Cleaning strategy determination method and device, storage medium and electronic device

Abstract

The invention discloses a cleaning strategy determination method and device, a storage medium and an electronic device.The method comprises the steps that an acquired visible light image of a photovoltaic panel is detected through a pre-trained first model, and smudginess area information on the photovoltaic panel is obtained; an infrared image of the photovoltaic panel is analyzed, smudginess degree information of the photovoltaic panel is obtained, and the infrared image and the visible light image are collected at the same time; and determining a cleaning strategy for the photovoltaic panel according to the smudginess area information and the smudginess degree information. The problem that the cleaning effect is poor due to the fact that a cleaning decision is made on the photovoltaic panel in a manual regular inspection mode is solved.

Description

Technical Field

[0001] The present application relates to the field of photovoltaic power generation technology, and in particular to a method and device for determining a cleaning strategy, a storage medium, and an electronic device. Background Art

[0002] In the long-term operation of photovoltaic power stations, the cleanliness of the photovoltaic panel surface is directly related to its power generation efficiency and service life. Photovoltaic panels are easily contaminated by dust, sand, leaves, bird droppings and other debris in outdoor environments. These debris not only block the area of ​​the photovoltaic panel that absorbs sunlight, reducing the photoelectric conversion efficiency, but also absorb the heat of solar energy, causing the temperature of the photovoltaic panel to rise abnormally. The increase in temperature has a negative impact on the performance of photovoltaic cells, and in severe cases it may even accelerate the aging and damage of the cells, thereby shortening the service life of the photovoltaic panels.

[0003] At present, the cleaning inspection of photovoltaic panels mostly relies on regular manual inspections. Based on experience, the inspectors judge whether the stains and dust on the surface of the photovoltaic panels have reached the level that affects the power generation efficiency, and then decide whether to carry out cleaning work. However, this method has significant limitations. First, photovoltaic power stations usually occupy a large area and have a large number of photovoltaic panels. It is difficult for manual inspections to cover all areas, which can easily cause blind spots and omissions in inspections. Secondly, the results of manual inspections are affected by the personal experience, skills and sense of responsibility of the inspectors, and the accuracy of recognition and cleaning decisions is difficult to guarantee. Furthermore, close-range manual inspections pose a threat to the health of inspectors under severe weather conditions (such as high temperature, thunderstorms, strong winds, etc.), increasing the risk of personnel safety. In addition, if the cleaning work is not in place or the inspection is not comprehensive, it will lead to frequent subsequent cleaning and inspections, significantly increasing the operation and maintenance costs of photovoltaic power stations.

[0004] In the related technology, cleaning decisions for photovoltaic panels are made through manual regular inspections, resulting in poor cleaning effects, and no effective solution has yet been found. Summary of the invention

[0005] The embodiments of the present application provide a method and device for determining a cleaning strategy, a storage medium, and an electronic device, so as to at least solve the problem in the prior art that cleaning decisions for photovoltaic panels are made through manual periodic inspections, resulting in poor cleaning effects.

[0006] According to one embodiment of the embodiments of the present application, a method for determining a cleaning strategy is provided, comprising: detecting a collected visible light image of a photovoltaic panel through a pre-trained first model to obtain dirt area information on the photovoltaic panel; analyzing an infrared image of the photovoltaic panel to obtain dirt degree information of the photovoltaic panel, wherein the infrared image is an image collected simultaneously with the visible light image; and determining a cleaning strategy for the photovoltaic panel based on the dirt area information and the dirt degree information.

[0007] In an exemplary embodiment, before detecting the collected visible light images of the photovoltaic panel through the pre-trained first model, the method also includes: acquiring historical visible light images of the photovoltaic panel; performing a labeling operation on the dirty area in the historical visible light image, wherein the labeling operation at least includes: drawing a bounding box of the dirty area and assigning a classification label to the dirty area; and training a second model through the historical visible light images after the labeling operation to obtain the first model, wherein the second model is an untrained neural network model for dirt detection.

[0008] In an exemplary embodiment, a collected visible light image of a photovoltaic panel is detected by a pre-trained first model to obtain dirty area information on the photovoltaic panel, including: determining a first dirty area in the visible light image by the first model; determining a region edge corresponding to the first dirty area by an edge segmentation algorithm; and determining the dirty area information by an image scale factor and pixel coordinates corresponding to the region edge corresponding to the first dirty area, wherein the image scale factor is an image scale factor of the visible light image obtained by the actual physical size corresponding to each pixel in the visible light image.

[0009] In an exemplary embodiment, the infrared image of the photovoltaic panel is analyzed to obtain the dirtiness information of the photovoltaic panel, including: extracting the area edge corresponding to the second dirty area of ​​the infrared image through an edge segmentation algorithm; determining the second dirty area and non-dirty area in the infrared image through the area edge corresponding to the second dirty area; determining the first temperature information of the second dirty area, and determining the second temperature information of the non-dirty area; determining the dirtiness information through the first temperature information and the second temperature information.

