panel surface image, and the solar panel surface image recording data set. The image generation channel is used to generate an image based on the color eigenvalue distribution deviation vector matrix and the reference solar panel surface image to obtain the reference image of the solar panel surface under the current ambient light intensity information, ambient illumination angle, and image acquisition control parameters. Finally, combine the color distribution deviation vector matrix analysis channel and the image generation channel to generate an image analysis model, and perform image analysis according to the image acquisition control parameters and the ambient light intensity information to generate the reference image of the solar panel surface.

[0039] The method provided in the embodiment of the present application further includes:

[0040] Traverse the acquisition control parameter record data set based on the reference acquisition control parameters for deviation analysis, and generate a control parameter deviation vector array set;

[0041] Traverse the ambient light intensity record data set based on the reference ambient light intensity for deviation analysis, and generate an ambient light intensity deviation vector set;

[0042] Traverse the illumination angle record data set based on the reference illumination angle for deviation analysis, and generate an illumination angle deviation vector set;

[0043] Traverse the recorded color eigenvalue distribution matrix set based on the reference color eigenvalue distribution matrix for deviation analysis, and generate a color eigenvalue distribution deviation vector matrix set;

[0044] Train a color distribution deviation vector matrix analysis channel according to the control parameter deviation vector array set, the ambient light intensity deviation vector set, the illumination angle deviation vector set, and the color eigenvalue distribution deviation vector matrix set.

[0045] When training the color distribution deviation vector matrix analysis channel, based on the reference acquisition control parameters, perform deviation analysis on the acquisition control parameter record data set to generate a control parameter deviation vector array set. Based on the reference ambient light intensity, perform deviation analysis on the ambient light intensity record data set to generate an ambient light intensity deviation vector set. Based on the reference illumination angle, perform deviation analysis on the illumination angle record data set to generate an illumination angle deviation vector set. Based on the reference color eigenvalue distribution matrix, perform deviation analysis on the recorded color eigenvalue distribution matrix set to generate a color eigenvalue distribution deviation vector matrix set. Finally, based on the obtained control parameter deviation vector array set, ambient light intensity deviation vector set, illumination angle deviation vector set, and color eigenvalue distribution deviation vector matrix set from the analysis, train the color distribution deviation vector matrix analysis channel. The color distribution deviation vector matrix analysis channel is constructed based on a neural network model, and the construction data is the control parameter deviation vector array set, the ambient light intensity deviation vector set, the illumination angle deviation vector set, and the color eigenvalue distribution deviation vector matrix set. Among them, the control parameter deviation vector array set, the ambient light intensity deviation vector set, and the illumination angle deviation vector set are input data, and the color eigenvalue distribution deviation vector matrix set is output data. Perform supervised training on the neural network model based on the construction data, obtain the result output by the model, and perform output accuracy judgment. When the output accuracy of the model meets the preset requirements, obtain the trained color distribution deviation vector matrix analysis channel.

[0046] The method provided by the embodiments of the present application further includes:

[0047] Train the image generation channel according to the color feature value distribution deviation vector matrix set, the reference solar panel surface image, and the solar panel surface image recording data set.

[0048] Construct the image generation channel according to the color feature value distribution deviation vector matrix set, the reference solar panel surface image, and the solar panel surface image recording data set. Among them, the construction of the image generation channel is based on the use of the generative adversarial network in deep learning. The color feature value distribution deviation vector matrix set and the reference solar panel surface image are used to construct the generator. The discriminator is constructed based on the simulated image generated by the generator and the solar panel surface image recording data set. The generator and the discriminator are improved through adversarial training, and finally a high-quality solar panel surface image is generated to obtain the constructed image generation channel.

[0049] Obtain the real-time image of the solar panel surface of the image acquisition device;

[0050] Analyze the real-time image of the solar panel surface and the reference image of the solar panel surface through the siamese comparison network to generate the target cleaning area;

[0051] Control the self-cleaning nozzle to clean based on the target cleaning area.

