Image generation method, device, photovoltaic system, electronic device, storage medium and computer program product for a photovoltaic system

Abstract

The present application discloses an image generation method, device, photovoltaic system, electronic device, storage medium and computer program product for a photovoltaic system, belonging to the technical field of photovoltaic systems. The image generation method of the photovoltaic system comprises: obtaining a first image corresponding to the photovoltaic system; the first image comprises a feature code image corresponding to each inverter, and the pixel relative position information of each feature code image in the first image matches the actual relative position information of multiple inverters; the first image is input into a target fusion model, and the target fusion model performs multi-layer feature extraction and feature fusion processing on the first image to obtain multiple fused feature images of different scales; based on at least part of the fused feature images of the multiple fused feature images of different scales, a target layout diagram is generated. The image generation method of the photovoltaic system of the present application automatically generates a target layout diagram, improves the efficiency of generating the target layout diagram, reduces delivery time, and improves the user experience.

Description

Technical Field

[0001] The present application belongs to the technical field of photovoltaic systems, and in particular relates to a method and device for generating an image of a photovoltaic system, a photovoltaic system, an electronic device, a storage medium, and a computer program product. Background Art

[0002] An electronic layout diagram for a photovoltaic system allows users to easily view the electrical information of each photovoltaic module in the system. In related technologies, this diagram is typically obtained by manually drawing it on power plant management software. However, this method requires manual operation, is time-consuming, and inefficient, impacting the user experience. Summary of the Invention

[0003] This application aims to solve at least one of the technical problems existing in the related art. To this end, this application proposes a photovoltaic system image generation method, device, photovoltaic system, electronic device, storage medium, and computer program product, which automatically generate a target layout map without manual operation, reduce the time to obtain the target layout map, improve the efficiency of target layout map generation, reduce delivery time, and improve the user experience.

[0004] In a first aspect, the present application provides an image generation method for a photovoltaic system, wherein the photovoltaic system includes a plurality of inverters, each of the inverters being electrically connected to at least one photovoltaic module; the method comprising:

[0005] Acquire a first image corresponding to the photovoltaic system; the first image includes a feature code image corresponding to each of the inverters, and the relative position information of pixels of each of the feature code images in the first image matches the actual relative position information of the multiple inverters;

[0006] Inputting the first image into a target fusion model, and performing multi-layer feature extraction and feature fusion processing on the first image by the target fusion model to obtain a plurality of fused feature images of different scales;

[0007] A target layout map is generated based on at least some of the fused feature images in the plurality of fused feature images at different scales.

[0008] According to the image generation method of the photovoltaic system of the present application, multi-layer feature extraction and feature fusion are performed on the first image that intuitively reflects the actual setting of the photovoltaic system and the connection status of the inverter and the photovoltaic module through the feature code image through the target fusion model, so as to obtain multiple fused feature images of different scales, obtain the rich features and details included in the first image, and then automatically generate the target layout diagram based on the fused feature image without manual operation, thereby reducing the time to obtain the target layout diagram, improving the efficiency of generating the target layout diagram, reducing the delivery time, and improving the user experience.

[0009] According to the photovoltaic system image generation method of the present application, the first image is input into the target fusion model, and the target fusion model performs multi-layer feature extraction and feature fusion processing on the first image to obtain multiple fused feature images of different scales, including:

[0010] The first image is input into the target fusion model, and the target fusion model performs feature extraction and feature fusion processing of at least four modules on the first image to obtain at least four fused feature images of different scales.

[0011] According to the image generation method of the photovoltaic system of the present application, the target fusion model includes: a first feature processing module, a second feature processing module, a third feature processing module and a fourth feature processing module;

[0012] The first feature processing module includes: a first submodule, a second submodule, and a third submodule; the second feature processing module includes: a fourth submodule, a fifth submodule, and a sixth submodule; the third feature processing module includes: a seventh submodule, an eighth submodule, and a ninth submodule; the fourth feature processing module includes: a tenth submodule, an eleventh submodule, and a twelfth submodule;

[0013] The first input port corresponding to the first submodule is used to input the first image;

[0014] The first sub-input port corresponding to the second sub-module is connected to the first sub-output port corresponding to the first sub-module;

[0015] The third input port corresponding to the third submodule is connected to the first suboutput port corresponding to the first submodule and the second output port corresponding to the second submodule respectively;

[0016] The fourth input port corresponding to the fourth submodule is connected to the second suboutput port corresponding to the first submodule;

[0017] The first sub-input port corresponding to the fifth sub-module is connected to the first sub-output port corresponding to the fourth sub-module, and the second sub-output port corresponding to the fifth sub-module is connected to the second sub-input port corresponding to the second sub-module;

[0018] The first sub-input port corresponding to the sixth sub-module is connected to the first sub-output port corresponding to the fifth sub-module and the first sub-output port corresponding to the fourth sub-module respectively; the second sub-input port corresponding to the sixth sub-module is connected to the second sub-output port corresponding to the third sub-module;

[0019] The seventh input port corresponding to the seventh submodule is connected to the second sub-output port corresponding to the fourth submodule;

[0020] The first sub-input port corresponding to the eighth sub-module is connected to the first sub-output port corresponding to the seventh sub-module, and the second sub-output port corresponding to the eighth sub-module is connected to the second sub-input port corresponding to the fifth sub-module;

