Photovoltaic panel category identification method and system, and storage medium

Abstract

The invention discloses a photovoltaic panel category identification method and system and a storage medium. The method comprises the following steps: acquiring remote sensing data of one or more photovoltaic panels in a target area; extracting spectral feature information of one or more photovoltaic panels from the remote sensing data; and inputting the spectral feature information into a trained photovoltaic panel category identification model to obtain a photovoltaic panel category identification result output by the photovoltaic panel category identification model. According to the method and the system disclosed by the embodiment of the invention, the obtained remote sensing data of the photovoltaic panels are input into the photovoltaic panel category recognition model obtained by deep learning according to the characteristic differences of different categories of photovoltaic panels in the remote sensing data, so that the category of the photovoltaic panels is quickly recognized, and the recognition efficiency of the category of the photovoltaic panels in a large-area range is improved; and technical support is provided for application scenes such as monitoring of the position of the photovoltaic panel or control of the installation progress of the photovoltaic panel.

Description

Technical Field

[0001] The present invention relates to the field of photovoltaic technology, and in particular to a photovoltaic panel category identification method, system and storage medium based on remote sensing data. Background Art

[0002] Photovoltaic panels, also known as solar photovoltaic panels or solar panels, are classified mainly based on factors such as the semiconductor materials used, manufacturing processes, and application scenarios.

[0003] Photovoltaic panels are the core components of solar photovoltaic systems, and their quality and performance directly affect the safety and stability of the entire solar photovoltaic system. By identifying the material, model, size, location, etc. of photovoltaic panels, it can be determined whether their quality or performance meets the requirements. In the prior art, the category of photovoltaic panels is generally identified manually by appearance or model, but manual identification of the category of photovoltaic panels requires a lot of human resources, especially when faced with the identification of photovoltaic panel categories in a large area, so the identification efficiency cannot meet the requirements.

[0004] Therefore, the prior art needs to be further improved. Summary of the invention

[0005] In view of the above-mentioned deficiencies in the prior art, an object of the present invention is to provide users with a method, system and storage medium for photovoltaic panel category identification based on remote sensing data, so as to overcome the defect of low efficiency of photovoltaic panel category identification in the prior art.

[0006] The technical solution adopted by the present invention to solve the technical problem is as follows: In a first aspect, this embodiment discloses a method for identifying photovoltaic panel categories based on remote sensing data, which includes: Obtain remote sensing data of one or more photovoltaic panels in a target area; Extracting spectral characteristic information of one or more photovoltaic panels from the remote sensing data; The spectral feature information is input into a trained photovoltaic panel category recognition model to obtain a photovoltaic panel category recognition result output by the photovoltaic panel category recognition model.

[0007] Optionally, the step of acquiring remote sensing data of one or more photovoltaic panels in the target area includes: Obtain remote sensing data of the target area; Identify the location area information of the photovoltaic panels from the remote sensing data of the target area; The remote sensing data of the photovoltaic panel is filtered out according to the location area information of the photovoltaic panel.

[0008] Optionally, the step of extracting spectral characteristic information of one or more photovoltaic panels from the remote sensing data comprises: The remote sensing data is input into a preset spectral characteristic information analysis tool to obtain spectral characteristic information of one or more photovoltaic panels output by the preset spectral characteristic information analysis tool.

[0009] Optionally, the method for generating the photovoltaic panel category recognition model includes: Obtain spectral characteristic information of different types of photovoltaic panels; Establish a training data set based on the categories and spectral feature information corresponding to the photovoltaic panels; The preset network model is trained using the training data set to obtain a trained photovoltaic panel category recognition model.

[0010] Optionally, the preset network model is a convolutional neural network model, a residual network model, a classification model based on machine learning, or a recognition model based on physical features.

[0011] Optionally, the step of acquiring spectral characteristic information of photovoltaic panels of different categories includes: Collect samples of different types of photovoltaic materials; the photovoltaic materials include: one or more of monocrystalline silicon, polycrystalline silicon, cadmium telluride, and copper indium gallium selenide; Hyperspectral imaging is performed on each category of photovoltaic material samples to obtain spectral characteristic information of each photovoltaic material sample at different wavelengths.

