Control method, device, electronic device, storage medium and program product of photovoltaic cleaning robot

Abstract

The present disclosure provides a control method, device, electronic device, storage medium and program product of a photovoltaic cleaning robot, including: analyzing the photovoltaic panel image through a pre-trained dirt detection model, determining the dirty area and its physical coordinates, and calculating the severity score. Dividing the photovoltaic panel into grid units, planning the cleaning path according to the score and location information. Collecting images in real time, if new dirt is detected or the score changes exceed the threshold, dynamically adjusting the cleaning path. The present disclosure can dynamically adjust the cleaning strategy according to the actual situation, so as to improve the cleaning effect and energy efficiency of the photovoltaic panel.

Description

Technical Field

[0001] The present disclosure relates to the technical field of intelligent cleaning of photovoltaic power stations, and in particular to a control method, device, electronic equipment, storage medium and program product of a photovoltaic cleaning robot. Background Art

[0002] This section is intended to provide a background or context to the embodiments of the disclosure that are recited in the claims. No description herein is admitted to be prior art by inclusion in this section.

[0003] Photovoltaic panels are the core components of solar power generation systems, and their cleanliness directly affects the power generation efficiency; however, dust, bird droppings, leaves and other dirt easily accumulate on the surface of photovoltaic panels, resulting in a significant decrease in photoelectric conversion efficiency. In order to solve this problem, a variety of photovoltaic cleaning robots have appeared on the market, which have improved the cleaning efficiency of photovoltaic panels to a certain extent.

[0004] However, in the related art, since photovoltaic cleaning robots cannot dynamically adjust cleaning strategies according to actual conditions, there are often problems of incomplete cleaning or excessive cleaning, resulting in waste of resources and poor cleaning effects, which in turn affects the overall power generation efficiency of photovoltaic panels. Summary of the invention

[0005] In view of this, the purpose of the present disclosure is to propose a control method, device, electronic device, storage medium and program product for a photovoltaic cleaning robot, which at least solves one of the technical problems in the related art to a certain extent.

[0006] Based on the above purpose, the first aspect of the exemplary embodiment of the present disclosure provides a control method for a photovoltaic cleaning robot, which is applied to a server. The method includes:

[0007] Determine image information of the surface of the photovoltaic panel, and detect the image information using a pre-trained dirt detection model to obtain a dirt probability map;

[0008] Judging the dirt probability map based on a set first threshold value to obtain a dirt area;

[0009] Determine geometric characteristic parameters of the dirty area, and transform the geometric characteristic parameters based on a transformation matrix to obtain physical coordinates of the dirty area;

[0010] Determine dirty region characteristic information of the dirty region, the total area and the diagonal distance of the photovoltaic panel surface, and perform calculation based on the dirty region characteristic information, the total area and the diagonal distance to obtain a first severity score;

[0011] Dividing the photovoltaic panel surface into a plurality of grid units; mapping the dirty area to the corresponding grid unit based on the physical coordinates to obtain a dirty area identifier corresponding to each grid unit;

[0012] Calculating based on the first severity score and the dirty area identifier corresponding to each of the grid units to obtain a second severity score corresponding to each of the grid units;

[0013] Determining the position information of each of the grid cells, and performing path planning based on the position information and the second severity score to obtain an optimized cleaning path;

[0014] The photovoltaic panel surface is collected in real time to obtain a real-time collected image; in response to determining that new dirty area information is detected based on the dirt detection model and / or the change in the first severity score exceeds a set third threshold, the optimized cleaning path is adjusted to obtain a dynamically optimized path.

[0015] Based on the same inventive concept, the second aspect of the exemplary embodiment of the present disclosure provides a control device for a photovoltaic cleaning robot, comprising:

[0016] A dirt probability map determination module is configured to determine image information of the surface of the photovoltaic panel, and detect the image information through a pre-trained dirt detection model to obtain a dirt probability map;

[0017] a dirty area determination module, configured to determine the dirt probability map based on a set first threshold value to obtain a dirty area;

[0018] A physical coordinate determination module is configured to determine geometric characteristic parameters of the dirty area, and transform the geometric characteristic parameters based on a transformation matrix to obtain physical coordinates of the dirty area;

[0019] A first score determination module is configured to determine the dirty region characteristic information of the dirty region, the total area and the diagonal distance of the photovoltaic panel surface, and calculate based on the dirty region characteristic information, the total area and the diagonal distance to obtain a first severity score;

[0020] A dirty area identification determination module is configured to divide the surface of the photovoltaic panel into a plurality of grid units; based on the physical coordinates, the dirty area is mapped to the corresponding grid unit to obtain a dirty area identification corresponding to each grid unit;

[0021] a second score determination module configured to calculate based on the first severity score and the dirty area identifier corresponding to each of the grid units to obtain a second severity score corresponding to each of the grid units;

[0022] a cleaning path determination module, configured to determine position information of each of the grid cells, perform path planning based on the position information and the second severity score, and obtain an optimized cleaning path;

[0023] The dynamic optimization path determination module is configured to perform real-time acquisition of the surface of the photovoltaic panel to obtain a real-time acquired image; in response to determining that new dirty area information is detected based on the dirt detection model and / or the change in the first severity score exceeds a set third threshold, the optimized cleaning path is adjusted to obtain a dynamically optimized path.

[0024] Based on the same inventive concept, a third aspect of the exemplary embodiment of the present disclosure provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method described in the first aspect is implemented.

[0025] Based on the same inventive concept, a fourth aspect of the exemplary embodiments of the present disclosure provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the method described in the first aspect.

[0026] Based on the same inventive concept, a fifth aspect of the exemplary embodiments of the present disclosure provides a computer program product, including computer program instructions. When the computer program instructions are executed on a computer, the computer executes the method described in the first aspect.

[0027] From the above, it can be seen that the control method, device, electronic device, storage medium and program product of the photovoltaic cleaning robot provided by the embodiments of the present disclosure include:

[0028] Determine the image information of the surface of the photovoltaic panel, detect the image information through a pre-trained dirt detection model to obtain a dirt probability map; judge the dirt probability map based on a set first threshold to obtain a dirty area; determine the geometric feature parameters of the dirty area, and transform the geometric feature parameters based on a transformation matrix to obtain the physical coordinates of the dirty area; determine the dirty area feature information of the dirty area, the total area and the diagonal distance of the photovoltaic panel surface, and calculate based on the dirty area feature information, the total area and the diagonal distance to obtain a first severity score; divide the photovoltaic panel surface into a plurality of grid units; map the dirty area based on the physical coordinates to the corresponding grid unit, and obtain the dirty area identification corresponding to each grid unit; calculate based on the first severity score and the dirty area identification corresponding to each grid unit, and obtain the second severity score corresponding to each grid unit; determine the position information of each grid unit, and plan the path based on the position information and the second severity score to obtain an optimized cleaning path; collect the surface of the photovoltaic panel in real time to obtain a real-time collected image; in response to determining that the newly added dirty area information and / or the change in the first severity score detected based on the dirt detection model exceeds the set third threshold, adjust the optimized cleaning path to obtain a dynamically optimized path. The present disclosure can dynamically adjust the cleaning strategy according to actual conditions to improve the cleaning effect and energy efficiency of photovoltaic panels. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the technical solutions in the present disclosure or related technologies, the drawings required for use in the embodiments or related technical descriptions are briefly introduced below. Obviously, the drawings described below are only embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0030] Figure 1 A schematic diagram of an application scenario of a control method for a photovoltaic cleaning robot provided by an exemplary embodiment of the present disclosure;

[0031] Figure 2 A schematic flow chart of a control method for a photovoltaic cleaning robot provided by an exemplary embodiment of the present disclosure;

[0032] Figure 3 A schematic structural diagram of a control device of a photovoltaic cleaning robot provided by an exemplary embodiment of the present disclosure;

[0033] Figure 4 A schematic diagram of the hardware structure of an electronic device provided for an exemplary embodiment of the present disclosure. DETAILED DESCRIPTION

[0034] It is understandable that before using the technical solutions disclosed in the embodiments of this application, the type, scope of use, usage scenarios, etc. of the personal information involved in this application should be informed to the user and the user's authorization should be obtained in an appropriate manner in accordance with relevant laws and regulations.