[0010] In an exemplary embodiment, the dirtiness information is determined by the first temperature information and the second temperature information, including: determining the difference between the first temperature information and the second temperature information as the dirtiness information; or assigning a first weight to the first temperature information through the first area information of the second dirty area, and assigning a second weight to the second temperature information through the second area information of the non-dirty area; determining the difference between the first product and the second product as the dirtiness information, wherein the first product is the product of the first weight and the first temperature information, and the second product is the product of the second weight and the second temperature information.

[0011] In an exemplary embodiment, a cleaning strategy for the photovoltaic panel is determined by the dirt area information and the dirt degree information, including one of the following: when the area value corresponding to the dirt area information is greater than a first threshold, determining that the cleaning strategy is that the photovoltaic panel needs to be cleaned; when the degree value corresponding to the dirt degree information is greater than a second threshold, determining that the cleaning strategy is that the photovoltaic panel needs to be cleaned; when the area value corresponding to the dirt area information is greater than the first threshold and the degree value corresponding to the dirt degree information is greater than the second threshold, determining that the cleaning strategy is that the photovoltaic panel must be cleaned.

[0012] According to another embodiment of the embodiment of the present application, a device for determining a cleaning strategy is also provided, including: a detection module, used to detect the collected visible light image of the photovoltaic panel through a pre-trained first model to obtain the dirt area information on the photovoltaic panel; an analysis module, used to analyze the infrared image of the photovoltaic panel to obtain the dirtiness level information of the photovoltaic panel, wherein the infrared image is an image collected simultaneously with the visible light image; a determination module, used to determine the cleaning strategy for the photovoltaic panel through the dirtiness area information and the dirtiness level information.

[0013] According to another aspect of the embodiments of the present application, a computer-readable storage medium is provided, in which a computer program is stored, wherein the computer program is configured to execute the above method when running.

[0014] According to another aspect of the embodiments of the present application, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the method through the computer program.

[0015] According to another aspect of the embodiments of the present application, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in any of the above method embodiments are implemented.

[0016] In an embodiment of the present application, the collected visible light image of the photovoltaic panel is detected by a pre-trained first model to obtain the dirt area information on the photovoltaic panel; the infrared image of the photovoltaic panel is analyzed to obtain the dirtiness information of the photovoltaic panel, wherein the infrared image is an image collected simultaneously with the visible light image; and the cleaning strategy for the photovoltaic panel is determined by the dirtiness area information and the dirtiness information. In other words, by identifying the dirtiness area information in the visible light image of the photovoltaic panel through the pre-trained first model and combining it with the dirtiness information identified from the infrared image of the photovoltaic panel, the cleaning strategy for the photovoltaic panel can be directly determined. Therefore, through the above-mentioned embodiment, the problem of poor cleaning effect caused by making cleaning decisions for photovoltaic panels through manual regular inspections in the prior art is solved, and the cleaning effect of the photovoltaic panels is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0018] Figure 1 It is a hardware structure block diagram of a computer terminal of a method for determining a cleaning strategy according to an embodiment of the present application;

[0019] Figure 2 is a flow chart of a method for determining a cleaning strategy according to an embodiment of the present application;

[0020] Figure 3 is another flow chart of a method for determining a cleaning strategy according to an embodiment of the present application;

[0021] Figure 4 is a schematic diagram of a dirty photovoltaic panel according to a method for determining a cleaning strategy in an embodiment of the present application;

[0022] Figure 5 is another flow chart of a method for determining a cleaning strategy according to an embodiment of the present application;

[0023] Figure 6 It is a structural block diagram of a device for determining a cleaning strategy according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0025] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products, or devices; "plurality" means two or more.

[0026] The method embodiments provided in the embodiments of the present application can be executed in a computer terminal or similar computing device or cloud platform or independent physical server or software platform, wherein the above software platform is run through one or more servers. Taking running on a computer terminal as an example, Figure 1 1 is a hardware structure block diagram of a computer terminal for determining a cleaning strategy according to an embodiment of the present application. Figure 1 As shown, the computer terminal may include one or more ( Figure 1 Only one is shown in the figure) processor 102 and memory 104 for storing data. In an exemplary embodiment, the computer terminal may also include a transmission device 106 for communication functions and an input and output device 108, wherein the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA. A person skilled in the art can understand that Figure 1 The structure shown is only for illustration and does not limit the structure of the above-mentioned computer terminal. Figure 1 More or fewer components as shown, or with Figure 1 Equivalent functions or comparisons shown Figure 1 A different configuration with more features is shown.