[0012] Optionally, the step of establishing a training data set according to the category and spectral feature information corresponding to the photovoltaic panel includes: Constructing a first data set according to the categories corresponding to the photovoltaic panels and the spectral characteristic information of different photovoltaic panels at different wavelengths; Acquire spectral remote sensing information of the location of the photovoltaic panels in the specified area, and extract spectral characteristic information of the photovoltaic panels in the specified area based on the spectral remote sensing information; Comparing the spectral characteristic information of the photovoltaic panels in the specified area with the spectral characteristic information of the photovoltaic panels contained in the first data set to identify the type of the photovoltaic panels in the specified area; Labeling the photovoltaic panels in the designated area according to the identified categories and spectral feature information of the photovoltaic panels, and adding the labeled photovoltaic panels to the first data set to obtain a second data set; A training data set is constructed according to the photovoltaic panel information in the second data set.

[0013] Optionally, the spectral feature information includes multiple; The step of inputting the spectral feature information into a trained photovoltaic panel category recognition model to obtain a photovoltaic panel category recognition result output by the photovoltaic panel category recognition model comprises: Each spectral feature information is input into the trained photovoltaic panel category recognition model in turn to obtain the identified photovoltaic panel category corresponding to each spectral feature information, and multiple photovoltaic panel categories constitute a photovoltaic panel category recognition set.

[0014] In a second aspect, this embodiment provides a photovoltaic panel category identification system based on remote sensing data, which includes: A data acquisition module, used to acquire remote sensing data of one or more photovoltaic panels in a target area; A feature extraction module, used to extract spectral feature information of one or more photovoltaic panels from the remote sensing data; The category recognition module is used to determine the category data corresponding to one or more photovoltaic panels in the target area according to the spectral feature information.

[0015] In a third aspect, the present embodiment provides a computer storage medium, wherein a photovoltaic panel category identification program based on remote sensing data is stored on the computer-readable storage medium, and when the photovoltaic panel category identification program based on remote sensing data is executed by a processor, the steps of the photovoltaic panel category identification method based on remote sensing data are implemented.

[0016] Beneficial effects: The present embodiment discloses a method, system and storage medium for identifying the category of photovoltaic panels based on remote sensing data, by acquiring remote sensing data of one or more photovoltaic panels in a target area; extracting spectral feature information of one or more photovoltaic panels from the remote sensing data; inputting the spectral feature information into a trained photovoltaic panel category recognition model to obtain a photovoltaic panel category recognition result output by the photovoltaic panel category recognition model. The method and system disclosed in the present embodiment input the acquired remote sensing data of photovoltaic panels into a photovoltaic panel category recognition model obtained by deep learning to achieve rapid identification of photovoltaic panel categories based on the feature differences presented by different categories of photovoltaic panels in remote sensing data, thereby improving the efficiency of identifying photovoltaic panel categories in a large area and providing technical support for application scenarios such as monitoring the location of photovoltaic panels or controlling the progress of photovoltaic panel installation. The method and system of the present embodiment are particularly suitable for the identification of large-area photovoltaic panel categories and have high applicability. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A flowchart of the steps of a method for identifying photovoltaic panel categories based on remote sensing data in an embodiment of the present invention; Figure 2 A flowchart of a specific application example of the method of the embodiment of the present invention; Figure 3 It is a principle structure block diagram of the system in an embodiment of the present invention; Figure 4 This is a spectral characteristic diagram of a photovoltaic panel based on hyperspectral remote sensing data in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0019] Since photovoltaic panels are important components in solar photovoltaic systems, it is of great significance to accurately identify the types of photovoltaic panels to ensure the safety and stability of solar photovoltaic systems. For example, different types of photovoltaic panels can be applied to different scenarios based on the identified types of photovoltaic panels, so that different types of photovoltaic panels can meet different application requirements. Since different types of photovoltaic panels have different photoelectric conversion efficiencies, users can also choose the appropriate photovoltaic panels based on the identified types of photovoltaic panels to maximize energy output. In the existing technology, manual identification is usually used to identify the category of photovoltaic panels. However, the manual identification method is not only inefficient, but also greatly affected by subjective factors, resulting in the inability to guarantee the recognition accuracy. In addition, with the advancement of sustainable development plans, photovoltaic products have shown a rapid growth trend in recent years, and photovoltaic panels are widely used in photovoltaic power generation systems. A photovoltaic panel category recognition method suitable for a large area is needed to improve the efficiency and accuracy of recognition, which is crucial to optimizing the layout of the photovoltaic industry.