[0052] Obtain the real-time image of the solar panel surface of the image acquisition device, and the real-time image corresponds to specific ambient light intensity information, ambient light illumination angle, and image acquisition control parameters. Subsequently, analyze the real-time image of the solar panel surface and the reference image of the solar panel surface through the siamese comparison network, analyze the differences between the real-time image of the solar panel surface and the reference image of the solar panel surface, and obtain the target cleaning area. Finally, control the self-cleaning nozzle to clean based on the obtained target cleaning area. This solves the technical problem in the prior art that the cleaning method of photovoltaic modules has low cleaning intelligence, resulting in difficult improvement of cleaning efficiency and cleaning effect. It achieves the technical effects of greatly improving the cleaning efficiency and effect of photovoltaic modules in fishery applications, reducing labor costs, extending the service life of photovoltaic modules, improving the overall power generation efficiency, and ensuring the stability of the aquaculture environment.

[0053] The method provided in the embodiment of the present application further includes:

[0054] Extract the first color feature distribution matrix of the real-time image of the solar panel surface through the first feature extraction channel of the siamese comparison network;

[0055] Extract the second color feature distribution matrix of the reference image on the surface of the solar panel through the second feature extraction channel of the siamese comparison network, where the first color feature distribution matrix and the second color feature distribution matrix have coordinate consistency;

[0056] Analyze the first color feature distribution matrix and the second color feature distribution matrix through the feature comparison channel of the siamese comparison network to obtain the target cleaning area.

[0057] Analyze the real-time image on the surface of the solar panel and the reference image on the surface of the solar panel through a siamese comparison network to generate a target cleaning area, including: extracting the first color feature distribution matrix of the real-time image on the surface of the solar panel through the first feature extraction channel of the siamese comparison network. The color feature distribution matrix records the color features of each pixel by mapping the color information of the image into the matrix. These matrices can be represented in different color spaces, such as RGB, HSV, etc., and each element of the matrix represents the color value distribution of the corresponding position in the image. Subsequently, extract the second color feature distribution matrix of the reference image on the surface of the solar panel through the second feature extraction channel of the siamese comparison network, where the first color feature distribution matrix and the second color feature distribution matrix have coordinate consistency. Further, analyze the first color feature distribution matrix and the second color feature distribution matrix through the feature comparison channel of the siamese comparison network to obtain the target cleaning area.

[0058] The method provided by the embodiment of the present application further includes:

[0059] Obtain the first color feature and the second color feature of the first coordinate;

[0060] When the color feature distance between the first color feature and the second color feature is greater than or equal to the color feature distance threshold, add the first coordinate to the initial target cleaning area;

[0061] After all coordinates are divided, cluster the target cleaning coordinates of the initial target cleaning area according to the distribution distance threshold to generate an aggregation area set, where the aggregation area set has an area area label set;

[0062] According to the area area label set, extract the aggregation areas greater than or equal to the area threshold from the aggregation area set and set them as the target cleaning area.

[0063] Analyzing the first color feature distribution matrix and the second color feature distribution matrix through the feature comparison channel of the twin comparison network to obtain the target cleaning area includes: obtaining the first color feature and the second color feature of the first coordinate, where the first coordinate is the pixel coordinate in the image, corresponding to the corresponding position of an element in the first color feature distribution matrix and the second color feature distribution matrix. When the color feature distance between the first color feature and the second color feature is greater than or equal to the color feature distance threshold, adding the first coordinate to the initial target cleaning area. The color feature distance threshold is the preset minimum deviation threshold of the color feature. When it is greater than this threshold, the deviation between the first color feature and the second color feature is relatively high. Since the second color feature is the reference color feature of the solar panel, when the deviation is relatively high, the first color feature obtained in real time may be contaminated by foreign objects, so the first coordinate at the corresponding position is added to the initial target cleaning area. Judging and dividing the first and second color feature distances for all coordinates. When all coordinates are divided, clustering the target cleaning coordinates of the initial target cleaning area according to the distribution distance threshold, that is, classifying the coordinate positions less than or equal to the distribution distance threshold into one cleaning area. The distribution distance threshold is the preset classification distance. When the distance is less than this distance, the coordinate distances are relatively close and they belong to the same area, otherwise the distances are relatively far and they do not belong to the same area. Clustering all the cleaning coordinates in the initial target cleaning area based on the distribution distance threshold to generate an aggregated area set. Among them, the aggregated area set has an area label set of regions, and the area label of the region is the area formed by the cleaning coordinates of each region in the aggregated area set, which can be represented by the number of cleaning coordinates or the actual cleaning area corresponding to the coordinates. Finally, according to the area label set of the regions, extracting the aggregated regions greater than or equal to the area threshold from the aggregated area set and setting them as the target cleaning area.