[0021] The first sub-input port corresponding to the ninth sub-module is connected to the first sub-output port corresponding to the eighth sub-module and the first sub-output port corresponding to the seventh sub-module, and the second sub-input port corresponding to the ninth sub-module is connected to the second sub-output port corresponding to the sixth sub-module;

[0022] The tenth input port corresponding to the tenth submodule is connected to the second sub-output port corresponding to the seventh submodule;

[0023] The eleventh input port corresponding to the eleventh submodule is connected to the tenth output port corresponding to the tenth submodule, and the second sub-output port corresponding to the eleventh submodule is connected to the second sub-input port corresponding to the eighth submodule;

[0024] The first sub-input port corresponding to the twelfth sub-module is connected to the first sub-output port corresponding to the eleventh sub-module and the tenth output port corresponding to the tenth sub-module, and the second sub-input port corresponding to the twelfth sub-module is connected to the second sub-output port corresponding to the ninth sub-module;

[0025] The first sub-output port corresponding to the third sub-module, the first sub-output port corresponding to the sixth sub-module, the first sub-output port corresponding to the ninth sub-module, and the twelfth output port corresponding to the twelfth sub-module are respectively used to output the fused feature images of different scales.

[0026] According to the photovoltaic system image generation method of the present application, the feature fusion is performed based on the following steps:

[0027] The target feature image and the adjacent feature image input to the target feature processing module in the target fusion model are processed using the model parameters and the first constant corresponding to the target feature processing module to obtain a target fusion image; the target feature image is output by the target feature processing module, and the adjacent feature image is output by a module adjacent to the target feature processing module; the model parameters are obtained by training based on multiple sample feature code images.

[0028] According to the photovoltaic system image generation method of the present application, generating a target layout map based on at least some of the fused feature images of the multiple fused feature images of different scales includes:

[0029] performing feature recognition on at least a portion of the fused feature images of the multiple fused feature images of different scales to obtain pixel position information of each feature code image in the at least a portion of the fused feature image and identification information corresponding to each feature code image; the identification information is obtained by recognizing the feature code image and is used to characterize identification information of the inverter and identification information of the photovoltaic module electrically connected to the inverter;

[0030] The target layout map is generated based on the pixel position information and the identification information.

[0031] According to the photovoltaic system image generation method of the present application, the step of performing feature recognition on at least a portion of the fused feature images of the multiple fused feature images of different scales to obtain pixel position information of each feature code image in the at least a portion of the fused feature images includes:

[0032] Calculating a rotation angle corresponding to each of the feature code images based on at least a portion of the positioning points corresponding to each of the feature code images in the first fused feature image in the at least partially fused feature image and an affine transformation matrix;

[0033] Based on each of the rotation angles, the pixel position information of each of the feature code images in the first fused feature image is updated to obtain new pixel position information.

[0034] According to the photovoltaic system image generation method of the present application, after generating the target layout map based on at least some of the fused feature images of the multiple fused feature images of different scales, the method further includes:

[0035] respectively obtaining electrical information collected by each of the inverters;

[0036] The electrical information is added to the target layout diagram according to pixel position information corresponding to each feature code image in the target layout diagram.

[0037] According to the photovoltaic system image generation method of the present application, generating a target layout map based on at least some of the fused feature images of the multiple fused feature images of different scales further includes:

[0038] Inputting at least part of the fused feature images of the plurality of fused feature images of different scales into a target prediction model to obtain the target layout map output by the target prediction model;

[0039] The target prediction model is trained based on at least one of a classification loss function, a confidence loss function, and an intersection-over-union loss function.

[0040] In a second aspect, the present application provides an image generation device for a photovoltaic system, wherein the photovoltaic system includes a plurality of inverters, each of the inverters being electrically connected to at least one photovoltaic module; the device includes:

[0041] a first processing module configured to obtain a first image corresponding to the photovoltaic system; the first image comprising a feature code image corresponding to each of the inverters, wherein pixel relative position information of each of the feature code images in the first image matches actual relative position information of the plurality of inverters;

[0042] a second processing module, configured to input the first image into a target fusion model, and have the target fusion model perform multi-layer feature extraction and feature fusion processing on the first image to obtain a plurality of fused feature images of different scales;

[0043] The third processing module is configured to generate a target layout image based on at least a portion of the fused feature images of the multiple fused feature images of different scales.

[0044] According to the image generation device of the photovoltaic system of the present application, multi-layer feature extraction and feature fusion are performed on the first image that intuitively reflects the actual setting of the photovoltaic system and the connection status of the inverter and the photovoltaic module through the feature code image through the target fusion model, so as to obtain multiple fused feature images of different scales, obtain the rich features and details included in the first image, and automatically generate the target layout diagram based on the fused feature image without manual operation, thereby reducing the time to obtain the target layout diagram, improving the efficiency of generating the target layout diagram, reducing the delivery time, and improving the user experience.

[0045] In a third aspect, the present application provides a photovoltaic system, comprising:

[0046] Multiple photovoltaic panels;

[0047] a plurality of inverters, each of the inverters being electrically connected to at least one of the photovoltaic modules;

[0048] The photovoltaic system generates a layout diagram based on the photovoltaic system image generation method according to the first aspect.