[0035] For example, in response to receiving an active request from a user, a prompt message is sent to the user to clearly prompt the user that the operation requested to be performed will require obtaining and using the user's personal information. Thus, the user can autonomously choose whether to provide personal information to software or hardware such as an electronic device, application, server, or storage medium that performs the operation of the technical solution of the present application according to the prompt message.

[0036] As an optional but non-limiting implementation, in response to receiving an active request from the user, the prompt information may be sent to the user in the form of a pop-up window, in which the prompt information may be presented in text form. In addition, the pop-up window may also carry a selection control for the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0037] It is understandable that the above notification and the process of obtaining user authorization are merely illustrative and do not constitute a limitation on the implementation method of the present application. Other methods that meet the relevant laws and regulations may also be applied to the implementation method of the present application.

[0038] It is understandable that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and relevant provisions.

[0039] In order to make the purpose, technical solutions and advantages of the present disclosure more clear, the principles and spirit of the present disclosure will be described with reference to several exemplary embodiments. It should be understood that these embodiments are provided only to enable those skilled in the art to better understand and implement the present disclosure, and are not intended to limit the scope of the present disclosure in any way. On the contrary, these embodiments are provided to make the present disclosure more thorough and complete, and to fully convey the scope of the present disclosure to those skilled in the art.

[0040] It should be understood herein that any number of elements in the drawings is for illustration rather than limitation, and any naming is only for distinction rather than having any limiting meaning.

[0041] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present disclosure should be understood by people with ordinary skills in the field to which the present disclosure belongs. The "first", "second" and similar words used in the embodiments of the present disclosure do not represent any order, quantity or importance, but are only used to distinguish different components. "Including" or "comprising" and similar words mean that the elements or objects appearing in front of the word cover the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connecting" or "connected" and similar words are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly. The article "one" or "a" before an element does not exclude the existence of multiple such elements.

[0042] The principle and spirit of the present disclosure are explained in detail below with reference to several representative embodiments of the present disclosure.

[0043] As described in the background technology, in the related art, since the photovoltaic cleaning robot cannot dynamically adjust the cleaning strategy according to the actual situation, there are often problems of incomplete cleaning or excessive cleaning, resulting in waste of resources and poor cleaning effect, which in turn affects the overall power generation efficiency of the photovoltaic panel. Specifically, when performing cleaning tasks, photovoltaic cleaning robots often adopt a preset cleaning strategy, which lacks consideration of the actual dirtiness. Therefore, the robot may over-clean some areas and under-clean other areas. This is because they cannot dynamically adjust the cleaning behavior according to the actual degree of dirtiness on the photovoltaic panel, and cannot cover all dirty areas on the photovoltaic panel, especially those areas that are not on the preset path, resulting in incomplete cleaning. In addition, due to the lack of perception of the real-time condition of the photovoltaic panel surface, this cleaning method cannot adapt to changes in dirt distribution.

[0044] Since the robot does not have a dirt detection function, it cannot identify which areas really need to be cleaned and which areas are relatively clean. Therefore, they cannot adjust the cleaning strategy according to the actual dirtiness, which leads to over-cleaning in less dirty areas and under-cleaning in heavily dirty areas, resulting in a waste of resources. In addition, if the cleaning robot applies the same cleaning intensity to all areas, ignoring the difference in dirtiness in different areas, this may lead to over-cleaning of lightly dirty areas and failure to clean heavily dirty areas thoroughly due to insufficient cleaning intensity.

[0045] Furthermore, due to the inability to dynamically identify and assess the severity of dirt, the robot may not be able to prioritize cleaning the areas that have the greatest impact on the performance of the photovoltaic panel. This leads to inconsistent cleaning results, affecting the overall power generation efficiency of the photovoltaic panel.

[0046] In order to solve the above problems, the present disclosure provides a control method, device, electronic device, storage medium and program product solution of a photovoltaic cleaning robot, specifically including:

[0047] Determine the image information of the surface of the photovoltaic panel, detect the image information through a pre-trained dirt detection model to obtain a dirt probability map; judge the dirt probability map based on a set first threshold to obtain a dirty area; determine the geometric feature parameters of the dirty area, and transform the geometric feature parameters based on a transformation matrix to obtain the physical coordinates of the dirty area; determine the dirty area feature information of the dirty area, the total area and the diagonal distance of the photovoltaic panel surface, and calculate based on the dirty area feature information, the total area and the diagonal distance to obtain a first severity score; divide the photovoltaic panel surface into a plurality of grid units; map the dirty area based on the physical coordinates to the corresponding grid unit, and obtain the dirty area identification corresponding to each grid unit; calculate based on the first severity score and the dirty area identification corresponding to each grid unit, and obtain the second severity score corresponding to each grid unit; determine the position information of each grid unit, and plan the path based on the position information and the second severity score to obtain an optimized cleaning path; collect the surface of the photovoltaic panel in real time to obtain a real-time collected image; in response to determining that the newly added dirty area information and / or the change in the first severity score detected based on the dirt detection model exceeds the set third threshold, adjust the path of the optimized cleaning path to obtain a dynamic optimized path. The present disclosure identifies dirty areas in real time through the collected data, calculates the dirt location and area, and performs a dirt severity score; plans a dynamic cleaning path based on the dirt severity score and position, and can effectively identify the dirty areas and their distribution, and can dynamically adjust the cleaning strategy according to the actual situation. Based on the dynamically planned cleaning path, it ensures that key dirty areas are cleaned first, thereby improving the cleaning effect and energy efficiency of photovoltaic panels.

[0048] After introducing the basic principles of the present disclosure, various non-limiting embodiments of the present disclosure are described in detail below.

[0049] refer to Figure 1 , which is a schematic diagram of an application scenario of the control method of the photovoltaic cleaning robot provided by the exemplary embodiment of the present disclosure.

[0050] This application scenario includes a terminal device 101 and a server 102. The terminal device 101 and the server 102 may be connected via a wired or wireless communication network to achieve data interaction.

[0051] The terminal device 101 may be an electronic device with data transmission and multimedia input / output functions close to the user side, including but not limited to a desktop computer, a mobile phone, a mobile computer, a tablet computer, a media player, a smart wearable device, a personal digital assistant (PDA) or other electronic devices capable of implementing the above functions. The electronic device may include a processor and a display screen with a touch input function, the display screen is used to present a graphical user interface, the graphical user interface can display an application interface, and the processor is used to process application data, generate a graphical user interface, and control the display of the graphical user interface on the display screen.

[0052] Server 102 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers. It can also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), as well as big data and artificial intelligence platforms.

[0053] In some exemplary embodiments, the control method of the photovoltaic cleaning robot may be executed on the terminal device 101 or the server 102 .

[0054] When the control method of the photovoltaic cleaning robot is run on the server 102 , the server 102 is used to provide control services of the photovoltaic cleaning robot to the user of the terminal device 101 .