[0064] The method provided by the embodiment of the present application further includes:

[0065] Configuring the dust color feature;

[0066] Sorting from the target cleaning area based on the dust color feature to obtain the proportion of the dust distribution area;

[0067] When the proportion of the dust distribution area is greater than or equal to the area proportion threshold, controlling the self-cleaning nozzle to perform cleaning based on the first impact force;

[0068] When the proportion of the dust distribution area is less than the area proportion threshold, controlling the self-cleaning nozzle to perform cleaning based on the second impact force, where the second impact force is greater than the first impact force.

[0069] Based on the target cleaning area, controlling the self-cleaning nozzle to perform cleaning includes: configuring a dust color feature, which is color feature data generated by dust pollution and is a parameter obtained by analyzing the dust pollution distribution in advance. Obtaining the mean value of the color features in the target cleaning area. Subsequently, sorting from the target cleaning area based on the dust color feature, obtaining and summing the area labels of the cleaning areas in the target cleaning area that have the same dust color feature, calculating the ratio based on the sum result and the sum result of the area labels of all target cleaning areas to obtain the proportion of the dust distribution area, and this proportion is the ratio of dust pollution. When the proportion of the dust distribution area is greater than or equal to the area proportion threshold, at this time most of the solar panels are covered with dust and are easy to rinse, controlling the self-cleaning nozzle to perform cleaning based on the first impact force. When the area of the dust distribution area is less than the area proportion threshold, it indicates that there may be more pollutants with strong adsorption force such as leaves and bird droppings, then use a larger impact force to clean and control the self-cleaning nozzle to perform cleaning based on the second impact force, where the second impact force is greater than the first impact force.

[0070] In the above text, with reference to Figure 1 a self-cleaning method of a fishery-solar complementary photovoltaic module according to an embodiment of the present invention is described in detail. Next, with reference to Figure 2 a self-cleaning device of a fishery-solar complementary photovoltaic module according to an embodiment of the present invention will be described.

[0071] A self-cleaning device of a fishery-solar complementary photovoltaic module according to an embodiment of the present invention solves the technical problem in the prior art that the cleaning method of photovoltaic modules has low cleaning intelligence, resulting in difficult improvement of cleaning efficiency and cleaning effect. It achieves the technical effects of greatly improving the cleaning efficiency and effect of photovoltaic modules in fishery applications, reducing labor costs, extending the service life of photovoltaic modules, improving the overall power generation efficiency, and ensuring the stability of the aquaculture environment. A self-cleaning device of a fishery-solar complementary photovoltaic module includes: a data acquisition module 11, a control parameter acquisition module 12, a reference image acquisition module 13, an image acquisition module 14, a cleaning area acquisition module 15, and a cleaning module 16.

[0072] The data acquisition module 11 is used to obtain ambient light intensity information and ambient light illumination angle through a light sensor, where the ambient light illumination angle refers to the angle between the line connecting the sun position and the image acquisition device and the solar panel;

[0073] The control parameter acquisition module 12 is used to obtain the image acquisition control parameters of the image acquisition device of the photovoltaic module;

[0074] A reference image acquisition module 13, configured to perform image analysis based on the image acquisition control parameters, the ambient light intensity information, and the ambient light illumination angle, and generate a reference image of the solar panel surface;

[0075] An image acquisition module 14, configured to obtain a real-time image of the solar panel surface of the image acquisition device;

[0076] A cleaning area acquisition module 15, configured to analyze the real-time image of the solar panel surface and the reference image of the solar panel surface through a twin comparison network, and generate a target cleaning area;

[0077] A cleaning module 16, configured to control a self-cleaning nozzle to perform cleaning based on the target cleaning area.