[0049] In a fourth aspect, the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the image generation method for the photovoltaic system as described in the first aspect above is implemented.

[0050] In a fifth aspect, the present application provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the image generation method for a photovoltaic system as described in the first aspect above.

[0051] In a sixth aspect, the present application provides a computer program product, comprising a computer program, which, when executed by a processor, implements the method for generating an image of a photovoltaic system as described in the first aspect above.

[0052] The above one or more technical solutions in the embodiments of the present application have at least one of the following technical effects:

[0053] Through the target fusion model, multi-layer feature extraction and feature fusion are performed on the first image that intuitively reflects the actual setting of the photovoltaic system and the connection status of the inverter and photovoltaic modules through the feature code image, so as to obtain multiple fused feature images of different scales, and obtain the rich features and details included in the first image. Based on the fused feature image, the target layout diagram is automatically generated without manual operation, which reduces the time to obtain the target layout diagram, improves the efficiency of target layout diagram generation, shortens delivery time, and improves user experience.

[0054] Furthermore, the target feature image and the adjacent feature image are processed by the model parameters and the first constant corresponding to the target feature processing module, and the target feature image and the adjacent feature image are effectively fused to obtain a target fusion image, thereby improving the detection accuracy and positioning accuracy of the feature code image in the target fusion image, reducing the information loss of the obtained target fusion image, and enhancing the robustness and adaptability of the target fusion image.

[0055] Furthermore, by performing feature recognition on at least part of the fused feature image, pixel position information and identification information corresponding to the feature code image are obtained, the photovoltaic modules connected to the inverter are effectively determined based on the identification information, and the photovoltaic modules and inverter are automatically paired. There is no need for the user to manually scan the feature codes one by one for pairing, which reduces manual operations, improves the efficiency of generating the target layout diagram, and shortens the delivery time of the target layout diagram.

[0056] Furthermore, by determining multiple positioning points in each feature code image in the first fused image, the rotation angle of the feature code image is accurately calculated based on the positioning points and the affine transformation matrix, thereby updating the pixel position information, making the new pixel position information more accurate, improving the positioning accuracy of the feature code image, and making the subsequently generated target layout map more consistent with the actual setting situation, thereby improving the accuracy of the generated target layout map.

[0057] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become obvious from the description below, or will be learned through practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0059] Figure 1 This is one of the flow charts of the method for generating an image of a photovoltaic system provided in an embodiment of the present application;

[0060] Figure 2 This is one of the principle schematic diagrams of the image generation method for a photovoltaic system provided in an embodiment of the present application;

[0061] Figure 3 This is the second principle schematic diagram of the image generation method for a photovoltaic system provided in an embodiment of the present application;

[0062] Figure 4 This is the third principle schematic diagram of the method for generating an image of a photovoltaic system provided in an embodiment of the present application;

[0063] Figure 5 This is the fourth principle schematic diagram of the image generation method for a photovoltaic system provided in an embodiment of the present application;

[0064] Figure 6 This is the fifth principle schematic diagram of the method for generating an image of a photovoltaic system provided in an embodiment of the present application;

[0065] Figure 7 This is the sixth principle schematic diagram of the method for generating an image of a photovoltaic system provided in an embodiment of the present application;

[0066] Figure 8 This is the second flow chart of the method for generating an image of a photovoltaic system provided in an embodiment of the present application;

[0067] Figure 9 This is the third flow chart of the method for generating an image of a photovoltaic system provided in an embodiment of the present application;

[0068] Figure 10 This is the seventh principle schematic diagram of the image generation method for a photovoltaic system provided in an embodiment of the present application;

[0069] Figure 11 This is the eighth principle schematic diagram of the method for generating an image of a photovoltaic system provided in an embodiment of the present application;

[0070] Figure 12 This is a fourth flow chart of the method for generating an image of a photovoltaic system provided in an embodiment of the present application;

[0071] Figure 13 This is the ninth principle diagram of the method for generating an image of a photovoltaic system provided in an embodiment of the present application;

[0072] Figure 14 This is the tenth principle schematic diagram of the method for generating an image of a photovoltaic system provided in an embodiment of the present application;

[0073] Figure 15This is the eleventh principle diagram of the method for generating an image of a photovoltaic system provided in an embodiment of the present application;

[0074] Figure 16 This is the twelfth principle diagram of the method for generating an image of a photovoltaic system provided in an embodiment of the present application;

[0075] Figure 17 is a schematic structural diagram of an image generating device for a photovoltaic system provided in an embodiment of the present application;

[0076] Figure 18 It is a structural diagram of an electronic device provided in an embodiment of the present application.