[0055] The terminal device 101 acquires image information of the surface of the photovoltaic panel in real time, and transmits the image information to the server 102;

[0056] The server 102 receives the image information transmitted by the terminal device 101;

[0057] The server 102 detects the image information using a pre-trained dirt detection model to obtain a dirt probability map;

[0058] The server 102 determines the dirt probability map based on a set first threshold to obtain a dirt area;

[0059] The server 102 determines geometric feature parameters of the dirty area, and the server 102 converts the geometric feature parameters based on a conversion matrix to obtain physical coordinates of the dirty area;

[0060] The server 102 determines the dirty region characteristic information of the dirty region, the total area and the diagonal distance of the photovoltaic panel surface, and the server 102 performs calculation based on the dirty region characteristic information, the total area and the diagonal distance to obtain a first severity score;

[0061] The server 102 divides the surface of the photovoltaic panel into a plurality of grid units; based on the physical coordinates, the server 102 maps the dirty area to the corresponding grid unit to obtain a dirty area identifier corresponding to each grid unit;

[0062] The server 102 performs calculation based on the first severity score and the dirty area identifier corresponding to each grid unit to obtain a second severity score corresponding to each grid unit;

[0063] The server 102 determines the position information of each of the grid cells, and performs path planning based on the position information and the second severity score to obtain an optimized cleaning path;

[0064] The server 102 collects data on the surface of the photovoltaic panel in real time to obtain a real-time collected image; in response to determining that new dirty area information is detected based on the dirt detection model and / or the change in the first severity score exceeds a set third threshold, the optimized cleaning path is adjusted to obtain a dynamic optimized path, and the server 102 transmits the dynamic optimized path to the terminal device 101.

[0065] It should be noted that the above application scenarios are only shown to facilitate understanding of the spirit and principle of the present disclosure, and the embodiments of the present disclosure are not limited in this respect. On the contrary, the embodiments of the present disclosure can be applied to any applicable scenario.

[0066] refer to Figure 2 , a control method for a photovoltaic cleaning robot, the method comprising the following steps:

[0067] Step S210: determine image information of the surface of the photovoltaic panel, and detect the image information using a pre-trained dirt detection model to obtain a dirt probability map.

[0068] In specific implementation, the method of determining the image information on the surface of the photovoltaic panel is as follows:

[0069] A high-resolution camera is installed on the surface of the photovoltaic panel, and the position and angle of the camera are precisely adjusted to ensure that image acquisition can cover the entire surface of the photovoltaic panel, thereby obtaining image information on the surface of the photovoltaic panel.

[0070] In the above exemplary embodiment, a method of determining image information of the photovoltaic panel surface is introduced. Next, a method of obtaining a dirt probability map is introduced:

[0071] In this exemplary embodiment, the image information is subjected to denoising, contrast enhancement, perspective correction, image segmentation and feature extraction to obtain a preprocessed image; the preprocessed image is subjected to image adjustment and normalization to obtain a normalized image; the normalized image is input into a pre-trained dirt detection model to obtain the dirt probability map.

[0072] In a specific implementation, performing denoising, contrast enhancement, perspective correction, image segmentation and feature extraction on the image information to obtain a preprocessed image means:

[0073] The original image collected may be affected by sensor noise or environmental clutter, and needs to be denoised to improve the accuracy of subsequent dirt detection. Specifically, median filtering is performed on the image. Median filtering is an effective denoising method that can eliminate salt and pepper noise (i.e., random white noise or black noise in the image) without losing the edge details of the image.

[0074] Due to the complexity of the ambient lighting in the photovoltaic power station, the collected images may have insufficient contrast. Contrast enhancement is required to improve the difference between the dirty area and the background. The local histogram equalization (CLAHE) technology is used. The specific operations are as follows:

[0075] The image is divided into several small blocks, each small block is subjected to histogram equalization processing, and then the processed image blocks are reassembled into an enhanced image;

[0076] Since there is a certain angle between the camera and the surface of the photovoltaic panel, the captured image may have perspective distortion, and perspective correction processing is required to ensure that the geometric shape of the image is consistent with the actual photovoltaic panel surface;

[0077] In order to improve the accuracy of subsequent dirt detection, the corrected image is divided into several small blocks, each of which covers a sub-area of ​​the photovoltaic panel. The size of the block should be set according to the size of the photovoltaic panel and the camera resolution. Usually, the image is divided into 64×64 pixel grid units;

[0078] A feature extraction method based on local binary pattern (LBP) is used to extract features from each image block to capture the texture information in the image block for subsequent dirt recognition.

[0079] In a specific implementation, the pre-processed image is subjected to image adjustment and normalization processing to obtain a normalized image, which means:

[0080] The unnecessary edge parts of the preprocessed image are removed, and only the photovoltaic panel area is retained. The image is adjusted to a size suitable for the input of the dirt detection model, and the feature vector of the preprocessed image is normalized to eliminate the influence of illumination changes on the feature value for subsequent dirt detection and classification.

[0081] The image pixel values ​​are normalized so that the pixel value range is between 0 and 1 to accelerate the convergence of the model. The expression is:

[0082]

[0083] in, is the pixel value matrix of the input image, is the normalized image pixel value matrix.

[0084] In the above exemplary embodiment, a method of obtaining a dirt probability map is introduced. The following specifically introduces a method of training a dirt detection model:

[0085] In this exemplary embodiment, the dirt detection model is trained by the following method:

[0086] Construct a sample set including several samples; wherein the samples include: sample data and label data; the sample data includes a normalized image for training; the label data includes a dirt probability map for training; the sample data is input into a dirt detection model to be trained to obtain prediction data output by the model, wherein the prediction data includes an observation probability map output by the model; determine the difference between the prediction data and the label data; based on the difference, update the parameters of the dirt detection model to be trained through a loss function until the difference between the prediction data and the label data is minimized, thereby obtaining the pre-trained dirt detection model.

[0087] In a specific implementation, a sample set including several samples is constructed; wherein the samples include: sample data and label data; the sample data includes a normalized image for training; the label data includes a dirt probability map for training; the sample data is input into a dirt detection model to be trained to obtain prediction data output by the model, wherein the prediction data includes an observation probability map output by the model; the difference between the prediction data and the label data is determined; based on the difference, the parameters of the dirt detection model to be trained are updated through a loss function until the difference between the prediction data and the label data is minimized, and obtaining the pre-trained dirt detection model means:

[0088] Construct a sample set containing multiple samples, each of which consists of a normalized image (sample data) and a corresponding dirt probability map (label data); these samples are used to train the dirt detection model; the normalized images in the sample set are input into the dirt detection model of the U-Net architecture, and the model processes each input image and outputs an observation probability map, i.e., predicted data; the difference between the observed probability map (predicted data) output by the model and the actual dirt probability map (label data) is calculated, which is usually quantified by a loss function; a weighted binary cross entropy loss function is used To measure the difference between the predicted probability and the true label, and update the model parameters based on this difference. and It is used to balance the sample ratio of dirty and non-dirty categories to deal with the problem of category imbalance, where the binary cross entropy loss function It can be expressed as:

[0089]

[0090] in, is the sample size, For the The true label of the sample, 1 means dirty, 0 means not dirty, For the The probability predicted by the model for a sample indicates the probability predicted by the model that the sample belongs to the positive class. The range of this probability value is [0,1]. and is the weight coefficient, usually set to =0.75 and = 0.25 to alleviate class imbalance;

[0091] Through the iterative training process, the model parameters are continuously updated until the difference between the predicted data and the label data is minimized, that is, the loss function value converges to a smaller value, at which time a pre-trained dirt detection model is obtained.

[0092] Step S220: judging the dirt probability map based on a set first threshold to obtain a dirt area.

[0093] In a specific implementation, the dirt probability map is judged based on the set first threshold to obtain the dirt area:

[0094] Binarize the probability map output by the model and set a threshold , when the pixel value is greater than the threshold, it is marked as a dirty area , expressed as:

[0095]

[0096] in, For the model at pixel The predicted probability of contamination at is the pixel value of the image after binarization.

[0097] Step S230: determine the geometric characteristic parameters of the dirty area, and transform the geometric characteristic parameters based on the transformation matrix to obtain the physical coordinates of the dirty area.

[0098] In a specific implementation, the geometric feature parameters of the dirty area refer to: a boundary box of the dirty area and a center of gravity position of the dirty area.