[0020] In order to improve the accuracy and efficiency of photovoltaic panel category recognition, the present application provides a photovoltaic panel category recognition method, system and storage medium based on remote sensing data, by acquiring remote sensing data of all photovoltaic panels within the target area, and extracting spectral feature information of each photovoltaic panel based on remote sensing data; using the trained photovoltaic panel category recognition module to identify the spectral feature information to obtain the identified photovoltaic panel category. The method and system disclosed in this embodiment combine remote sensing technology and deep learning to achieve efficient and accurate recognition of photovoltaic panel categories, and based on remote sensing technology, it has a better advantage in acquiring spatial information. Therefore, the method and system of this embodiment are suitable for the identification of photovoltaic panel categories in a large area, and can provide technical support for the layout and detection of photovoltaic panels.

[0021] The method, system and storage medium of this embodiment are further described in detail below in conjunction with the drawings in the specification.

[0022] This embodiment discloses a photovoltaic panel category identification method based on remote sensing data, such as Figure 1 As shown, the steps of the photovoltaic panel category identification method include: Step S1, obtaining remote sensing data of one or more photovoltaic panels in a target area.

[0023] In order to identify the category of the photovoltaic panels in the target area, in this step, the remote sensing data of one or more photovoltaic panels in the target area is first obtained. Specifically, the target area can be a power plant with a variety of different photovoltaic panels, and the target area can also be a partial area of ​​a photovoltaic power plant. Since there are many different photovoltaic panels arranged in a photovoltaic power plant, the category identification of the photovoltaic panels in the area can be realized according to the different categories of photovoltaic panels. Arrange them in different positions, and by identifying the category of photovoltaic panels, monitor whether the installation position of the photovoltaic panels is correct, thereby improving the power generation performance of the entire photovoltaic power plant.

[0024] The remote sensing data may be data obtained from a professional remote sensing data platform, or may be drone images, aerial images or satellite images. The drone images, aerial images or satellite images are subjected to noise reduction, atmospheric correction and geometric correction, and then image enhancement, image landmark information enhancement, and image cropping to obtain the photovoltaic panel image.

[0025] Specifically, the specific steps of acquiring remote sensing data of one or more photovoltaic panels in the target area in this step include: Step S11, acquiring remote sensing data of the target area.

[0026] In this step, the remote sensing data of the target area can be obtained from a professional remote sensing data provider platform, or the remote sensing data of the target area can be searched from open source data resources, or the remote sensing data of the target area can be obtained by using a drone.

[0027] Step S12: Identify the location area information of the photovoltaic panels from the remote sensing data of the target area.

[0028] After the remote sensing data of the target area is acquired, the location area information of the photovoltaic panels in the target area is identified from the remote sensing data of the target area. Specifically, if the remote sensing data of the target area is image information, the shape features of the photovoltaic panels of the objects in the image can be identified to locate the location area information of the photovoltaic panels. If the remote sensing data contains surface reflectivity information, the surface reflectivity data can be used to detect the absorption signal of the packaging film of the photovoltaic panel, so as to quickly and accurately obtain the distribution boundary information of the photovoltaic panel. It can be imagined that there can be multiple photovoltaic panels in the target area, and the location area information of the photovoltaic panels identified from the remote sensing data of the target area can correspond to the location information of multiple photovoltaic panels.