[0077] Reference numerals:

[0078] The first input port 901 corresponding to the first submodule; the first sub-input port 9021 corresponding to the second submodule;

[0079] The second sub-input port 9022 corresponding to the second sub-module; the third input port 903 corresponding to the third sub-module;

[0080] The fourth input port 904 corresponding to the fourth submodule; the first sub-input port 9051 corresponding to the fifth submodule;

[0081] The fifth submodule corresponds to the second sub-input port 9052; the sixth submodule corresponds to the first sub-input port 9061;

[0082] The second sub-input port 9062 corresponding to the sixth sub-module; the seventh input port 907 corresponding to the seventh sub-module;

[0083] The first sub-input port 9081 corresponding to the eighth sub-module; the second sub-input port 9082 corresponding to the eighth sub-module;

[0084] The first sub-input port 9091 corresponding to the ninth sub-module; the second sub-input port 9092 corresponding to the ninth sub-module;

[0085] The tenth input port 910 corresponding to the tenth submodule; the eleventh input port 911 corresponding to the eleventh submodule;

[0086] The first sub-input port 9121 corresponding to the twelfth sub-module;

[0087] The second sub-input port 9122 corresponding to the twelfth sub-module;

[0088] The first sub-output port 9211 corresponding to the first sub-module; the second sub-output port 9212 corresponding to the first sub-module;

[0089] The second submodule corresponds to the second output port 922; the third submodule corresponds to the first suboutput port 9231;

[0090] The third submodule corresponds to the second sub-output port 9232; the fourth submodule corresponds to the first sub-output port 9241;

[0091] The fourth submodule corresponds to the second sub-output port 9242; the fifth submodule corresponds to the first sub-output port 9251;

[0092] The fifth submodule corresponds to the second sub-output port 9252; the sixth submodule corresponds to the first sub-output port 9261;

[0093] The sixth submodule corresponds to the second sub-output port 9262; the seventh submodule corresponds to the first sub-output port 9271;

[0094] The seventh submodule corresponds to the second sub-output port 9272; the eighth submodule corresponds to the first sub-output port 9281;

[0095] The eighth submodule corresponds to the second sub-output port 9282; the ninth submodule corresponds to the first sub-output port 9291;

[0096] The second sub-output port 9292 corresponding to the ninth sub-module; the tenth output port 930 corresponding to the tenth sub-module;

[0097] The first sub-output port 9311 corresponding to the eleventh sub-module;

[0098] The second sub-output port 9312 corresponding to the eleventh sub-module;

[0099] The twelfth output port 932 corresponding to the twelfth submodule;

[0100] First submodule 9001; second submodule 9002; third submodule 9003; fourth submodule 9004;

[0101] Fifth submodule 9005; sixth submodule 9006; seventh submodule 9007; eighth submodule 9008;

[0102] Ninth submodule 9009 ; tenth submodule 9010 ; eleventh submodule 9011 ; twelfth submodule 9012 . DETAILED DESCRIPTION

[0103] The following will be combined with the accompanying drawings in the embodiments of the present application to clearly describe the technical solutions in the embodiments of the present application. Obviously, the embodiments described are 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 are within the scope of protection of this application.

[0104] The terms "first," "second," and the like in the specification and claims of this application are used to distinguish similar objects, and are not used to describe a specific order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments of this application can be implemented in an order other than that illustrated or described herein, and that the objects distinguished by "first," "second," and the like are generally of the same type, and do not limit the number of objects; for example, the first object can be one or more. In addition, the term "and / or" in the specification and claims refers to at least one of the connected objects, and the character " / " generally indicates that the objects connected are in an "or" relationship.

[0105] In related technologies, there are two main ways to obtain an electronic layout diagram:

[0106] First, when installing each inverter, a characteristic code label representing its serial number is affixed to the corresponding position on the tabular paper layout diagram; then the user scans the code one by one to form an equipment list, uses a pre-made tabular template to generate an electronic layout draft, and matches the PV modules and inverter serial numbers in the electronic layout diagram one by one, so that the data reported by the inverter to the background can be displayed in the correct relative position on the electronic layout diagram.

[0107] Second, for photovoltaic systems installed on complex rooftops, installers attach the characteristic code label of each inverter to a piece of white paper when installing it, and record as much information as possible about the relative position and placement direction (horizontal or vertical) of the photovoltaic components. They then bring this white paper back to the office, scan the codes one by one to form an equipment list, and manually draw a rough draft of the electronic layout. Finally, they match the photovoltaic components and inverter serial numbers in the electronic layout one by one, so that the data reported by the inverter to the background can be displayed in the correct relative position on the electronic layout.

[0108] However, both of the above methods require a lot of manual work, especially in complex rooftop scenarios, which require traveling back and forth between the office and the installation site. The delivery and commissioning of the photovoltaic system may take several days.

[0109] The following, in conjunction with the accompanying drawings, describes in detail the image generation method of the photovoltaic system, the image generation device of the photovoltaic system, the photovoltaic system, the electronic device, and the readable storage medium provided by the embodiments of the present application through specific embodiments and their application scenarios.

[0110] The method for generating an image of a photovoltaic system may be applied to a terminal, and may be specifically executed by hardware or software in the terminal.

[0111] The terminal includes but is not limited to portable communication devices such as mobile phones or tablet computers. It should also be understood that in some embodiments, the terminal may not be a portable communication device, but a desktop computer.

[0112] In the following embodiments, a terminal including a display and a touch-sensitive surface is described. However, it should be understood that the terminal may include one or more other physical user interface devices such as a physical keyboard, a mouse, and a joystick.

[0113] The image generation method of a photovoltaic system provided in an embodiment of the present application can be executed by the photovoltaic system, or by an image generation device of the photovoltaic system arranged on the photovoltaic system, or by a server electrically connected (wired or wirelessly connected) to the photovoltaic system, or by a user terminal communicatively connected to the photovoltaic system, including but not limited to a mobile terminal and a non-mobile terminal.