[0099] In a specific implementation, the geometric characteristic parameters of the dirty area are determined, and the physical coordinates of the dirty area are obtained by converting the geometric characteristic parameters based on the conversion matrix:

[0100] Scan the binary image and mark all connected dirty pixel blocks, expressed as:

[0101]

[0102] in, For the identified Dirty areas, is the number of dirty areas;

[0103] Calculate the area of ​​each dirty area. The area is the number of pixels in the area. The expression is:

[0104]

[0105] in, For the The area of ​​the dirty area;

[0106] Calculate the bounding box of each dirty area and center of gravity , the expression is:

[0107]

[0108] in, For the The bounding box of the dirty area, For the The center of gravity of the dirty area;

[0109] The centroid coordinates of each dirty area and bounding box Converted into actual physical coordinates in the photovoltaic panel coordinate system, expressed as:

[0110]

[0111] in, is the transformation matrix from the image coordinate system to the physical coordinate system, is the actual physical coordinate of the center of gravity of the dirty area; The actual physical coordinates of the dirty area bounding box.

[0112] Step S240: determining the dirty region characteristic information of the dirty region, the total area and the diagonal distance of the photovoltaic panel surface, and performing calculation based on the dirty region characteristic information, the total area and the diagonal distance to obtain a first severity score.

[0113] In a specific implementation, the characteristic information of the dirty area refers to: the area of ​​the dirty area, the type of dirt, and the distance from the center of gravity of the dirty area to the center of the photovoltaic panel.

[0114] In a specific implementation, a method of obtaining a first severity score by performing calculation based on the characteristic information of the dirty area, the total area and the diagonal distance is as follows:

[0115] Calculate a severity score for each soiled area based on its size, type and location (i.e. the first severity score), the scoring formula is as follows:

[0116]

[0117] in, For the Severity rating of dirty areas, For the The area of ​​the dirty area, is the total area of ​​the photovoltaic panels, is the dirt type weight, The center of gravity of the dirty area The distance to the center of the photovoltaic panel, is the diagonal distance of the photovoltaic panel, , , To adjust the parameters;

[0118] Based on the calculated severity score , sort all dirty areas and clean the areas with the highest scores first.

[0119] Step S250, dividing the photovoltaic panel surface into a plurality of grid units; based on the physical coordinates, mapping the dirty area to the corresponding grid unit, and obtaining a dirty area identifier corresponding to each grid unit.

[0120] In a specific implementation, the photovoltaic panel surface is divided into a plurality of grid units:

[0121] The surface of the photovoltaic panel is divided into several grid units. Each grid unit corresponds to a rectangular area of ​​fixed size. The granularity of the grid division should be set according to the size of the photovoltaic panel and the cleaning range of the robot. Assume that the physical size of the photovoltaic panel is × , the size of each grid cell is × , then the size of the grid map is × , expressed as:

[0122]

[0123] in, Indicates the number of grids in the width direction of the grid map. Indicates the number of grids in the height direction of the grid map.

[0124] Step S260: Calculate based on the first severity score and the dirty area identifier corresponding to each of the grid units to obtain a second severity score corresponding to each of the grid units.

[0125] In a specific implementation, a calculation is performed based on the first severity score and the dirty area identifier corresponding to each grid unit to obtain a second severity score corresponding to each grid unit:

[0126] The physical location of each dirty area will be obtained and severity rating Mapped to the corresponding grid cells, for each grid cell, calculate its dirtiness severity score (i.e. the second severity score), expressed as:

[0127]

[0128] in, For grid cells The severity score, For grid cells A collection of all dirty region indexes within .

[0129] Step S270: Determine the position information of each of the grid cells, perform path planning based on the position information and the second severity score, and obtain an optimized cleaning path.

[0130] In this exemplary embodiment, path planning is performed based on the position information and the second severity score to obtain an optimized cleaning path, including:

[0131] The position information is calculated to obtain edge weight information; based on the edge weight information, the total path weight information is obtained; the second severity corresponding to each of the grid units is scored, and accessed in order from large to small until each grid unit with the second severity score greater than the set second threshold is covered, so as to obtain a cleaning path plan; based on the total path weight information and the cleaning path plan, an initial path plan is obtained; based on the Bezier curve, the initial path plan is smoothed, and the energy consumption of the initial path plan is calculated to obtain energy consumption weight information; based on the total path weight information and the energy consumption weight information, the initial path plan is optimized to obtain the optimized cleaning path.

[0132] In a specific implementation, the location information is calculated to obtain edge weight information; based on the edge weight information, the total path weight information is obtained in the following manner:

[0133] Each grid cell in the grid map As a node in the graph, define To the adjacent node The edge weight is the Euclidean distance between the two, expressed as:

[0134]

[0135] in, For Node The physical coordinates of For Node The physical coordinates of

[0136] Calculate the total weight of the path, the expression is:

[0137]

[0138] in, is the total weight of the path, is the number of nodes passed in the path, Indicates from position Move to location The required edge weights, and Respectively represent two adjacent nodes on the path.

[0139] In a specific implementation, the second severity corresponding to each of the grid cells is scored, and the cells are visited in descending order until each grid cell whose second severity score is greater than the set second threshold is covered, and a cleaning path plan is obtained; based on the path weight information and the cleaning path plan, an initial path plan is obtained:

[0140] Starting from the current position of the robot, the high-priority nodes in the grid map are visited in turn, that is, those grid cells whose second severity score is higher than the set threshold are selected. These grid cells will be included in the cleaning path planning until all severity scores are covered. For grid cells larger than a certain threshold, the path weight information is combined with the clean path plan to obtain the initial path plan. This plan considers the order of visiting all high-priority grid cells from the current position and the weight of each path segment.

[0141] In a specific implementation, the initial path planning is smoothed based on the Bezier curve, and the energy consumption of the initial path planning is calculated to obtain the energy consumption weight information:

[0142] Starting from the robot's current position, visit high priority nodes in the grid map in sequence until all severity scores are covered Grid cells larger than a certain threshold;

[0143] Use Bezier curves to smooth the path, introduce an energy consumption model, and estimate the energy consumption on each path. (i.e., energy consumption weight information) and incorporate it into the calculation of path weight, the expression is:

[0144]

[0145] in, For robot slave nodes Move to Node Energy required For the robot at the node The energy required for cleaning.

[0146] In a specific implementation, the initial path planning is optimized based on the path weight information and the energy consumption weight information to obtain the optimized cleaning path, which means:

[0147] The optimization goal is to minimize the weighted energy consumption of the path, which can be expressed as:

[0148]

[0149] in, is the energy consumption weight coefficient.

[0150] Step S280, real-time data collection is performed on the surface of the photovoltaic panel to obtain a real-time collected image; in response to determining that new dirty area information is detected based on the dirt detection model and / or the change in the first severity score exceeds a set third threshold, the optimized cleaning path is adjusted to obtain a dynamically optimized path.

[0151] In a specific implementation, real-time acquisition is performed on the surface of the photovoltaic panel to obtain a real-time acquired image; in response to determining that the newly added dirty area information and / or the change in the first severity score detected based on the dirt detection model exceeds a set third threshold, it means:

[0152] While the robot is performing cleaning tasks, the camera continuously collects images of the photovoltaic panel surface and updates the dirt information in real time through the dirt detection model; if a new dirty area is detected or the first severity score of an existing dirty area exceeds the set threshold, the current path needs to be dynamically adjusted.