[0029] Step S13: Filter out the remote sensing data of the photovoltaic panel according to the location area information of the photovoltaic panel.

[0030] According to the location area information of the located photovoltaic panels, the remote sensing data of each photovoltaic panel can be quickly filtered out from the remote sensing data of the target area.

[0031] Step S2: extracting spectral characteristic information of one or more photovoltaic panels from the remote sensing data.

[0032] After obtaining the remote sensing data of the photovoltaic panels in the target area, the spectral characteristic information of each photovoltaic panel is extracted based on the remote sensing data of the photovoltaic panels. The spectral characteristic information is the electromagnetic radiation characteristics of the photovoltaic panel at different wavelengths, including reflectivity, absorptivity and spectral index. These parameters can quantify the spectral characteristics of the photovoltaic panel and provide support for the identification of the photovoltaic panel category.

[0033] The step of acquiring the spectral characteristic information of the photovoltaic panel based on the remote sensing data may be: analyzing the reflection or absorption characteristics of the photovoltaic panel in different bands respectively, so as to intuitively display the spectral characteristics of the photovoltaic panel by drawing a spectral curve graph.

[0034] Specifically, the step of extracting spectral characteristic information of one or more photovoltaic panels from the remote sensing data includes: The remote sensing data is input into a preset spectral characteristic information analysis tool to obtain spectral characteristic information of one or more photovoltaic panels output by the preset spectral characteristic information analysis tool.

[0035] In one embodiment, a spectral feature information analysis tool can be used to extract spectral feature information of photovoltaic panels from remote sensing data. The spectral feature information analysis tool is a software tool specifically used to process and analyze spectral feature information. For example: OPUS spectral software, GRAMS spectral analysis software, Omnic infrared analysis software, OceanView spectral analysis software, Unscrambler HSI spectral analysis software, etc. You can select a suitable spectral feature information analysis tool according to the specific application scenario and data characteristics. Preferably, ENVI can be used to process remote sensing data to obtain spectral feature information of photovoltaic panels. ENVI (The Environment for Visualizing Images) is a complete remote sensing image processing platform that can realize image data input / output, image enhancement, information extraction and other processing operations.

[0036] Step S3: input the spectral feature information into a trained photovoltaic panel category recognition model to obtain a photovoltaic panel category recognition result output by the photovoltaic panel category recognition model.

[0037] After the spectral feature information of each photovoltaic panel is obtained in the above steps, the obtained spectral feature information is input into the trained photovoltaic panel category recognition model to obtain the photovoltaic panel category recognition result output by the photovoltaic panel category recognition model.

[0038] Since the photovoltaic panel category recognition model used in this step is a model obtained based on deep learning training, accurate category recognition results can be obtained by analyzing and identifying the spectral feature information of the photovoltaic panels using this model.

[0039] Specifically, the method for generating the photovoltaic panel category recognition model includes: Step S31, obtaining spectral characteristic information of photovoltaic panels of different categories.

[0040] In this step, the spectral characteristic information of different types of photovoltaic panels is collected. There are many types of photovoltaic panels, including monocrystalline silicon, polycrystalline silicon, cadmium telluride, copper indium gallium selenide, etc. In order to obtain more spectral characteristic information of different types of photovoltaic panels, different types of photovoltaic material samples can also be collected; the photovoltaic materials include: one or more of monocrystalline silicon, polycrystalline silicon, cadmium telluride, copper indium gallium selenide; each type of photovoltaic material sample is subjected to hyperspectral imaging to obtain the spectral characteristic information of each photovoltaic material sample at different wavelengths. In the specific implementation, hyperspectral imaging is performed on each photovoltaic material sample to obtain the spectral characteristic information of each photovoltaic material sample at different wavelengths.

[0041] Step S32: establishing a training data set according to the category and spectral feature information corresponding to the photovoltaic panel.