[0114] like Figure 1 As shown, the method for generating an image of a photovoltaic system includes: step 110 , step 120 and step 130 .

[0115] A photovoltaic system consists of multiple inverters.

[0116] Each inverter is electrically connected to at least one photovoltaic module.

[0117] For example, an inverter is connected to a photovoltaic panel.

[0118] For another example, an inverter is connected to two or more photovoltaic modules, that is, the inverter is a one-to-many inverter.

[0119] Step 110: Acquire a first image corresponding to the photovoltaic system;

[0120] In this step, the first image includes a feature code image corresponding to each inverter, and the pixel relative position information of each feature code image in the first image matches the actual relative position information of the plurality of inverters.

[0121] The characteristic code image is an image obtained by encoding the relevant information of the inverter into specific image symbols.

[0122] The feature code image may include: barcode, QR code and other forms of identification images.

[0123] The information in the feature code image may include: the serial number of the inverter and the initial password of the inverter wireless hotspot.

[0124] It should be noted that when the number of photovoltaic modules connected to the inverter is greater than one (i.e., a one-to-many inverter), the characteristic code image corresponding to the inverter can be designed accordingly.

[0125] For example, a one-to-many inverter has multiple removable labels with PV channel numbers. The PV channel numbers are used to identify the PV modules connected to the inverter.

[0126] The pixel relative position information is the relative position information between the target feature code image in the first image and the other feature code images in the multiple feature code images.

[0127] The target characteristic code image is any characteristic code image among the multiple characteristic code images.

[0128] The other feature codes are feature code images other than the target feature code image among the multiple feature code images.

[0129] The actual relative positions are the actual relative positions of the inverters in actual application scenarios.

[0130] In actual implementation, based on the actual installation position relationship between the inverter and the photovoltaic assembly, the characteristic code image corresponding to the inverter can be posted on the drawing to obtain the first image.

[0131] The first image may include a tabular image and a non-tabular image.

[0132] The feature code images corresponding to the inverters included in the first tabular image may be posted in a corresponding table.

[0133] The characteristic code image corresponding to each inverter included in the non-tabular image may be posted on a blank paper or a design drawing of the photovoltaic system.

[0134] The design drawings of the photovoltaic system are drawings that record the layout of each photovoltaic component in the photovoltaic system.

[0135] The layout of the photovoltaic modules recorded in the design drawings of the photovoltaic system is the same as the actual installation of the photovoltaic system.

[0136] It should be noted that the design drawings may not necessarily reflect the exact position and relative arrangement angle of the photovoltaic modules from a top-down perspective, but they can reflect the relative position of the photovoltaic modules and at least accurately locate any photovoltaic module.

[0137] The symbols representing the position of each PV module in the design drawing can match the characteristic code size corresponding to the inverter and can reflect the information of whether the PV module is placed horizontally or vertically.

[0138] Positioning point symbols can also be set in the design drawings to more accurately locate the position of each photovoltaic component symbol.

[0139] For a one-to-many photovoltaic system, the labels corresponding to the inverters can be attached to the corresponding photovoltaic component symbols on the design drawings.

[0140] The environments in which photovoltaic systems are set up include: complex environments and simple environments.

[0141] like Figure 2As shown, the rooftop environment where the photovoltaic system is located is relatively simple.

[0142] like Figure 3 As shown in Figure 1, the rooftop environment where the photovoltaic system is located is relatively complex.

[0143] It should be noted that photovoltaic systems installed in complex environments are not aligned horizontally or vertically like photovoltaic systems installed in simple environments. In addition, there are also situations where multiple photovoltaic modules are installed.

[0144] For Figure 2 The first image corresponding to the photovoltaic system shown can be Figure 4 As shown, that is, the first image in the table format, based on the actual installation location of the photovoltaic system, the characteristic code image corresponding to the inverter is posted in the corresponding table, such as in the first column A row, the second column A row, the third column A row and the fourth column A row, etc.

[0145] For Figure 3 For the photovoltaic system shown in FIG. 1 , when the characteristic code image is a barcode image and the first image is a design drawing, the first image corresponding to the photovoltaic system can be as follows: Figure 5 As shown, this is the first image in non-tabular format.

[0146] In the case where the characteristic code image is a barcode image and the first image is a blank drawing, the first image corresponding to the photovoltaic system can be as follows: Figure 6 As shown, this is the first image in non-tabular format.

[0147] In the case where the characteristic code image is a two-dimensional code image and the first image is a design drawing, the first image corresponding to the photovoltaic system can be as follows: Figure 7 As shown, this is the first image in non-tabular format.

[0148] In the actual execution process, taking the first image as a table image as an example, a blank table image can be obtained first. Based on the actual installation situation of the photovoltaic system, the user posts the characteristic code image corresponding to the inverter in the corresponding grid of the table to represent the connection relationship between the photovoltaic module and the inverter, and obtains a tabular image.

[0149] Taking the first image as a non-tabular image as an example, the design drawings of the photovoltaic system can be obtained in advance. Based on the actual installation situation of the photovoltaic system, the user can post the feature code image corresponding to the inverter at the corresponding position of the design drawing to obtain a non-tabular image, such as Figure 5 The first image is shown.