[0153] In the above exemplary embodiment, it is introduced that the surface of the photovoltaic panel is collected in real time to obtain a real-time collected image; in response to determining that the newly added dirty area information and / or the change in the first severity score detected based on the dirt detection model exceeds the set third threshold, the following describes a method for obtaining a dynamic optimization path:

[0154] In this exemplary embodiment, the optimized cleaning path is adjusted to obtain a dynamically optimized path, including:

[0155] Determine the first severity score and the newly added physical coordinates corresponding to the newly added dirty area information, and based on the newly added physical coordinates, map the dirty area to the corresponding grid unit to obtain the newly added dirty area identification; perform calculations based on the first severity score corresponding to the newly added dirty area information and the newly added dirty area identification to obtain a second severity score of the grid unit corresponding to the newly added dirty area identification; determine the position information of the cleaning robot, and adjust the optimized cleaning path based on the second severity score of the grid unit corresponding to the newly added dirty area identification to obtain the dynamically optimized path.

[0156] In a specific implementation, determining the first severity score and the newly added physical coordinates corresponding to the newly added dirty area information, mapping the dirty area to the corresponding grid unit based on the newly added physical coordinates, and obtaining the newly added dirty area identifier; performing calculation based on the first severity score corresponding to the newly added dirty area information and the newly added dirty area identifier, and obtaining the second severity score of the grid unit corresponding to the newly added dirty area identifier; determining the position information of the cleaning robot, and adjusting the optimized cleaning path based on the second severity score of the grid unit corresponding to the newly added dirty area identifier, and obtaining the dynamic optimization path means:

[0157] Through image acquisition and analysis, newly added dirty areas are determined, and the first severity scores of these areas are calculated. At the same time, the geometric feature parameters of these areas are converted into physical coordinates. Using the physical coordinates of the newly added dirty areas, these areas are mapped to the grid cells of the photovoltaic panel, so as to obtain the newly added dirty area identification. According to the first severity score of the newly added dirty area and the corresponding grid cell identification, the second severity score of the grid cell corresponding to these newly added dirty areas is calculated, and the current position of the cleaning robot is determined. Based on the second severity score of the grid cell corresponding to the newly added dirty area identification, the impact on the current optimized cleaning path is evaluated. If the severity score of the newly added dirty area exceeds the preset threshold or has a significant impact on the cleaning efficiency, the optimized cleaning path will be locally adjusted to ensure that the new dirty area is included in the cleaning plan. If the newly added dirty area needs to be cleaned, consider inserting it into the best position in the path. This is usually to find the nearest point in the path and use it as the insertion point to minimize the additional travel; if the existing path contains the changed dirty area, the cleaning order of these areas is reordered according to the new severity score. Use path optimization algorithms, such as the Dijkstra algorithm or the A* algorithm, for the affected path segments to find the shortest or optimal path connecting the affected areas. Avoid changing the unaffected path segments as much as possible to reduce the amount of calculation and the complexity of the adjustment. Smooth the adjusted path and use methods such as Bezier curves to reduce sharp turns and unnecessary returns in the path. When adjusting the path, consider the energy consumption model to ensure that the path adjustment does not significantly increase the energy consumption of the robot, and ensure that all dirty areas with severity scores exceeding the threshold are included in the cleaning path.

[0158] Furthermore, if local adjustments cannot significantly improve path performance, global path replanning is performed to regenerate the optimal path covering all high-priority dirty areas; after each path adjustment, the robot's travel path is updated, and the robot moves along the updated path, performing cleaning operations in each dirty area; at each cleaning point, the cleaning effect is monitored in real time. If the cleaning effect is not ideal, the area is remarked and included in the subsequent path planning.

[0159] In the above exemplary embodiment, a dynamic optimization path is introduced. The following describes a method for obtaining a power adaptation path after obtaining the dynamic optimization path:

[0160] In this exemplary embodiment, after obtaining the dynamic optimization path, the method further includes:

[0161] Determine the displacement distance between the grid units to be worked on the surface of the photovoltaic panel, the cleaning area of ​​the grid units to be worked, the second severity score corresponding to the grid units to be worked, the moving energy consumption of the cleaning robot and the cleaning energy consumption of the cleaning robot, and calculate based on the displacement distance between the grid units to be worked, the cleaning area of ​​the grid units to be worked, the second severity score corresponding to the grid units to be worked, the moving energy consumption of the cleaning robot and the cleaning energy consumption of the cleaning robot to obtain a task energy demand estimate; determine the number information of the grid units to be worked on the surface of the photovoltaic panel, and calculate based on the number information of the grid units to be worked, the moving energy consumption of the cleaning robot and the cleaning energy consumption of the cleaning robot to obtain the power demand of the grid units to be worked; determine the current remaining power of the cleaning robot and the homeward power demand of the last working grid unit, and in response to determining that the remaining power is less than the combined power demand of the grid unit to be worked and the homeward power demand, adjust the dynamic optimization path based on the task energy demand estimate to obtain a power adaptation path.

[0162] In a specific implementation, the displacement distance between the grid units to be worked on the surface of the photovoltaic panel, the cleaning area of ​​the grid units to be worked, the second severity score corresponding to the grid units to be worked, the moving energy consumption of the cleaning robot and the cleaning energy consumption of the cleaning robot are determined, and the task energy demand estimate is obtained based on the displacement distance between the grid units to be worked, the cleaning area of ​​the grid units to be worked, the second severity score corresponding to the grid units to be worked, the moving energy consumption of the cleaning robot and the cleaning energy consumption of the cleaning robot; the number information of the grid units to be worked on the surface of the photovoltaic panel is determined, and the power demand of the grid units to be worked is obtained based on the number information of the grid units to be worked, the moving energy consumption of the cleaning robot and the cleaning energy consumption of the cleaning robot; the current remaining power of the cleaning robot and the homing power demand of the last working grid unit are determined, and in response to determining that the remaining power is less than the combined power demand of the grid unit to be worked and the homing power demand:

[0163] Determine the displacement distance of each grid unit to be operated , Cleaning area and second severity score , use the power consumption model to estimate the mobile energy consumption of each grid unit to be operated and clean energy consumption , for each grid cell to be worked on, estimate the energy requirements of future tasks , the expression is,

[0164]

[0165] in, is the moving distance of the robot to the next task point, Clean the area for the next task point, Score the severity of the dirtiness of the next task point, is the energy consumption for moving distance, To clean energy consumption.

[0166] For the entire cleaning path, calculate the total power required to perform all tasks , the expression is,

[0167]

[0168] in, is the total number of task points in the cleaning path, From the mission point −1 to the mission point Mobile power demand, For mission points Clean electricity demand;

[0169] Before executing a task, Estimated total power demand For comparison, if , then execute the predetermined complete cleaning path. If , then enter the dynamic path adjustment process;

[0170] The power required to return to the charging station from the last mission point.

[0171] In a specific implementation, the dynamic optimization path is adjusted based on the estimated energy demand of the task to obtain a power adaptation path:

[0172] According to the severity score of dirtiness, all uncleaned areas are prioritized, and the cleaning path is adjusted to remove low-priority areas or shorten the path length to reduce energy consumption. The path is smoothed using methods such as Bezier curves to reduce sharp turns and reduce energy consumption, ensuring that the adjusted path enables the robot to return safely to the charging station.

[0173] Based on the above exemplary embodiments, it is not only possible to effectively identify dirty areas and their distribution, but also to dynamically adjust the cleaning strategy according to the actual situation. Based on the dynamically planned cleaning path, it is possible to ensure that the key dirty areas are cleaned first, improve the cleaning effect and energy utilization efficiency of the photovoltaic panels, and also to dynamically allocate power through the real-time power monitoring function to avoid task interruption due to insufficient power. Specific: In order to verify the effectiveness of the control method of the photovoltaic cleaning robot described in this scheme, a comparative experiment was carried out. The experimental site selected a photovoltaic power station with an area of ​​100 square meters, and the total area of ​​its photovoltaic panels was 80 square meters, with different degrees of dirty areas. Two photovoltaic cleaning robots were used in the experiment, one of which adopted the traditional fixed path cleaning method (called type A robot), and the other robot adopted the dynamic cleaning path planning method based on visual recognition of the present invention (called type B robot). The main goal of the experiment is to compare the performance differences between the two methods in terms of dirt recognition accuracy, cleaning coverage, power consumption and cleaning efficiency.