[0042] In this step, a first data set is first constructed based on the collected photovoltaic material samples of different categories, and then the first training set is optimized based on the actual scenario to obtain a second data set.

[0043] Specifically, combined Figure 2 As shown, the steps of constructing training data in this step include: First, a first data set is constructed according to the categories corresponding to photovoltaic panels and the spectral characteristic information of different photovoltaic panels at different wavelengths; that is, the first data set only contains the categories and spectral characteristic information corresponding to the collected photovoltaic material samples.

[0044] Secondly, obtain the spectral remote sensing information of the location of the photovoltaic panels in the designated area, and extract the spectral characteristic information of the photovoltaic panels in the designated area based on the spectral remote sensing information; the designated area can be a target area, or a different area with a similar application scenario to the target area. The designated area is an area where multiple different types of photovoltaic panels are actually used, so the first data set is optimized by the category and spectral characteristic information of the photovoltaic panels in the actual scenario, and the preset network model is trained using the optimized data set to obtain a photovoltaic panel category recognition model that is closer to the application scenario of the target area.

[0045] Again, the spectral characteristic information of the photovoltaic panels in the specified area is compared with the spectral characteristic information of the photovoltaic panels contained in the first data set to identify the category of the photovoltaic panels in the specified area. Since the category information of the photovoltaic panels in the specified area is unknown, in this step, the spectral characteristic information of the photovoltaic panels in the specified area is compared with the spectral characteristic information of the photovoltaic panels in the first data set to identify the category of the photovoltaic panels in the specified area.

[0046] Next, the photovoltaic panels in the designated area are labeled according to the identified categories and spectral feature information of the photovoltaic panels, and the labeled photovoltaic panels are added to the first data set to obtain a second data set.

[0047] The identified categories and spectral feature information are marked on the photovoltaic panels in the specified area, and the marked photovoltaic panel categories and spectral feature information are added to the first data set, thereby obtaining an updated second data set.

[0048] Finally, a training data set is constructed according to the photovoltaic panel information in the second data set.

[0049] The training data set is constructed based on the category information and spectral feature information of the photovoltaic panels stored in the second data set. In order to achieve more accurate training results, the corresponding relationship data between the spectral feature information and the category of photovoltaic panels in the first data set can be continuously optimized, and the training data set can be constructed after the number of photovoltaic panel information labeled with category and spectral feature information is not less than 500.

[0050] Step S33: Use the training data set to train the preset network model to obtain a trained photovoltaic panel category recognition model.

[0051] The preset network model is trained using the training data set constructed in the above step S32, thereby obtaining a trained photovoltaic panel category recognition model. In the specific implementation process, the preset network model is a convolutional neural network model, a residual network model, a classification model based on machine learning, or a recognition model based on physical features.

[0052] When training the preset network model, combined with Figure 2As shown, the sorted training data set is divided into a training set, a validation set and a test set according to a certain ratio. Then the data is preprocessed and a suitable preset network model is selected. Secondly, some training parameters are set to start training, and the divided training set data is input into the selected preset network model, and training is performed according to the set training parameters. During the training process of the preset network model, the validation set data is regularly input into the model in training to observe the degree of consistency between the output results of the model on the validation set and the labeled results, so as to evaluate the performance of the trained model, and adjust the model according to the performance of the model on the validation set to improve the performance of the model. When the preset network model is trained, a trained photovoltaic panel category recognition model is obtained. The test set data is input into the trained photovoltaic panel category recognition model to finally evaluate the performance of the photovoltaic panel category recognition model. If the evaluation result of the photovoltaic panel category recognition model on the test machine is not ideal, the photovoltaic panel category recognition model needs to be optimized, and then retrained and evaluated until a satisfactory result is obtained.