[0150] In the actual implementation process, the characteristic code corresponding to the inverter can be posted at the corresponding position on the white paper according to the actual installation situation of the photovoltaic system to indicate the relative position relationship between the photovoltaic components, such as Figure 6 The first image is shown.

[0151] Step 120: Input the first image into the target fusion model, and perform multi-layer feature extraction and feature fusion processing on the first image by the target fusion model to obtain multiple fused feature images of different scales;

[0152] In this step, the target fusion model processes the first image to obtain a model with more comprehensive and richer features of the first image.

[0153] The target fusion model can be a neural network model or a deep learning model.

[0154] The target fusion model can also be obtained by fusing the feature extraction model and the feature fusion model.

[0155] Among them, the feature extraction model can be a convolutional neural network, a recurrent neural network, and a feature extraction network formed by combining CSPDarknet53 as the basic structure.

[0156] The feature fusion model can be the additive fusion model in ResNet and FPN, the compact bilinear pooling (CBP), and the weighted bidirectional feature pyramid model (BiFPN).

[0157] The specific selection of the target fusion model can be determined based on actual conditions and is not limited in this application.

[0158] Multi-layer feature extraction is an operation of extracting low-level features and high-level features from the first image respectively.

[0159] The low-level features may include features such as edges, corners, and textures of each feature code image in the first image.

[0160] Advanced features may include information such as the shape, position, overall structure, and features of the extracted feature code image.

[0161] The features extracted from each layer are different.

[0162] It should be noted that the number of layers for multi-layer feature extraction can be determined based on actual conditions and is not limited in this application.

[0163] Feature fusion is to fuse the features extracted from each layer to obtain more accurate features.

[0164] Feature fusion can include: additive fusion, multiplicative fusion, splicing fusion and weighted fusion.

[0165] It should be noted that by efficiently fusing feature maps from different scales, the extracted low-level features and high-level features can be effectively combined, thereby better processing feature codes at different positions and sizes in the first image.

[0166] During the actual execution process, a suitable feature fusion method can be selected based on the actual situation of the first image, and this application does not limit it.

[0167] The fused feature image is an image obtained by fusing the features extracted from each layer.

[0168] It should be noted that the scale of the fused feature image obtained will be different depending on the level of feature fusion.

[0169] Taking the target fusion model for three-layer feature extraction and feature fusion processing as an example, the scale of the fused feature image obtained in the first layer may be 320×320; the scale of the fused feature image obtained in the second layer may be 160×160, and the scale of the fused feature image obtained in the third layer may be 80×80.

[0170] The specific size of the feature scale obtained by feature extraction at each layer can be customized by the user or determined based on actual conditions, and is not limited in this application.

[0171] In the actual execution process, after obtaining the first image, the first image can be input into the target fusion model, and the target fusion model is used to perform multi-layer feature extraction and feature fusion on the first image to obtain a fused feature image with richer semantics and more details.

[0172] Step 130 : Generate a target layout map based on at least some of the fused feature images in the plurality of fused feature images at different scales.

[0173] In this step, at least part of the fused feature images is one or more fused feature images among a plurality of fused feature images of different scales.

[0174] The specific number of at least partially fused feature images can be determined based on actual conditions. For example, the number of at least partially fused feature images can be 2 or 4, etc., which is not limited in this application.

[0175] The target layout diagram is an electronic layout diagram that records the installation status of each photovoltaic module in the photovoltaic system.

[0176] During the actual execution process, after obtaining multiple fused feature images of different scales, at least part of the fused feature images can be screened from the multiple fused feature images of different scales based on the clarity of each fused feature image and the features included in each fused feature image, and feature recognition can be performed on at least part of the fused images, such as identifying feature code images, text, feature code size and other information in at least part of the fused images. Based on the recognition results, a drawing tool is called to automatically generate a target layout diagram.

[0177] During the actual execution process, after obtaining the target layout diagram, the position and angle of the photovoltaic component symbols in the target layout diagram can be manually fine-tuned to make the target layout diagram closer to the actual installation situation or more in line with the user's usage preferences.

[0178] During the research and development process, the inventors discovered that in related technologies, the electronic layout diagram is mainly obtained by the user manually drawing the electronic layout diagram on the power station management software; however, the above method requires manual operation, is time-consuming, inefficient, and affects the user experience.

[0179] The present application obtains the first image corresponding to the photovoltaic system to effectively obtain the relative position information of the installation of each photovoltaic component in the photovoltaic system, and determines the connection status of each photovoltaic component and the inverter through the feature code image corresponding to the inverter in the first image; thereby, the first image is subjected to multi-layer feature extraction and feature fusion processing through the target fusion model, and on the basis of capturing the detail features of the first image at different levels, the detail features at different levels are fused to generate a fused feature image with more comprehensive representation capability and resistance to noise interference; based on one or more fused feature images, a target layout diagram is automatically generated without manual operation, thereby reducing the time to obtain the target layout diagram, improving the generation efficiency of the target layout diagram, and being able to ensure the accuracy of the generated target layout diagram, thereby improving the user experience.