[0174] First, a high-resolution camera was installed on the surface of the photovoltaic panel, and the position and angle of the camera were precisely adjusted to ensure that the image acquisition could cover the entire surface of the photovoltaic panel. Secondly, the collected images were preprocessed, including denoising, contrast enhancement, perspective correction, etc., to improve the clarity and reliability of the image. Subsequently, the dirt detection module deployed on the B-type robot, based on the U-Net architecture, a deep learning model of convolutional neural networks, analyzed the image, identified the dirty area in real time, and calculated its location and area. By calculating the severity score of the dirty area, the B-type robot planned a dynamic cleaning path, monitored the battery power in real time, and dynamically allocated power in combination with path planning to ensure that there is still enough power to return to the charging station after completing all cleaning tasks.

[0175] The experiment was conducted for 10 cleaning cycles, each lasting 1 hour. After each cycle, the cleaning effect of the photovoltaic panels was evaluated, and the accuracy of dirt recognition, cleaning coverage, energy consumption per unit area, and cleaning efficiency were calculated; the performance comparison of type A and type B robots is shown in Table 1:

[0176] Table 1 Performance comparison of photovoltaic cleaning robots

[0177]

[0178] Through the analysis of experimental data, it can be clearly seen that the photovoltaic cleaning robot control method of this scheme is superior to the traditional fixed path cleaning method in all performance indicators. In terms of dirt recognition accuracy, the B-type robot reached 92%, which is much higher than the 75% of the A-type robot. This shows that the image preprocessing and deep learning model of this scheme can more accurately identify the dirty areas on the surface of the photovoltaic panel, effectively avoiding the common false detection and missed detection phenomena in traditional methods. This high-accuracy dirt recognition can ensure that the cleaning robot can clean more targeted and avoid ineffective cleaning actions.

[0179] Secondly, in terms of cleaning coverage, the B-type robot reached 98%, which is much higher than the 85% of the A-type robot. This shows that the method of this scheme can more comprehensively cover the dirty areas on the surface of the photovoltaic panel and maximize the power generation efficiency of the photovoltaic panel. The traditional fixed-path cleaning method lacks real-time perception of the distribution of dirt, and is prone to miss some difficult-to-detect dirty areas, resulting in a low cleaning coverage rate.

[0180] In terms of energy consumption per unit area, the B-type robot is only 8 Wh / ㎡, which is significantly lower than the 12 Wh / ㎡ of the A-type robot. This is mainly due to the dynamic optimization of the cleaning path and the reasonable allocation of electricity in this solution method. By dynamically adjusting the cleaning path, the B-type robot can avoid repeated cleaning and ineffective movement, thereby reducing energy consumption. Traditional methods often have unnecessary path redundancy due to the use of fixed paths, resulting in energy waste.

[0181] Finally, in terms of cleaning efficiency, the B-type robot reached 75㎡ / h, which is 25% higher than the 60㎡ / h of the A-type robot. This means that in the same amount of time, the method of this solution can clean more photovoltaic panel area, thereby improving overall operating efficiency. Efficient cleaning can not only save time and labor costs, but also improve the power generation efficiency of photovoltaic power stations.

[0182] In summary, the photovoltaic cleaning robot control method of this scheme significantly improves the accuracy of dirt recognition and cleaning efficiency while reducing energy consumption by introducing advanced image processing technology and intelligent path planning algorithm.

[0183] It should be noted that the method of the embodiment of the present disclosure can be performed by a single device, such as a computer or a server. The method of the present embodiment can also be applied in a distributed scenario and completed by multiple devices cooperating with each other. In the case of such a distributed scenario, one of the multiple devices can only perform one or more steps in the method of the embodiment of the present disclosure, and the multiple devices will interact with each other to complete the described method.

[0184] It should be noted that the above describes some embodiments of the present disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the above embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0185] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the present disclosure also provides a control device for a photovoltaic cleaning robot.

[0186] refer to Figure 3 , the control device of the photovoltaic cleaning robot comprises:

[0187] The dirt probability map determination module 310 is configured to determine image information of the surface of the photovoltaic panel, detect the image information through a pre-trained dirt detection model, and obtain a dirt probability map;

[0188] A dirt region determination module 320 is configured to determine the dirt probability map based on a set first threshold value to obtain a dirt region;

[0189] A physical coordinate determination module 330 is configured to determine geometric characteristic parameters of the dirty area, and transform the geometric characteristic parameters based on a transformation matrix to obtain physical coordinates of the dirty area;

[0190] A first score determination module 340 is configured to determine the dirty region characteristic information of the dirty region, the total area and the diagonal distance of the photovoltaic panel surface, and calculate based on the dirty region characteristic information, the total area and the diagonal distance to obtain a first severity score;

[0191] The dirty area identification determination module 350 is configured to divide the photovoltaic panel surface into a plurality of grid units; based on the physical coordinates, map the dirty area to the corresponding grid unit to obtain a dirty area identification corresponding to each grid unit;

[0192] A second score determination module 360 ​​is configured to calculate based on the first severity score and the dirty area identifier corresponding to each grid unit to obtain a second severity score corresponding to each grid unit;

[0193] A cleaning path determination module 370 is configured to determine the position information of each of the grid cells, perform path planning based on the position information and the second severity score, and obtain an optimized cleaning path;

[0194] The dynamic optimization path determination module 380 is configured to perform real-time acquisition of the surface of the photovoltaic panel to obtain a real-time acquired image; in response to determining that the newly added dirty area information and / or the change in the first severity score detected based on the dirt detection model exceeds a set third threshold, the optimized cleaning path is adjusted to obtain a dynamically optimized path.

[0195] In this exemplary embodiment, the dirt probability map determination module 310 is specifically configured as follows:

[0196] Determine the image information of the surface of the photovoltaic panel, perform denoising, contrast enhancement, perspective correction, image segmentation and feature extraction on the image information to obtain a preprocessed image; perform image adjustment and normalization on the preprocessed image to obtain a normalized image; input the normalized image into a pre-trained dirt detection model to obtain the dirt probability map; and train the dirt detection model by the following method: construct a sample set including several samples; wherein the sample includes: sample data and label data; the sample data includes a normalized image for training; the label data includes a dirt probability map for training; input the sample data into the dirt detection model to be trained to obtain the predicted data output by the model, wherein the predicted data includes the observed probability map output by the model; determine the difference between the predicted data and the label data; based on the difference, update the parameters of the dirt detection model to be trained through a loss function until the difference between the predicted data and the label data is minimized to obtain the pre-trained dirt detection model.

[0197] In this exemplary embodiment, the dirty area determination module 320 is specifically configured as follows:

[0198] The dirt probability map is judged based on a set first threshold to obtain a dirt area.

[0199] In this exemplary embodiment, the physical coordinate determination module 330 is specifically configured as follows:

[0200] The geometric characteristic parameters of the dirty area are determined, and the physical coordinates of the dirty area are obtained by transforming the geometric characteristic parameters based on a transformation matrix.

[0201] In this exemplary embodiment, the first score determination module 340 is specifically configured to:

[0202] Determine dirty region characteristic information of the dirty region, the total area and the diagonal distance of the photovoltaic panel surface, and perform calculation based on the dirty region characteristic information, the total area and the diagonal distance to obtain a first severity score.

[0203] In this exemplary embodiment, the dirty area identification determination module 350 is specifically configured as follows:

[0204] The photovoltaic panel surface is divided into a plurality of grid units; based on the physical coordinates, the dirty area is mapped to the corresponding grid unit to obtain a dirty area identifier corresponding to each grid unit.