[0078] Next, the specific configuration of the reference image acquisition module 13 will be described in detail. The reference image acquisition module 13 may further include: performing image analysis based on the image acquisition control parameters and the ambient light intensity information to generate a reference image of the solar panel surface, including: configuring reference acquisition control parameters, reference ambient light intensity, reference illumination angle, and a reference solar panel surface image, where the reference solar panel surface image has a reference color feature value distribution matrix; collecting an image acquisition control parameter record data set, an ambient light intensity record data set, an illumination angle record data set, and a solar panel surface image record data set, where the solar panel surface image record data set has a recorded color feature value distribution matrix set; training a color distribution deviation vector matrix analysis channel according to the reference acquisition control parameters, the reference ambient light intensity, the reference illumination angle, the reference color feature value distribution matrix, in combination with the set of image acquisition control parameter record data sets, the ambient light intensity record data set, the illumination angle record data set, and the recorded color feature value distribution matrix set; training an image generation channel according to the reference color feature value distribution matrix, the recorded color feature value distribution matrix set, the reference solar panel surface image, and the solar panel surface image record data set; merging the color distribution deviation vector matrix analysis channel and the image generation channel to generate an image analysis model, and performing image analysis according to the image acquisition control parameters and the ambient light intensity information to generate the reference image of the solar panel surface.

[0079] Next, the specific configuration of the reference image acquisition module 13 will be further described in detail. The reference image acquisition module 13 further includes: training a color distribution deviation vector matrix analysis channel according to the reference acquisition control parameters, the reference ambient light intensity, the reference illumination angle, the reference color eigenvalue distribution matrix, in combination with the acquisition image acquisition control parameter record data set, the ambient light intensity record data set, the illumination angle record data set, and the recorded color eigenvalue distribution matrix set, including: based on the reference acquisition control parameters, traversing the acquisition control parameter record data set for deviation analysis to generate a control parameter deviation vector array set; based on the reference ambient light intensity, traversing the ambient light intensity record data set for deviation analysis to generate an ambient light intensity deviation vector set; based on the reference illumination angle, traversing the illumination angle record data set for deviation analysis to generate an illumination angle deviation vector set; based on the reference color eigenvalue distribution matrix, traversing the recorded color eigenvalue distribution matrix set for deviation analysis to generate a color eigenvalue distribution deviation vector matrix set; training the color distribution deviation vector matrix analysis channel according to the control parameter deviation vector array set, the ambient light intensity deviation vector set, the illumination angle deviation vector set, and the color eigenvalue distribution deviation vector matrix set.

[0080] Next, the specific configuration of the reference image acquisition module 13 will be described in detail. The reference image acquisition module 13 may further include: training an image generation channel according to the reference color eigenvalue distribution matrix, the recorded color eigenvalue distribution matrix set, the reference solar panel surface image, and the solar panel surface image record data set, including: training the image generation channel according to the color eigenvalue distribution deviation vector matrix set, the reference solar panel surface image, and the solar panel surface image record data set.

[0081] Next, the specific configuration of the cleaning area acquisition module 15 will be described in detail. The cleaning area acquisition module 15 further includes: analyzing the real-time image of the solar panel surface and the reference image of the solar panel surface through a siamese comparison network to generate a target cleaning area, including: extracting a first color feature distribution matrix of the real-time image of the solar panel surface through the first feature extraction channel of the siamese comparison network; extracting a second color feature distribution matrix of the reference image of the solar panel surface through the second feature extraction channel of the siamese comparison network, where the first color feature distribution matrix and the second color feature distribution matrix have coordinate consistency; parsing the first color feature distribution matrix and the second color feature distribution matrix through the feature comparison channel of the siamese comparison network to obtain the target cleaning area.