[0053] Furthermore, since there may be multiple photovoltaic panels in the target area, the extracted spectral feature information includes multiple ones. In this step, the spectral feature information is input into the trained photovoltaic panel category recognition model to obtain the photovoltaic panel category recognition result output by the photovoltaic panel category recognition model. Specifically, the step includes: inputting each spectral feature information into the trained photovoltaic panel category recognition model in turn to obtain the identified photovoltaic panel category corresponding to each spectral feature information. Multiple photovoltaic panel categories constitute a photovoltaic panel category recognition set, so the final recognition result is a data set containing multiple photovoltaic panel categories.

[0054] Since remote sensing technology has great advantages in obtaining information about time and space, the method of this embodiment can obtain large-area surface information from high altitudes based on remote sensing technology. Through remote sensing images of different bands, the spatial distribution, geometric shape, and spectral characteristics of photovoltaic power stations can be captured. Deep learning, as a powerful artificial intelligence technology, has the ability to learn complex patterns and automatically extract features from large amounts of data, and make predictions based on this. The method of this embodiment combines remote sensing technology with deep learning to achieve efficient and accurate monitoring of photovoltaic product categories.

[0055] The photovoltaic panel category recognition method proposed in this embodiment combines remote sensing technology with deep learning technology to achieve rapid recognition of photovoltaic panel categories in a large area. First, photovoltaic panels of different categories are classified according to their characteristic differences in remote sensing spectra, and then a deep learning model is obtained through training. The remote sensing data of the photovoltaic panel to be identified is input into the trained deep learning model to obtain the photovoltaic panel recognition result output by the deep learning model, thereby realizing the recognition of photovoltaic panel categories in a large area.

[0056] Based on the above photovoltaic panel classification method, this application also discloses a photovoltaic panel classification system based on remote sensing data, such as Figure 3 As shown, the photovoltaic panel category identification system includes: The data acquisition module 310 is used to acquire remote sensing data of one or more photovoltaic panels in the target area; its function is as described in step S1.

[0057] The feature extraction module 320 is used to extract spectral feature information of one or more photovoltaic panels from the remote sensing data; its function is as described in step S2.

[0058] The category identification module 330 is used to determine the category data corresponding to one or more photovoltaic panels in the target area according to the spectral feature information, and its function is as described in step S3.

[0059] Furthermore, the data acquisition module includes: a regional data acquisition unit, a location identification unit and a data screening unit.

[0060] The regional data acquisition unit is used to obtain remote sensing data of the target area.

[0061] A location identification unit, used to identify the location area information of the photovoltaic panel from the remote sensing data of the target area; The data screening unit is used to screen out the remote sensing data of the photovoltaic panel according to the location area information of the photovoltaic panel.

[0062] Furthermore, the feature extraction module includes: a spectral feature extraction unit.

[0063] The spectral feature extraction unit is used to input the remote sensing data into a preset spectral feature information analysis tool to obtain spectral feature information of one or more photovoltaic panels output by the preset spectral feature information analysis tool.

[0064] Furthermore, the method for generating the photovoltaic panel category recognition model includes: Obtain spectral characteristic information of different types of photovoltaic panels; Establish a training data set based on the categories and spectral feature information corresponding to the photovoltaic panels; The preset network model is trained using the training data set to obtain a trained photovoltaic panel category recognition model.

[0065] Furthermore, the preset network model is a convolutional neural network model, a residual network model, a classification model based on machine learning, or a recognition model based on physical features.

[0066] Furthermore, the step of obtaining spectral characteristic information of photovoltaic panels of different categories includes: Collect samples of different types of photovoltaic materials; the photovoltaic materials include: one or more of monocrystalline silicon, polycrystalline silicon, cadmium telluride, and copper indium gallium selenide; Hyperspectral imaging is performed on each category of photovoltaic material samples to obtain spectral characteristic information of each photovoltaic material sample at different wavelengths.