[0205] In this exemplary embodiment, the second score determination module 360 ​​is specifically configured to:

[0206] A calculation is performed based on the first severity score and the dirty area identifier corresponding to each of the grid units to obtain a second severity score corresponding to each of the grid units.

[0207] In this exemplary embodiment, the cleaning path determination module 370 is specifically configured as follows:

[0208] Determine the position information of each of the grid units, calculate the position information, and obtain edge weight information; obtain the total path weight information based on the edge weight information; score the corresponding second severity of each of the grid units, and visit them in order from large to small until each grid unit with the second severity score greater than the set second threshold is covered, so as to obtain a cleaning path plan; obtain an initial path plan based on the total path weight information and the cleaning path plan; perform path smoothing on the initial path plan based on a Bezier curve, and calculate the energy consumption of the initial path plan to obtain energy consumption weight information; optimize the initial path plan based on the total path weight information and the energy consumption weight information to obtain the optimized cleaning path.

[0209] In this exemplary embodiment, the dynamic optimization path determination module 380 is specifically configured as follows:

[0210] The photovoltaic panel surface is collected in real time to obtain a real-time collected image; in response to determining that the newly added dirty area information and / or the first severity score detected based on the dirt detection model exceeds a set third threshold, the first severity score and the newly added physical coordinates corresponding to the newly added dirty area information are determined, and based on the newly added physical coordinates, the dirty area is mapped to the corresponding grid unit to obtain a newly added dirty area identifier; calculation is performed based on the first severity score corresponding to the newly added dirty area information and the newly added dirty area identifier to obtain a second severity score of the grid unit corresponding to the newly added dirty area identifier; the position information of the cleaning robot is determined, and the optimized cleaning path is adjusted based on the second severity score of the grid unit corresponding to the newly added dirty area identifier to obtain the dynamic optimization path.

[0211] For the convenience of description, the above device is described by dividing it into various modules according to its functions. Of course, when implementing the present disclosure, the functions of each module can be implemented in the same or multiple software and / or hardware.

[0212] The device of the above embodiment is used to implement the control method of the corresponding photovoltaic cleaning robot in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.

[0213] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments and methods, the present disclosure also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the control method of the photovoltaic cleaning robot described in any of the above embodiments is implemented.

[0214] Figure 4 A more specific schematic diagram of the hardware structure of an electronic device provided in this embodiment is shown, and the device may include: a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, the memory 1020, the input / output interface 1030, and the communication interface 1040 are connected to each other through the bus 1050 in the device.

[0215] The processor 1010 can be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0216] The memory 1020 may be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 1020 may store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program codes are stored in the memory 1020 and are called and executed by the processor 1010.

[0217] The input / output interface 1030 is used to connect the input / output module to realize information input and output. The input / output module can be configured in the device as a component (not shown in the figure), or it can be externally connected to the device to provide corresponding functions. The input device may include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output device may include a display, a speaker, a vibrator, an indicator light, etc.

[0218] The communication interface 1040 is used to connect a communication module (not shown) to realize communication interaction between the device and other devices. The communication module can realize communication through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0219] The bus 1050 includes a path that transmits information between the various components of the device (eg, the processor 1010 , the memory 1020 , the input / output interface 1030 , and the communication interface 1040 ).

[0220] It should be noted that, although the above device only shows the processor 1010, the memory 1020, the input / output interface 1030, the communication interface 1040 and the bus 1050, in the specific implementation process, the device may also include other components necessary for normal operation. In addition, it can be understood by those skilled in the art that the above device may also only include the components necessary for implementing the embodiments of the present specification, and does not necessarily include all the components shown in the figure.

[0221] The electronic device of the above embodiment is used to implement the control method of the corresponding photovoltaic cleaning robot in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.

[0222] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the present disclosure also provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the control method of the photovoltaic cleaning robot as described in any of the above embodiments.

[0223] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0224] The above-mentioned non-transitory computer-readable storage medium can be any available medium or data storage device that can be accessed by a computer, including but not limited to magnetic storage (such as floppy disks, hard disks, magnetic tapes, magneto-optical disks (MO), etc.), optical storage (such as CD, DVD, BD, HVD, etc.), and semiconductor storage (such as ROM, EPROM, EEPROM, non-volatile memory (NAND FLASH), solid-state drive (SSD)), etc.

[0225] The computer instructions stored in the storage medium of the above embodiment are used to enable the computer to execute the control method of the photovoltaic cleaning robot as described in any embodiment in the above exemplary method part, and have the beneficial effects of the corresponding method embodiment, which will not be repeated here.

[0226] Based on the same inventive concept, corresponding to the control method of the photovoltaic cleaning robot described in any of the above embodiments, the present disclosure also provides a computer program product, which includes computer program instructions. In some embodiments, the computer program instructions can be executed by one or more processors of a computer so that the computer and / or the processor execute the control method of the photovoltaic cleaning robot. Corresponding to the execution subject corresponding to each step in each embodiment of the control method of the photovoltaic cleaning robot, the processor that executes the corresponding step may belong to the corresponding execution subject.

[0227] The computer program product of the above embodiment is used to enable the computer and / or the processor to execute the control method of the photovoltaic cleaning robot as described in any of the above embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0228] Those skilled in the art will appreciate that the embodiments of the present disclosure may be implemented as a system, method, or computer program product. Therefore, the present disclosure may be specifically implemented in the following forms, namely: complete hardware, complete software (including firmware, resident software, microcode, etc.), or a combination of hardware and software, generally referred to herein as a "circuit," "module," or "system." In addition, in some embodiments, the present disclosure may also be implemented in the form of a computer program product in one or more computer-readable media, which contains computer-readable program code.

[0229] Any combination of one or more computer-readable media may be used. A computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. More specific examples (non-exhaustive examples) of computer-readable storage media may include, for example: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device, or device.

[0230] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0231] The program code embodied on the computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0232] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0233] It should be understood that each box in the flowchart and / or block diagram and the combination of boxes in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer or other programmable data processing device to produce a machine, and these computer program instructions are executed by a computer or other programmable data processing device to produce a device that implements the functions / operations specified in the boxes in the flowchart and / or block diagram.

[0234] These computer program instructions may also be stored in a computer-readable medium that enables a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable medium produce a product that includes an instruction device that implements the functions / operations specified in the blocks in the flowchart and / or block diagram.

[0235] Computer program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device so that a series of operational steps are performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby enabling the instructions executed on the computer or other programmable device to provide a process for implementing the functions / operations specified in the blocks in the flowchart and / or block diagram.

[0236] In addition, although the operations of the disclosed method are described in a particular order in the accompanying drawings, this does not require or imply that the operations must be performed in this particular order, or that all the operations shown must be performed to achieve the desired results. On the contrary, the steps depicted in the flow chart can be performed in a different order. Additionally or alternatively, some steps can be omitted, multiple steps can be combined into one step, and / or one step can be decomposed into multiple steps.

[0237] The flowchart and block diagram in the accompanying drawings illustrate the possible architecture, functions and operations of the system, method and computer program product according to various embodiments of the present application. Wherein, each box in the flowchart or block diagram can represent a module, a program segment, or a part of the code, and the above-mentioned module, program segment, or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0238] It should be noted that, although several modules or units of the equipment for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present application, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into being embodied by multiple modules or units.

[0239] A person skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present application (including the claims) is limited to these examples. In line with the concept of the present application, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the embodiments of the present application as described above, which are not provided in detail for the sake of simplicity.