[0180] According to the image generation method of the photovoltaic system provided in the embodiment of the present application, multi-layer feature extraction and feature fusion are performed on the first image that intuitively reflects the actual setting of the photovoltaic system and the connection status of the inverter and the photovoltaic module through the feature code image through the target fusion model, so as to obtain multiple fused feature images of different scales, obtain the rich features and details included in the first image, and automatically generate the target layout diagram based on the fused feature image without manual operation, thereby reducing the time to obtain the target layout diagram, improving the efficiency of generating the target layout diagram, reducing the delivery time, and improving the user experience.

[0181] In some embodiments, step 120 may further include:

[0182] The first image is input into the target fusion model, and the target fusion model performs feature extraction and feature fusion processing of at least four modules on the first image to obtain at least four fused feature images of different scales.

[0183] In this embodiment, during the actual execution process, after obtaining the first image, the first image is input into the target fusion model, and feature extraction and feature fusion processing are performed on the first image through the modules included in the target fusion model.

[0184] It should be noted that each module includes: a feature extraction model and a feature fusion model.

[0185] The input end of the feature fusion model is connected to the output end of the feature extraction module.

[0186] like Figure 8 As shown in the figure, taking the target fusion model using CSPDarknet53 as the basic structure to form a feature extraction network as an example, the target fusion model can perform feature extraction of four modules to obtain large-scale features, medium-scale features, small-scale features and extremely small-scale features.

[0187] Medium-scale features can be obtained by convolving large-scale features, small-scale features can be obtained by convolving medium-scale features, and extremely small-scale features can be obtained by convolving small-scale features. After each module obtains features of the corresponding scale, each module can fuse the features of the corresponding scale to obtain a fused image of different scales.

[0188] Smaller-scale features can also be upsampled to obtain larger-scale features. For example, very small-scale features can be upsampled to obtain small-scale features.

[0189] Of course, in the actual implementation process, you can also Figure 8 The target fusion model shown includes four modules, and more modules are set up to extract features smaller than extremely small scale features and improve the accuracy of the acquired features.

[0190] The number of modules included in the target fusion model can be determined based on actual conditions. For example, the target fusion model can include 3 modules or 5 modules, etc.; this application does not limit this.

[0191] It should be noted that, the more modules the target fusion model includes, the more obvious and clearer the features of the fusion feature image output by the target fusion model will be.

[0192] After obtaining features of different scales, the features of different scales can be fused separately through weighted fusion, multiplication fusion, and splicing fusion to obtain at least four fused feature images.

[0193] According to the image generation method for a photovoltaic system provided in an embodiment of the present application, feature extraction and feature fusion of at least four modules are performed on the first image through a target fusion model, thereby effectively improving the image quality of the fused feature image output by the target fusion model. This enables the target fusion model to extract features of the first image at different levels and fuse features of different scales, thereby improving the adaptability of the fused feature image.

[0194] In some embodiments, the target fusion model may include: a first feature processing module, a second feature processing module, a third feature processing module, and a fourth feature processing module.

[0195] In this embodiment, the first feature processing module includes: a first submodule 9001 , a second submodule 9002 , and a third submodule 9003 .

[0196] The second feature processing module includes: a fourth submodule 9004 , a fifth submodule 9005 and a sixth submodule 9006 .

[0197] The third feature processing module includes: a seventh submodule 9007 , an eighth submodule 9008 and a ninth submodule 9009 .

[0198] The fourth feature processing module includes: a tenth submodule 9010 , an eleventh submodule 9011 and a twelfth submodule 9012 .

[0199] The first input port 901 corresponding to the first submodule is used to input the first image.

[0200] The first sub-input port 9021 corresponding to the second sub-module is connected to the first sub-output port 9211 corresponding to the first sub-module.

[0201] The third input port 903 corresponding to the third submodule is connected to the first suboutput port 9211 corresponding to the first submodule and the second output port 922 corresponding to the second submodule respectively.

[0202] The fourth input port 904 corresponding to the fourth submodule is connected to the second suboutput port 9212 corresponding to the first submodule.

[0203] The first sub-input port 9051 corresponding to the fifth sub-module is connected to the first sub-output port 9241 corresponding to the fourth sub-module.

[0204] The second sub-output port 9252 corresponding to the fifth sub-module is connected to the second sub-input port 9022 corresponding to the second sub-module.

[0205] The first sub-input port 9061 corresponding to the sixth sub-module is connected to the first sub-output port 9251 corresponding to the fifth sub-module and the first sub-output port 9241 corresponding to the fourth sub-module respectively.

[0206] The second sub-input port 9062 corresponding to the sixth sub-module is connected to the second sub-output port 9232 corresponding to the third sub-module.

[0207] The seventh input port 907 corresponding to the seventh submodule is connected to the second suboutput port 9242 corresponding to the fourth submodule.

[0208] The first sub-input port 9081 corresponding to the eighth sub-module is connected to the first sub-output port 9271 corresponding to the seventh sub-module.

[0209] The second sub-output port 9282 corresponding to the eighth sub-module is connected to the second sub-input port 9052 corresponding to the fifth sub-module.

[0210] The first sub-input port 9091 corresponding to the ninth sub-module is respectively connected to the first sub-output port 9281 corresponding to the eighth sub-module and the first sub-output port 9271 corresponding to the seventh sub-module.

[0211] The second sub-input port 9092 corresponding to the ninth sub-module is connected to the second sub-output port 9262 corresponding to the sixth sub-module.