[0027] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as computer programs corresponding to the methods in the embodiments of the present application. The processor 102 executes various functional applications and data processing by running the computer programs stored in the memory 104, that is, the above-mentioned methods are implemented. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the computer terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0028] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of a computer terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0029] In this embodiment, a method for determining a cleaning strategy is provided, which is applied to the above-mentioned computer terminal. Figure 2 : is a flow chart of a method for determining a cleaning strategy according to an embodiment of the present application, the process comprising the following steps:

[0030] Step S202, detecting the collected visible light image of the photovoltaic panel by using a pre-trained first model to obtain dirt area information on the photovoltaic panel;

[0031] Optionally, the dirt on the photovoltaic panel includes stains and dust accumulation, more specifically, includes: dust, sand, leaves, bird droppings, etc.

[0032] Step S204, analyzing the infrared image of the photovoltaic panel to obtain information on the degree of dirtiness of the photovoltaic panel, wherein the infrared image is an image collected simultaneously with the visible light image;

[0033] Optionally, the execution order of step S202 and step S204 is not limited in this embodiment of the application.

[0034] Step S206: determining a cleaning strategy for the photovoltaic panel based on the dirt area information and the dirt degree information.

[0035] Through the above steps, the collected visible light image of the photovoltaic panel is detected by the pre-trained first model to obtain the dirt area information on the photovoltaic panel; the infrared image of the photovoltaic panel is analyzed to obtain the dirtiness information of the photovoltaic panel, wherein the infrared image is an image collected simultaneously with the visible light image; and the cleaning strategy for the photovoltaic panel is determined by the dirtiness area information and the dirtiness information. In other words, by identifying the dirtiness area information in the visible light image of the photovoltaic panel through the pre-trained first model and combining it with the dirtiness information identified from the infrared image of the photovoltaic panel, the cleaning strategy for the photovoltaic panel can be directly determined. Therefore, through the above embodiments, the problem of poor cleaning effect caused by making cleaning decisions for photovoltaic panels through manual regular inspections in the prior art is solved, and the cleaning effect of the photovoltaic panels is improved.

[0036] In an exemplary embodiment, before detecting the collected visible light images of the photovoltaic panel through the pre-trained first model, the method also includes: acquiring historical visible light images of the photovoltaic panel; performing a labeling operation on the dirty area in the historical visible light image, wherein the labeling operation at least includes: drawing a bounding box of the dirty area and assigning a classification label to the dirty area; and training a second model through the historical visible light images after the labeling operation to obtain the first model, wherein the second model is an untrained neural network model for dirt detection.

[0037] By annotating historical visible light images and then using standardized historical visible light images for model training, the accuracy of the model can be continuously optimized to make it more suitable for photovoltaic panel contamination detection in specific environments, thereby providing more reliable data support in practical applications. For example, in photovoltaic power stations in deserts or industrial areas, the type and degree of contamination of photovoltaic panels caused by environmental factors may be different from those in general areas. The first model used to detect photovoltaic power stations in deserts or industrial areas can be trained using historical visible light images of photovoltaic panels in photovoltaic power stations in deserts or industrial areas. This enables the model to more accurately identify contamination conditions under these special conditions.

[0038] The specific training process includes:

[0039] 1) Obtain historical visible light images of photovoltaic panels: First, the system or operator needs to collect visible light images of photovoltaic panels at different time points. These images should include views of photovoltaic panels in various cleaning states to ensure that the model can learn different types of stains and dust features.

[0040] 2) Perform labeling operations on dirty areas in historical visible light images: On the collected images, dirty areas need to be labeled. Labeling operations include but are not limited to:

[0041] Draw bounding boxes for dirty areas: Accurately mark the extent of stains and dust accumulation on the image, which is usually done through visual inspection and manual drawing of bounding boxes.

[0042] Assign classification labels to dirty areas: Assign a classification label to each area within the bounding box based on the type of stain (such as dust, sand, leaves, bird droppings, etc.) and severity. This helps the model learn the characteristics of different types of stains so that they can be distinguished during detection.

[0043] 3) Training the second model by performing the historical visible light images after the annotation operation: The training process involves adjusting the model parameters to minimize the deviation between the predicted bounding box and the actual bounding box, as well as to ensure the accuracy of the classification label.

[0044] 4) Obtaining the first model: After training, the second model is transformed into a first model that can efficiently and accurately detect stains and dust on photovoltaic panels by learning the features and patterns in the historical visible light images. Optionally, the first model is based on the YOLOv10 algorithm, which means that it can simultaneously detect objects in a single pass through the image without additional post-processing steps such as non-maximum suppression (NMS), thereby improving detection speed and accuracy.