[0240] In addition, to simplify the description and discussion, and in order not to make the embodiments of the present application difficult to understand, the known power / ground connections to the integrated circuit (IC) chip and other components may or may not be shown in the provided drawings. In addition, the device may be shown in the form of a block diagram to avoid making the embodiments of the present application difficult to understand, and this also takes into account the fact that the details of the implementation of these block diagram devices are highly dependent on the platform on which the embodiments of the present application are to be implemented (that is, these details should be fully within the scope of understanding of those skilled in the art). Where specific details (e.g., circuits) are set forth to describe exemplary embodiments of the present application, it is obvious to those skilled in the art that the embodiments of the present application can be implemented without these specific details or with changes in these specific details. Therefore, these descriptions should be considered illustrative rather than restrictive.

[0241] Although the present application has been described in conjunction with specific embodiments of the present application, many alternatives, modifications and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may use the discussed embodiments.

[0242] The embodiments of the present application are intended to cover all such substitutions, modifications and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present application should be included in the scope of protection of the present application.

[0243] Although the spirit and principle of the present disclosure have been described with reference to several specific embodiments, it should be understood that the present disclosure is not limited to the disclosed specific embodiments, and the division of various aspects does not mean that the features in these aspects cannot be combined to benefit, and such division is only for the convenience of expression. The present disclosure is intended to cover various modifications and equivalent arrangements included in the spirit and scope of the attached claims. The scope of the attached claims conforms to the broadest interpretation, thereby including all such modifications and equivalent structures and functions.

Claims

1. A control method for a photovoltaic cleaning robot, characterized in that: include: Determine image information of the surface of the photovoltaic panel, and detect the image information using a pre-trained dirt detection model to obtain a dirt probability map; Judging the dirt probability map based on a set first threshold value to obtain a dirt area; Determine geometric characteristic parameters of the dirty area, and transform the geometric characteristic parameters based on a transformation matrix to obtain physical coordinates of the dirty area; Determine dirty region characteristic information of the dirty region, the total area and the diagonal distance of the photovoltaic panel surface, and perform calculation based on the dirty region characteristic information, the total area and the diagonal distance to obtain a first severity score; Dividing the photovoltaic panel surface into a plurality of grid units; mapping the dirty area to the corresponding grid unit based on the physical coordinates to obtain a dirty area identifier corresponding to each grid unit; Calculating based on the first severity score and the dirty area identifier corresponding to each of the grid units to obtain a second severity score corresponding to each of the grid units; Determining the position information of each of the grid cells, and performing path planning based on the position information and the second severity score to obtain an optimized cleaning path; Performing real-time acquisition of the surface of the photovoltaic panel to obtain a real-time acquired image; In response to determining that newly added dirty area information is detected based on the dirt detection model and / or a change in the first severity score exceeds a set third threshold, the optimized cleaning path is adjusted to obtain a dynamically optimized path.

2. The method according to claim 1, characterized in that The performing path planning based on the position information and the second severity score to obtain an optimized cleaning path includes: Calculating the position information to obtain edge weight information; Based on the edge weight information, obtaining path total weight information; Score the second severity corresponding to each of the grid cells, and visit them in order from large to small until each grid cell whose second severity score is greater than a set second threshold is covered, thereby obtaining a cleaning path plan; Based on the total path weight information and the cleaning path planning, an initial path planning is obtained; Performing path smoothing processing on the initial path planning based on a Bezier curve, and calculating the energy consumption of the initial path planning to obtain energy consumption weight information; The initial path planning is optimized based on the total path weight information and the energy consumption weight information to obtain the optimized cleaning path.

3. The method according to claim 1, characterized in that The step of adjusting the optimized cleaning path to obtain a dynamically optimized path includes: Determine the first severity score and the newly added physical coordinates corresponding to the newly added dirty area information, and map the dirty area to the corresponding grid unit based on the newly added physical coordinates to obtain a newly added dirty area identifier; Calculating based on the first severity score corresponding to the newly added dirty area information and the newly added dirty area identifier, obtaining a second severity score of the grid unit corresponding to the newly added dirty area identifier; The position information of the cleaning robot is determined, and the optimized cleaning path is adjusted based on the second severity score of the grid unit corresponding to the newly added dirty area identifier to obtain the dynamic optimized path.

4. The method according to claim 1, characterized in that After obtaining the dynamic optimization path, the method further includes: Determine the displacement distance between grid units to be operated on the surface of the photovoltaic panel, the cleaning area of ​​the grid units to be operated, the second severity score corresponding to the grid units to be operated, the moving energy consumption of the cleaning robot, and the cleaning energy consumption of the cleaning robot, and calculate based on the displacement distance between the grid units to be operated, the cleaning area of ​​the grid units to be operated, the second severity score corresponding to the grid units to be operated, the moving energy consumption of the cleaning robot, and the cleaning energy consumption of the cleaning robot to obtain a task energy demand estimate; Determine the number of grid units to be operated on the surface of the photovoltaic panel, and calculate based on the number of grid units to be operated, the moving energy consumption of the cleaning robot, and the cleaning energy consumption of the cleaning robot to obtain the power demand of the grid units to be operated; Determine the current remaining power of the cleaning robot and the homing power requirement of the last working grid unit. In response to determining that the remaining power is less than the combined power requirement of the grid unit to be worked and the homing power requirement, adjust the dynamic optimization path based on the task energy demand estimate to obtain a power adaptation path.

5. The method according to claim 1, characterized in that: The detecting the image information by using a pre-trained dirt detection model to obtain a dirt probability map includes: De-noising, contrast enhancement, perspective correction, image segmentation and feature extraction are performed on the image information to obtain a pre-processed image; Performing image adjustment and normalization processing on the preprocessed image to obtain a normalized image; The normalized image is input into a pre-trained dirt detection model to obtain the dirt probability map.

6. The method according to claim 5, characterized in that The method further comprises training the dirt detection model by: Constructing a sample set including several samples; wherein the samples include: sample data and label data; the sample data includes a normalized image for training; the label data includes a dirt probability map for training; Inputting the sample data into a dirt detection model to be trained to obtain prediction data output by the model, wherein the prediction data includes an observation probability map output by the model; Determining a difference between the predicted data and the label data; Based on the difference, the parameters of the dirt detection model to be trained are updated through a loss function until the difference between the predicted data and the label data is minimized, thereby obtaining the pre-trained dirt detection model.

7. A control device for a photovoltaic cleaning robot, characterized in that: include: A dirt probability map determination module is configured to determine image information of the surface of the photovoltaic panel, and detect the image information through a pre-trained dirt detection model to obtain a dirt probability map; a dirty area determination module, configured to determine the dirt probability map based on a set first threshold value to obtain a dirty area; A physical coordinate determination module is configured to determine geometric characteristic parameters of the dirty area, and transform the geometric characteristic parameters based on a transformation matrix to obtain physical coordinates of the dirty area; A first score determination module is configured to determine the dirty region characteristic information of the dirty region, the total area and the diagonal distance of the photovoltaic panel surface, and calculate based on the dirty region characteristic information, the total area and the diagonal distance to obtain a first severity score; A dirty area identification determination module is configured to divide the surface of the photovoltaic panel into a plurality of grid units; based on the physical coordinates, the dirty area is mapped to the corresponding grid unit to obtain a dirty area identification corresponding to each grid unit; a second score determination module configured to calculate based on the first severity score and the dirty area identifier corresponding to each of the grid units to obtain a second severity score corresponding to each of the grid units; a cleaning path determination module, configured to determine position information of each of the grid cells, perform path planning based on the position information and the second severity score, and obtain an optimized cleaning path; A dynamic optimization path determination module is configured to collect data on the surface of the photovoltaic panel in real time to obtain a real-time collected image; In response to determining that newly added dirty area information is detected based on the dirt detection model and / or a change in the first severity score exceeds a set third threshold, the optimized cleaning path is adjusted to obtain a dynamically optimized path.

8. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method according to any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to cause a computer to execute the method according to any one of claims 1 to 6.

10. A computer program product, characterized in that The method comprises computer program instructions, which, when executed on a computer, cause the computer to execute the method according to any one of claims 1 to 6.

Images