[0067] Furthermore, the step of establishing a training data set according to the category and spectral feature information corresponding to the photovoltaic panel includes: Constructing a first data set according to the categories corresponding to the photovoltaic panels and the spectral characteristic information of different photovoltaic panels at different wavelengths; Acquire spectral remote sensing information of the location of the photovoltaic panels in the specified area, and extract spectral characteristic information of the photovoltaic panels in the specified area based on the spectral remote sensing information; Comparing the spectral characteristic information of the photovoltaic panels in the specified area with the spectral characteristic information of the photovoltaic panels contained in the first data set to identify the type of the photovoltaic panels in the specified area; Labeling the photovoltaic panels in the designated area according to the identified categories and spectral feature information of the photovoltaic panels, and adding the labeled photovoltaic panels to the first data set to obtain a second data set; A training data set is constructed according to the photovoltaic panel information in the second data set.

[0068] Furthermore, the spectral characteristic information includes multiple; The step of inputting the spectral feature information into a trained photovoltaic panel category recognition model to obtain a photovoltaic panel category recognition result output by the photovoltaic panel category recognition model comprises: Each spectral feature information is input into the trained photovoltaic panel category recognition model in turn to obtain the identified photovoltaic panel category corresponding to each spectral feature information, and multiple photovoltaic panel categories constitute a photovoltaic panel category recognition set.

[0069] The system of this embodiment uses hyperspectral remote sensing data to distinguish the categories of photovoltaic panels, and can obtain category distinction data, such as key bands, etc. based on the distinction results, so that better recognition results can be achieved. In addition, this embodiment can realize photovoltaic panel category recognition in a large area, so the recognition efficiency of photovoltaic panel categories can be greatly improved.

[0070] This embodiment provides a computer storage medium, wherein a photovoltaic panel category identification program based on remote sensing data is stored on the computer-readable storage medium, and when the photovoltaic panel category identification program based on remote sensing data is executed by a processor, the steps of the photovoltaic panel category identification method based on remote sensing data are implemented.

[0071] Combination Figure 4The figure shows the use of a full-spectrum ground object spectrometer in an indoor environment using a halogen lamp to simulate sunlight, and then the spectral information is visualized to collect hyperspectral imaging of several photovoltaic panels. From the spectrum diagram, we can preliminarily distinguish the types of several photovoltaic panels in the wavelength range of 600nm-900nm and 1200nm-1500nm. Figure 4 It is proved that the method and system provided by this embodiment have strong feasibility.

[0072] The present invention discloses a photovoltaic panel category recognition method, system and storage medium based on remote sensing data, by acquiring remote sensing data of one or more photovoltaic panels in a target area; extracting spectral feature information of one or more photovoltaic panels from the remote sensing data; inputting the spectral feature information into a trained photovoltaic panel category recognition model to obtain a photovoltaic panel category recognition result output by the photovoltaic panel category recognition model. The method and system disclosed in this embodiment input the acquired remote sensing data of photovoltaic panels into a photovoltaic panel category recognition model obtained by deep learning to achieve rapid recognition of photovoltaic panel categories based on the feature differences presented by different categories of photovoltaic panels in remote sensing data, thereby improving the recognition efficiency of photovoltaic panel categories within a large area, and providing technical support for application scenarios such as monitoring of photovoltaic panel locations or controlling the progress of photovoltaic panel installation. The method and system of this embodiment are particularly suitable for the identification of large-area photovoltaic panel categories and have high applicability.

[0073] Those skilled in the art will readily appreciate other embodiments of the invention after considering the specification and practicing the invention disclosed herein. The present invention is intended to cover any variations, uses or adaptations of the present invention that follow the general principles of the present invention and include common knowledge or customary techniques in the art that are not disclosed in this disclosure. The description and examples are to be considered exemplary only, and the true scope and spirit of the present invention are indicated by the following claims.

[0074] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application.

Claims

1. A photovoltaic panel category identification method based on remote sensing data, characterized in that: include: Obtain remote sensing data of one or more photovoltaic panels in a target area; Extracting spectral characteristic information of one or more photovoltaic panels from the remote sensing data; The spectral feature information is input into a trained photovoltaic panel category recognition model to obtain a photovoltaic panel category recognition result output by the photovoltaic panel category recognition model.