[0045] The training step of the second model in the above embodiment ensures that the model can learn the characteristics of photovoltaic panel stains and dust from a large number of historical visible light images, so that in practical applications, it can quickly and accurately identify the stains and dust on the photovoltaic panels, providing an important basis for the cleaning and maintenance of photovoltaic panels. By combining the analysis of visible light images and infrared images, this method further improves the comprehensiveness and reliability of detection, making the cleaning strategy of photovoltaic panels more scientific and efficient.

[0046] In an exemplary embodiment, a collected visible light image of a photovoltaic panel is detected by a pre-trained first model to obtain dirty area information on the photovoltaic panel, including: determining a first dirty area in the visible light image by the first model; determining a region edge corresponding to the first dirty area by an edge segmentation algorithm; and determining the dirty area information by an image scale factor and pixel coordinates corresponding to the region edge corresponding to the first dirty area, wherein the image scale factor is an image scale factor of the visible light image obtained by the actual physical size corresponding to each pixel in the visible light image.

[0047] Optionally, in practical applications, a high-resolution camera can be used to capture images of photovoltaic panels to obtain visible light images of photovoltaic panels. Ensure that the images can clearly reflect the details of the photovoltaic panel surface. After image acquisition, necessary preprocessing is performed, such as grayscale, filtering, contrast enhancement, etc., to improve the accuracy of subsequent model detection.

[0048] The preprocessed visible light image of the photovoltaic panel is input into the pre-trained YOLOv10 model for detection. The model uses deep learning technology to identify stains and dust on the photovoltaic panel and outputs multiple bounding boxes (used to frame the first dirty area in the above embodiment), each of which corresponds to a potential dirty area. The bounding box contains the coordinate information and classification information of the dirty area, where the classification information includes but is not limited to stains or dust.

[0049] For each dirty region detected by the YOLOv10 model, an edge segmentation algorithm is further applied to accurately locate its boundaries (i.e., region edges). The edge segmentation algorithm can identify the outlines of objects in the image, thereby more accurately extracting the edge information of the dirty region. This step improves the accuracy and robustness of dirty region identification and avoids detection errors caused by blurred boundaries or changes in lighting conditions.

[0050] In order to convert the pixel coordinates on the image to the actual physical size, the image scale factor needs to be determined in advance. This is usually achieved by including a reference object of known size in the image (such as a fixed-size component of a photovoltaic panel). The size of the reference object in reality (Wreal) and its size in the image (Wpixel) are measured, and the image scale factor is calculated using the principle of similar triangles.

[0051] In an exemplary embodiment, the infrared image of the photovoltaic panel is analyzed to obtain the dirtiness information of the photovoltaic panel, including: extracting the area edge corresponding to the second dirty area of ​​the infrared image through an edge segmentation algorithm; determining the second dirty area and non-dirty area in the infrared image through the area edge corresponding to the second dirty area; determining the first temperature information of the second dirty area, and determining the second temperature information of the non-dirty area; determining the dirtiness information through the first temperature information and the second temperature information.

[0052] Optionally, the first temperature information may be an average temperature value or a median temperature value of the second dirty area; and the second temperature information may be an average temperature value or a median temperature value of the non-dirty area.

[0053] Furthermore, the dirtiness information is determined by the first temperature information and the second temperature information, including: determining the difference between the first temperature information and the second temperature information as the dirtiness information; or assigning a first weight to the first temperature information through the first area information of the second dirty area, and assigning a second weight to the second temperature information through the second area information of the non-dirty area; determining the difference between the first product and the second product as the dirtiness information, wherein the first product is the product of the first weight and the first temperature information, and the second product is the product of the second weight and the second temperature information.

[0054] Furthermore, a cleaning strategy for the photovoltaic panel is determined by the dirt area information and the dirt degree information, including one of the following: when the area value corresponding to the dirt area information is greater than a first threshold, determining that the cleaning strategy is that the photovoltaic panel needs to be cleaned; when the degree value corresponding to the dirt degree information is greater than a second threshold, determining that the cleaning strategy is that the photovoltaic panel needs to be cleaned; when the area value corresponding to the dirt area information is greater than the first threshold and the degree value corresponding to the dirt degree information is greater than the second threshold, determining that the cleaning strategy is that the photovoltaic panel must be cleaned.

[0055] This dual-threshold-based cleaning strategy formulation method can more accurately determine the cleaning needs of photovoltaic panels, avoid wasting resources, and ensure that the performance of photovoltaic panels is not affected by dirt. In practical applications, this method can be combined with real-time operating data of photovoltaic power stations, such as power generation and ambient temperature, to dynamically adjust the threshold and achieve a more flexible and efficient cleaning plan. For example, during peak power generation, the threshold can be appropriately lowered to ensure that all photovoltaic panels are in the best working condition, while during off-peak periods, the threshold can be appropriately raised to reduce the cleaning frequency and save costs.