[0082] Next, the specific configuration of the cleaning area acquisition module 15 will be further described in detail. The cleaning area acquisition module 15 further includes: parsing the first color feature distribution matrix and the second color feature distribution matrix through the feature comparison channel of the twin comparison network to obtain the target cleaning area, including: obtaining the first color feature and the second color feature of the first coordinate; when the color feature distance between the first color feature and the second color feature is greater than or equal to the color feature distance threshold, adding the first coordinate to the initial target cleaning area; after all coordinates are divided, clustering the target cleaning coordinates of the initial target cleaning area according to the distribution distance threshold to generate an aggregation area set, where the aggregation area set has an area area label set; according to the area area label set, extracting the aggregation areas greater than or equal to the area threshold from the aggregation area set and setting them as the target cleaning area.

[0083] Next, the specific configuration of the cleaning module 16 will be described in detail. The cleaning module 16 further includes: controlling the self-cleaning nozzle to perform cleaning based on the target cleaning area, including: configuring the dust color feature; sorting from the target cleaning area based on the dust color feature to obtain the dust distribution area proportion; when the dust distribution area proportion is greater than or equal to the area proportion threshold, controlling the self-cleaning nozzle to perform cleaning based on the first impact force; when the dust distribution area is less than the area proportion threshold, controlling the self-cleaning nozzle to perform cleaning based on the second impact force, where the second impact force is greater than the first impact force.

[0084] The self-cleaning device of a fish-solar complementary photovoltaic module provided by an embodiment of the present invention can execute the self-cleaning method of a fish-solar complementary photovoltaic module provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects for executing the method.

[0085] Although this application makes various references to certain modules in the device according to the embodiments of this application, however, any number of different modules can be used and run on the user terminal and / or server. The included various units and modules are only divided according to the functional logic, but are not limited to the above division as long as the corresponding functions can be realized; in addition, the specific names of the functional units are only for the convenience of mutual distinction and do not limit the protection scope of the present invention.

[0086] The above specific embodiments do not constitute a limitation to the protection scope of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principle of this application shall be included within the protection scope of this application.

Claims

1. A self-cleaning method for a fish-light complementary photovoltaic assembly, characterized in that: include: The ambient light intensity information and the ambient light angle are obtained through the light sensor, wherein the ambient light angle refers to the angle between the sun position and the line connecting the image acquisition device and the solar panel; Obtaining image acquisition control parameters of an image acquisition device of a photovoltaic module; Perform image analysis according to the image acquisition control parameters, the ambient light intensity information and the ambient light angle to generate a reference image of the solar panel surface; Obtaining a real-time image of the solar panel surface from the image acquisition device; Analyzing the real-time image of the solar panel surface and the reference image of the solar panel surface through a twin comparison network to generate a target cleaning area; Based on the target cleaning area, controlling the self-cleaning nozzle to perform cleaning; Performing image analysis according to the image acquisition control parameters and the ambient light intensity information to generate a reference image of the solar panel surface includes: Configure reference acquisition control parameters, reference ambient light intensity, reference illumination angle, and reference solar panel surface image, wherein the reference solar panel surface image has a reference color eigenvalue distribution matrix; Collecting an image acquisition control parameter recording data set, an ambient light intensity recording data set, an illumination angle recording data set, and a solar panel surface image recording data set, wherein the solar panel surface image recording data set has a recording color eigenvalue distribution matrix set; According to the reference acquisition control parameters, the reference ambient light intensity, the reference illumination angle, and the reference color eigenvalue distribution matrix, in combination with the set of image acquisition control parameter recording data sets, the ambient light intensity recording data sets, the illumination angle recording data sets, and the recorded color eigenvalue distribution matrix set, a color distribution deviation vector matrix parsing channel is trained; training an image generation channel according to the reference color eigenvalue distribution matrix, the recorded color eigenvalue distribution matrix set, the reference solar panel surface image, and the solar panel surface image record data set; The color distribution deviation vector matrix analysis channel and the image generation channel are combined to generate an image analysis model, and image analysis is performed according to the image acquisition control parameters and the ambient light intensity information to generate a reference image of the solar panel surface; According to the reference acquisition control parameter, the reference ambient light intensity, the reference illumination angle, and the reference color eigenvalue distribution matrix, in combination with the image acquisition control parameter recording data set, the ambient light intensity recording data set, the illumination angle recording data set, and the recorded color eigenvalue distribution matrix set, a color distribution deviation vector matrix analysis channel is trained, including: Based on the reference acquisition control parameter, traverse the acquisition control parameter record data set to perform deviation analysis and generate a control parameter deviation vector array set; Based on the reference ambient light intensity, traverse the ambient light intensity record data set to perform deviation analysis and generate an ambient light intensity deviation vector set; Based on the reference illumination angle, traverse the illumination angle record data set to perform deviation analysis and generate an illumination angle deviation vector set; Based on the reference color eigenvalue distribution matrix, traverse the recorded color eigenvalue distribution matrix set to perform deviation analysis and generate a color eigenvalue distribution deviation vector matrix set; Training a color distribution deviation vector matrix analysis channel according to the control parameter deviation vector array set, the ambient light intensity deviation vector set, the illumination angle deviation vector set, and the color eigenvalue distribution deviation vector matrix set; According to the reference color eigenvalue distribution matrix, the recorded color eigenvalue distribution matrix set, the reference solar panel surface image and the solar panel surface image record data set, training an image generation channel includes: The image generation channel is trained according to the color feature value distribution deviation vector matrix set, the reference solar panel surface image and the solar panel surface image record data set.