2. The photovoltaic panel category identification method based on remote sensing data according to claim 1, characterized in that: The step of acquiring remote sensing data of one or more photovoltaic panels in the target area includes: Obtain remote sensing data of the target area; Identify the location area information of the photovoltaic panels from the remote sensing data of the target area; The remote sensing data of the photovoltaic panel is filtered out according to the location area information of the photovoltaic panel.

3. The photovoltaic panel category identification method based on remote sensing data according to claim 1, characterized in that: The step of extracting spectral characteristic information of one or more photovoltaic panels from the remote sensing data comprises: The remote sensing data is input into a preset spectral characteristic information analysis tool to obtain spectral characteristic information of one or more photovoltaic panels output by the preset spectral characteristic information analysis tool.

4. The photovoltaic panel category identification method based on remote sensing data according to claim 3 is characterized in that: The method for generating the photovoltaic panel category recognition model includes: Obtain spectral characteristic information of different types of photovoltaic panels; Establish a training data set based on the categories and spectral feature information corresponding to the photovoltaic panels; The preset network model is trained using the training data set to obtain a trained photovoltaic panel category recognition model.

5. The photovoltaic panel category identification method based on remote sensing data according to claim 4 is characterized in that: The preset network model is a convolutional neural network model, a residual network model, a classification model based on machine learning, or a recognition model based on physical features.

6. The photovoltaic panel category identification method based on remote sensing data according to claim 4 is characterized in that: The step of obtaining spectral characteristic information of different types of photovoltaic panels includes: Collect samples of different types of photovoltaic materials; the photovoltaic materials include: one or more of monocrystalline silicon, polycrystalline silicon, cadmium telluride, and copper indium gallium selenide; Hyperspectral imaging is performed on each category of photovoltaic material samples to obtain spectral characteristic information of each photovoltaic material sample at different wavelengths.

7. The photovoltaic panel category identification method based on remote sensing data according to claim 4 is characterized in that: The steps of establishing a training data set based on the categories and spectral feature information corresponding to photovoltaic panels include: Constructing a first data set according to the categories corresponding to the photovoltaic panels and the spectral characteristic information of different photovoltaic panels at different wavelengths; Acquire spectral remote sensing information of the location of the photovoltaic panels in the specified area, and extract spectral characteristic information of the photovoltaic panels in the specified area based on the spectral remote sensing information; Comparing the spectral characteristic information of the photovoltaic panels in the specified area with the spectral characteristic information of the photovoltaic panels contained in the first data set to identify the type of the photovoltaic panels in the specified area; Labeling the photovoltaic panels in the designated area according to the identified categories and spectral feature information of the photovoltaic panels, and adding the labeled photovoltaic panels to the first data set to obtain a second data set; A training data set is constructed according to the photovoltaic panel information in the second data set.

8. The photovoltaic panel category identification method based on remote sensing data according to claim 4 is characterized in that: The spectral feature information includes multiple; The step of inputting the spectral feature information into a trained photovoltaic panel category recognition model to obtain a photovoltaic panel category recognition result output by the photovoltaic panel category recognition model comprises: Each spectral feature information is input into the trained photovoltaic panel category recognition model in turn to obtain the identified photovoltaic panel category corresponding to each spectral feature information, and multiple photovoltaic panel categories constitute a photovoltaic panel category recognition set.

9. A photovoltaic panel category identification system based on remote sensing data, characterized in that: include: A data acquisition module, used to acquire remote sensing data of one or more photovoltaic panels in a target area; A feature extraction module, used to extract spectral feature information of one or more photovoltaic panels from the remote sensing data; The category recognition module is used to determine the category data corresponding to one or more photovoltaic panels in the target area according to the spectral feature information.

10. A computer storage medium, characterized in that: The computer-readable storage medium stores a photovoltaic panel category identification program based on remote sensing data. When the photovoltaic panel category identification program based on remote sensing data is executed by the processor, the steps of the photovoltaic panel category identification method based on remote sensing data as described in any one of claims 1-8 are implemented.

Images