[0056] In order to better understand the process of the above-mentioned method for determining the cleaning strategy, the above-mentioned method for determining the cleaning strategy is described below in combination with an optional embodiment, but it is not used to limit the technical solution of the embodiment of the present application.

[0057] The optional embodiment of the present application proposes a method for detecting stains and dust accumulation on photovoltaic panels based on the YOLOv10 algorithm and infrared image analysis (equivalent to the determination of the cleaning strategy in the above embodiment). The YOLOv10 algorithm is mainly used to detect stains and dust accumulation on photovoltaic panels, and combined with infrared image analysis technology, the area of ​​stains and dust accumulation is calculated through edge segmentation algorithm and conversion of pixel coordinates to physical coordinates (equivalent to the dirt area information in the above embodiment). At the same time, by analyzing the temperature difference of the stain area on the infrared image, the severity of the dirt is evaluated (equivalent to the dirt degree information in the above embodiment). Based on the area of ​​stains and dust accumulation and their severity, auxiliary decision-making is provided for the cleaning of photovoltaic panels.

[0058] Optionally, a photovoltaic panel stain and dust detection method based on YOLOv10 algorithm and infrared image analysis is provided in an optional embodiment of the present application, such as Figure 3 As shown, it mainly includes: YOLOv10 detection algorithm training, stain and dust area calculation, infrared image temperature difference analysis, and cleaning decision.

[0059] Among them, YOLOv10 detection algorithm training includes:

[0060] Step 301: Collect a large number of photovoltaic panel images (equivalent to the historical visible light images in the above embodiment), such as Figure 4 As shown, these images should cover different situations of soiling and dust accumulation, including PV panels under different degrees and types and lighting conditions.

[0061] Step 302: accurately annotate the stains and dust in the image, including drawing bounding boxes and classification labels, to ensure the accuracy and consistency of the annotations.

[0062] Step 303: Use the labeled data set to train the YOLOv10 model (equivalent to the second model in the above embodiment) so that it can identify stains and dust on the photovoltaic panel. Environment configuration: Ensure that the computing environment has the necessary libraries and frameworks installed, such as Python, PyTorch, CUDA, etc. Model selection: Select the appropriate YOLOv10 model size according to project requirements, such as YOLOv10-S is suitable for scenarios with high speed requirements, while YOLOv10-L is suitable for scenarios with higher accuracy requirements.

[0063] The calculation of stain and dust accumulation area includes:

[0064] Step 401: It is necessary to know the actual physical size represented by each pixel in the image (equivalent to the visible light image in the above embodiment). This is usually obtained through a calibration process during image acquisition. If an object of known size in the image is known, the image scale factor can be determined using the principle of similar triangles.

[0065] Step 402: Convert the pixel coordinates of the stains and dust in the image into physical coordinates. This involves multiplying the pixel coordinates of the bounding box (i.e., the edge of the area corresponding to the first dirty area) by the image scale factor to obtain the area (i.e., the dirty area information). If the edge is not a regular rectangle, a polygon approximation algorithm can be used to approximate the edge to a polygon. Use a polygon area formula (e.g., the Shoelace formula) to calculate the area of ​​the approximated polygon (i.e., the dirty area information).

[0066] Specifically, the area calculation principle includes:

[0067] Determine the actual size and pixel size of a known object: Select an object of known size in the image (e.g. a standard cell of a photovoltaic panel). Measure the actual width of the object. l and the width Wpixel in the image.

[0068] Calculate the conversion ratio from pixel size to physical size (actual size) (equivalent to the image scale factor in the above embodiment):

[0069]

[0070] Calculate the pixel area of ​​stains and dust accumulation: Use edge segmentation algorithm to extract the boundary of stains and dust accumulation (that is, the edge of the area corresponding to the first dirty area). Calculate the pixel area Apixel within the boundary.

[0071] Convert to actual physical area:

[0072] Areal=Apixel×(Pixel_to_Real_Ratio)

[0073] Step 403: Apply the trained YOLOv10 model to the actual photovoltaic panel image to detect the location and size of the stains and dust, including the location (bounding box coordinates), size and category of the stains and dust.

[0074] Among them, infrared image temperature difference analysis includes:

[0075] The severity of the dirt is evaluated by calculating the temperature difference (ie, the difference between the first temperature information and the second temperature information) between the dirty area (equivalent to the second dirty area) and the clean area (equivalent to the non-dirty area) on the infrared image.

[0076] Step 501: pre-process the infrared image by edge segmentation to extract the edges of stains and dust accumulation (ie, the edges of the region corresponding to the second dirty region).

[0077] Step 502: Identify the dirty area and the clean area on the infrared image, calculate the average or median temperature of these areas, and calculate the temperature difference between the dirty area and the clean area.