2. The method according to claim 1, characterized in that The real-time image of the solar panel surface and the reference image of the solar panel surface are analyzed by a twin comparison network to generate a target cleaning area, including: Extracting a first color feature distribution matrix of the real-time image of the solar panel surface through a first feature extraction channel of the twin comparison network; Extracting a second color feature distribution matrix of the solar panel surface reference image through a second feature extraction channel of the twin comparison network, wherein the first color feature distribution matrix and the second color feature distribution matrix have coordinate consistency; The first color feature distribution matrix and the second color feature distribution matrix are analyzed through the feature comparison channel of the twin comparison network to obtain the target cleaning area.

3. The method according to claim 2, characterized in that The first color feature distribution matrix and the second color feature distribution matrix are analyzed through the feature comparison channel of the twin comparison network to obtain the target cleaning area, including: Obtain a first color feature and a second color feature of a first coordinate; When the color feature distance between the first color feature and the second color feature is greater than or equal to a color feature distance threshold, adding the first coordinate to an initial target cleaning area; When all coordinate divisions are completed, clustering the target cleaning coordinates of the initial target cleaning area according to a distribution distance threshold to generate a clustered area set, wherein the clustered area set has a region area label set; According to the region area label set, an aggregated region whose area is greater than or equal to a region area threshold is extracted from the aggregated region set and is set as the target cleaning region.

4. The method according to claim 1, characterized in that Based on the target cleaning area, controlling the self-cleaning nozzle to perform cleaning includes: Configure dust color characteristics; Sorting the dust from the target cleaning area based on the dust color characteristics to obtain the area ratio of dust distribution area; When the area ratio of the dust distribution area is greater than or equal to the area ratio threshold, controlling the self-cleaning nozzle to perform cleaning based on the first impact force; When the area of ​​the dust distribution region is smaller than the area ratio threshold, the self-cleaning nozzle is controlled to perform cleaning based on a second impact force, wherein the second impact force is greater than the first impact force.

5. A self-cleaning device for a fish-light complementary photovoltaic assembly, characterized in that: The device is used to execute the method according to any one of claims 1 to 4, and the device comprises: The data acquisition module is used to obtain ambient light intensity information and ambient light angle through a light sensor, wherein the ambient light angle refers to the angle between the sun position and the line connecting the image acquisition device and the solar panel; An acquisition control parameter acquisition module is used to obtain image acquisition control parameters of an image acquisition device for a photovoltaic module; A reference image acquisition module, used for performing image analysis according to the image acquisition control parameters, the ambient light intensity information and the ambient light angle, to generate a reference image of the solar panel surface; An image acquisition module, used to obtain a real-time image of the solar panel surface of the image acquisition device; A cleaning area acquisition module, used for analyzing the real-time image of the solar panel surface and the reference image of the solar panel surface through a twin comparison network to generate a target cleaning area; The cleaning module is used to control the self-cleaning nozzle to perform cleaning based on the target cleaning area.

Images