[0078] Among them, cleaning decisions include:

[0079] Step 601: Setting the area threshold (i.e., the first threshold): Setting the area threshold of the stain according to the operation requirements and historical data of the photovoltaic panel. When the stain area exceeds the threshold, the system will prompt that cleaning may be required.

[0080] Step 602: Setting the temperature difference threshold (i.e., the second threshold): Setting the temperature difference threshold between the dirty area and the clean area, which can be determined based on the threshold of the photovoltaic panel performance degradation. When the temperature difference exceeds the threshold, it indicates that the dirt may have a significant impact on the photovoltaic panel performance.

[0081] Step 603: Comprehensive evaluation: Combine the analysis results of the stain area and temperature difference to conduct a comprehensive evaluation. If both the area and the temperature difference exceed their respective thresholds, cleaning is strongly recommended. If only one exceeds the threshold, a comprehensive judgment on whether cleaning is needed can be made based on the actual situation and historical data.

[0082] Step 604: Output of cleaning recommendations: Based on the comprehensive evaluation results, a cleaning recommendation report is output, which should include the location, area, temperature difference of the stain, and detailed description of the cleaning recommendations.

[0083] In summary, if Figure 5 As shown, the optional embodiment of the present application finally outputs the cleaning decision for the photovoltaic panel through YOLOv10 detection algorithm training, stain and dust area calculation, infrared image temperature difference analysis, and cleaning decision assistance. Among them, the YOLOv10 detection algorithm training mainly detects the stains and dust on the photovoltaic panel; the area of ​​the stains and dust is calculated by the edge segmentation algorithm; the temperature difference of the stain area on the infrared image is analyzed by infrared image temperature difference analysis to evaluate the severity of the dirt; the cleaning decision is mainly based on the area and severity of the stains and dust, and provides auxiliary decision-making for the cleaning of the photovoltaic panel.

[0084] The optional embodiment of the present application realizes efficient target detection capability: YOLOv10 algorithm is the latest target detection model, known for its high efficiency and accuracy. In the detection of stains and dust accumulation on photovoltaic panels, YOLOv10 can quickly identify stains and dust accumulation areas on the surface of photovoltaic panels. Through infrared image analysis, this technology can capture the temperature difference caused by stains and dust accumulation, and then accurately calibrate the location and range of the contaminated area.

[0085] The optional embodiment of the present application implements dual label assignment and NMS-free training: YOLOv10 introduces dual label assignment and NMS-free (non-maximum suppression) training methods. This innovation significantly improves detection performance and efficiency. During the training phase, YOLOv10 combines one-to-many and one-to-one label assignment strategies, which not only retains rich supervisory signals, but also avoids the reasoning delay caused by relying on NMS post-processing. This training method makes the model more efficient during the optimization process, and can directly output high-quality detection results during the reasoning phase without the need for additional post-processing steps.

[0086] The optional embodiment of the present application realizes an optimized model architecture design: YOLOv10 is also fully optimized in terms of model architecture design, including lightweight classification head, spatial channel decoupling down sampling and level guidance block design. These design strategies not only reduce the computational overhead, but also improve the detection accuracy and efficiency of the model. For the photovoltaic panel stain and dust detection tasks, these optimizations enable the model to better adapt to the complex environment and diverse needs in actual application scenarios while maintaining high performance.

[0087] The optional embodiments of the present application realize the unique advantages of infrared image analysis: infrared image analysis technology uses changes in the surface temperature of photovoltaic panels to detect stains and dust accumulation. Since stains and dust accumulation affect the heat absorption and heat dissipation performance of photovoltaic panels, they will show obvious temperature differences in infrared images. This non-contact detection method is not only efficient and fast, but also can avoid damage or safety hazards caused by direct contact with photovoltaic panels. In addition, infrared image analysis can also provide comprehensive temperature distribution information, which helps to more accurately evaluate the overall performance status of photovoltaic panels.

[0088] The optional embodiments of the present application have high real-time and automation: The optional embodiments of the present application can have a high degree of real-time and automation based on the photovoltaic panel stain and dust detection technology of the YOLOv10 algorithm and infrared image analysis. This technology can automatically collect infrared image data of photovoltaic panels, and obtain detection results through rapid processing and analysis by algorithms. This automated detection method not only reduces the need for manual intervention, but also improves the accuracy and efficiency of detection. For large-scale photovoltaic power stations, this real-time and efficient detection technology has important application value.

[0089] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods of each embodiment of the present application.

[0090] The present application also provides a structural block diagram of a cleaning strategy determination device. Figure 6 is a structural block diagram of a device for determining a cleaning strategy according to an embodiment of the present application; Figure 6 As shown, including:

[0091] A detection module 62 is used to detect the collected visible light image of the photovoltaic panel using a pre-trained first model to obtain the dirt area information on the photovoltaic panel;

[0092] An analysis module 64 is used to analyze the infrared image of the photovoltaic panel to obtain information on the degree of dirtiness of the photovoltaic panel, wherein the infrared image is an image collected simultaneously with the visible light image;

[0093] The determination module 66 is used to determine a cleaning strategy for the photovoltaic panel according to the dirt area information and the dirt degree information.

[0094] Through the above-mentioned device, the collected visible light image of the photovoltaic panel is detected by the pre-trained first model to obtain the dirt area information on the photovoltaic panel; the infrared image of the photovoltaic panel is analyzed to obtain the dirt degree information of the photovoltaic panel, wherein the infrared image is an image collected simultaneously with the visible light image; the cleaning strategy for the photovoltaic panel is determined by the dirt area information and the dirt degree information. In other words, by identifying the dirt area information in the visible light image of the photovoltaic panel through the pre-trained first model and combining it with the dirt degree information identified from the infrared image of the photovoltaic panel, the cleaning strategy for the photovoltaic panel can be directly determined. Therefore, through the above-mentioned embodiment, the problem of poor cleaning effect caused by making cleaning decisions for photovoltaic panels through manual regular inspections in the prior art is solved, and the cleaning effect of the photovoltaic panel is improved.

[0095] In an exemplary embodiment, the device also includes a training module, which is used to: obtain historical visible light images of the photovoltaic panel before detecting the collected visible light images of the photovoltaic panel through a pre-trained first model; perform a labeling operation on the dirty area in the historical visible light image, wherein the labeling operation at least includes: drawing a bounding box of the dirty area and assigning a classification label to the dirty area; and train a second model by performing the labeling operation on the historical visible light images to obtain the first model, wherein the second model is an untrained neural network model for dirt detection.

[0096] In an exemplary embodiment, the detection module 62 is further used to: determine a first dirty area in the visible light image through the first model; determine a region edge corresponding to the first dirty area through an edge segmentation algorithm; and determine the dirty area information through an image scale factor and pixel coordinates corresponding to the region edge corresponding to the first dirty area, wherein the image scale factor is an image scale factor of the visible light image obtained through the actual physical size corresponding to each pixel in the visible light image.

[0097] In an exemplary embodiment, the analysis module 64 is also used to: extract the area edge corresponding to the second dirty area of ​​the infrared image through an edge segmentation algorithm; determine the second dirty area and the non-dirty area in the infrared image through the area edge corresponding to the second dirty area; determine the first temperature information of the second dirty area, and determine the second temperature information of the non-dirty area; determine the dirtiness information through the first temperature information and the second temperature information.

[0098] In an exemplary embodiment, the analysis module 64 is also used to: determine the difference between the first temperature information and the second temperature information as the dirtiness information; or assign a first weight to the first temperature information through the first area information of the second dirty area, and assign a second weight to the second temperature information through the second area information of the non-dirty area; determine the difference between the first product and the second product as the dirtiness information, wherein the first product is the product of the first weight and the first temperature information, and the second product is the product of the second weight and the second temperature information.

[0099] In an exemplary embodiment, the determination module 66 is also used for one of the following: when the area value corresponding to the dirty area information is greater than a first threshold, determining that the cleaning strategy is that the photovoltaic panel needs to be cleaned; when the degree value corresponding to the dirty level information is greater than a second threshold, determining that the cleaning strategy is that the photovoltaic panel needs to be cleaned; when the area value corresponding to the dirty area information is greater than the first threshold and the degree value corresponding to the dirty level information is greater than the second threshold, determining that the cleaning strategy is that the photovoltaic panel must be cleaned.

[0100] An embodiment of the present application further provides a storage medium, which includes a stored program, wherein the program executes any of the above methods when it is run.

[0101] Optionally, in this embodiment, the storage medium may be configured to store program codes for executing the following steps:

[0102] S1, detecting the collected visible light image of the photovoltaic panel by using a pre-trained first model to obtain the dirt area information on the photovoltaic panel;

[0103] S2, analyzing the infrared image of the photovoltaic panel to obtain information on the degree of dirtiness of the photovoltaic panel, wherein the infrared image is an image collected simultaneously with the visible light image;

[0104] S3, determining a cleaning strategy for the photovoltaic panel according to the dirt area information and the dirt degree information.

[0105] An embodiment of the present application further provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0106] Optionally, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0107] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0108] S1, detecting the collected visible light image of the photovoltaic panel by using a pre-trained first model to obtain the dirt area information on the photovoltaic panel;

[0109] S2, analyzing the infrared image of the photovoltaic panel to obtain information on the degree of dirtiness of the photovoltaic panel, wherein the infrared image is an image collected simultaneously with the visible light image;

[0110] S3, determining a cleaning strategy for the photovoltaic panel according to the dirt area information and the dirt degree information.

[0111] Optionally, in this embodiment, the above-mentioned storage medium may include but is not limited to: a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk, and other media that can store program codes.

[0112] An embodiment of the present application further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the steps in any one of the above method embodiments are implemented.

[0113] An embodiment of the present application further provides another computer program product, including a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in any of the above method embodiments are implemented.

[0114] An embodiment of the present application also provides a computer program, which includes computer instructions stored in a computer-readable storage medium; a processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device performs the steps in any one of the above method embodiments.

[0115] Optionally, the specific examples in this embodiment may refer to the examples described in the above embodiments and optional implementation modes, and this embodiment will not be described in detail here.

[0116] Obviously, those skilled in the art should understand that the above modules or steps of the present application can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, and optionally, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order from that herein, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present application is not limited to any specific combination of hardware and software.

[0117] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for determining a cleaning strategy, characterized in that: include: The collected visible light image of the photovoltaic panel is detected by using a pre-trained first model to obtain the dirt area information on the photovoltaic panel; Analyzing the infrared image of the photovoltaic panel to obtain information on the degree of dirtiness of the photovoltaic panel, wherein the infrared image is an image collected simultaneously with the visible light image; A cleaning strategy for the photovoltaic panel is determined based on the dirt area information and the dirt degree information.

2. The method for determining a cleaning strategy according to claim 1, characterized in that: Before detecting the collected visible light image of the photovoltaic panel by using the pre-trained first model, the method further includes: Acquire historical visible light images of the photovoltaic panel; Performing a labeling operation on the dirty area in the historical visible light image, wherein the labeling operation at least includes: drawing a boundary box of the dirty area and assigning a classification label to the dirty area; The first model is obtained by training a second model using historical visible light images after the annotation operation, wherein the second model is an untrained neural network model for dirt detection.

3. The method for determining a cleaning strategy according to claim 1, characterized in that: The collected visible light image of the photovoltaic panel is detected by using the pre-trained first model to obtain the dirt area information on the photovoltaic panel, including: Determine a first dirty area in the visible light image by using the first model; Determine the region edge corresponding to the first dirty region by using an edge segmentation algorithm; The dirty area information is determined by an image scale factor and pixel coordinates corresponding to an area edge corresponding to the first dirty area, wherein the image scale factor is an image scale factor of the visible light image obtained by an actual physical size corresponding to each pixel in the visible light image.

4. The method for determining a cleaning strategy according to claim 1, characterized in that: Analyzing the infrared image of the photovoltaic panel to obtain information on the degree of dirtiness of the photovoltaic panel includes: Extracting the area edge corresponding to the second dirty area of ​​the infrared image by an edge segmentation algorithm; Determine the second dirty area and non-dirty area in the infrared image through the area edge corresponding to the second dirty area; Determining first temperature information of the second dirty area, and determining second temperature information of the non-dirty area; The dirtiness information is determined by the first temperature information and the second temperature information.

5. The method for determining a cleaning strategy according to claim 4, characterized in that: Determining the dirtiness information by using the first temperature information and the second temperature information includes: determining a difference between the first temperature information and the second temperature information as the dirtiness information; or A first weight is assigned to the first temperature information through the first area information of the second dirty area, and a second weight is assigned to the second temperature information through the second area information of the non-dirty area; and the difference between the first product and the second product is determined as the dirtiness information, wherein the first product is the product of the first weight and the first temperature information, and the second product is the product of the second weight and the second temperature information.

6. The method for determining a cleaning strategy according to claim 1, characterized in that: Determining a cleaning strategy for the photovoltaic panel according to the dirt area information and the dirt degree information includes one of the following: When the area value corresponding to the dirty area information is greater than a first threshold, determining that the cleaning strategy is that the photovoltaic panel needs to be cleaned; When the degree value corresponding to the dirtiness information is greater than a second threshold, determining that the cleaning strategy is that the photovoltaic panel needs to be cleaned; When the area value corresponding to the dirt area information is greater than a first threshold value, and the degree value corresponding to the dirt degree information is greater than a second threshold value, it is determined that the cleaning strategy is that the photovoltaic panel must be cleaned.

7. A device for determining a cleaning strategy, characterized in that: include: A detection module, used to detect the collected visible light image of the photovoltaic panel through a pre-trained first model to obtain the dirt area information on the photovoltaic panel; An analysis module, used for analyzing the infrared image of the photovoltaic panel to obtain information on the degree of dirtiness of the photovoltaic panel, wherein the infrared image is an image collected simultaneously with the visible light image; A determination module is used to determine a cleaning strategy for the photovoltaic panel based on the dirt area information and the dirt degree information.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein the program executes the method described in any one of claims 1 to 6 when executed.

9. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to execute the method according to any one of claims 1 to 6 through the computer program.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images