[0212] The tenth input port 910 corresponding to the tenth submodule is connected to the second suboutput port 9272 corresponding to the seventh submodule.

[0213] The eleventh input port 911 corresponding to the eleventh submodule is connected to the tenth output port 930 corresponding to the tenth submodule.

[0214] The second sub-output port 9312 corresponding to the eleventh sub-module is connected to the second sub-input port 9082 corresponding to the eighth sub-module.

[0215] The first sub-input port 9121 corresponding to the twelfth sub-module is connected to the first sub-output port 9311 corresponding to the eleventh sub-module and the tenth output port 930 corresponding to the tenth sub-module respectively.

[0216] The second sub-input port 9122 corresponding to the twelfth sub-module is connected to the second sub-output port 9292 corresponding to the ninth sub-module.

[0217] The first sub-output port 9231 corresponding to the third sub-module, the first sub-output port 9261 corresponding to the sixth sub-module, the first sub-output port 9291 corresponding to the ninth sub-module, and the twelfth output port 932 corresponding to the twelfth sub-module are respectively used to output fused feature images of different scales.

[0218] like Figure 8 As shown, the fused feature images of different scales output by the first sub-output port 9231 corresponding to the third sub-module, the first sub-output port 9261 corresponding to the sixth sub-module, the first sub-output port 9291 corresponding to the ninth sub-module, and the twelfth output port 932 corresponding to the twelfth sub-module can be input into the prediction module.

[0219] The prediction module generates a target layout map based on multiple fused feature images of different scales.

[0220] In some embodiments, the first feature processing module, the second feature processing module, the third feature processing module, and the fourth feature processing module may further include more sub-modules.

[0221] For example, the first feature processing module, the second feature processing module, the third feature processing module, and the fourth feature processing module are each provided with four submodules or five submodules, and the connection method between the modules is the same as that of the three submodules.

[0222] That is, the sub-modules in each feature processing module are connected in sequence, and the subsequent sub-modules can receive the outputs of the previous sub-modules. The sub-modules in the target feature processing module in the first feature processing module, the second feature processing module, the third feature processing module and the fourth feature processing module are respectively connected to the sub-modules in the upper-layer feature processing module and the lower-layer feature processing module. To avoid repetition, they are not described here.

[0223] According to the image generation method for a photovoltaic system provided in an embodiment of the present application, by setting the target fusion model to a model including four modules, feature extraction and feature fusion are performed on the first image through multiple modules to obtain multiple fused feature images of different scales, thereby improving the accuracy of the obtained fused feature images, reducing the loss of the fused feature images, and enhancing the adaptability of the fused feature images.

[0224] In some embodiments, feature fusion may be performed based on the following steps:

[0225] The target feature image and the adjacent feature images input to the target feature processing module are processed using the model parameters and the first constant corresponding to the target feature processing module in the target fusion model to obtain a target fusion image;

[0226] In this embodiment, the target feature processing module is any one of the multiple modules included in the target fusion model.

[0227] The model parameters are obtained by training based on multiple sample feature code images.

[0228] The model parameters may be learnable parameters corresponding to the target feature processing module.

[0229] The first constant is a preset value.

[0230] The specific value of the first constant can be determined based on actual conditions and is not limited in this application.

[0231] The target feature image is output by the target feature processing module.

[0232] The adjacent feature image is output by the module adjacent to the target feature processing module.

[0233] The target feature image is a feature image extracted from the first image by a feature extraction model corresponding to the target feature processing module.

[0234] The adjacent feature image is a feature extraction model corresponding to a module adjacent to the target feature processing module, and is a feature image extracted from the first image.

[0235] The adjacent feature image can be a feature image extracted from the first image by a feature extraction model corresponding to the previous module of the target feature processing module; or it can be a feature image extracted from the first image by a feature extraction model corresponding to the next module of the target feature processing module.

[0236] like Figure 9 As shown in the figure, taking the target fusion model as an example of feature fusion through the weighted bidirectional feature pyramid (BiFPN) structure, the medium-scale feature P2_1 can be obtained based on the following formula:

[0237]

[0238] The medium-scale feature P2_2 can be obtained based on the following formula:

[0239]

[0240] Among them, Pi_j is the target fusion image; is the target feature image; Pi+1_j and Pi-1_j are adjacent feature images; Wi is the model parameter corresponding to the target feature processing module, and ϵ is the first constant used to ensure that the denominator is not a minimum value of zero.

[0241] It should be noted that the fused feature images of other modules can be obtained by the same calculation method as P2_1 and P2_2. To avoid repetition, they are not described here.

[0242] BiFPN uses a learnable weighted fusion mechanism to integrate features from different scales; this means that BiFPN can more accurately select which features need to be retained and which features should be suppressed, especially when the signature image is small or of low quality; thereby improving the positioning accuracy of small targets (such as the signature image in the first image).

[0243] Furthermore, BiFPN employs bidirectional feature propagation (i.e., top-down and bottom-up), effectively reducing the loss of information contained in the first image as it is transferred between feature layers (modules). For example, minimizing the loss of signature image information is particularly important during signature image detection, as signature images often contain subtle, high-frequency information (such as fine edges). BiFPN can better preserve these details, reducing the loss of key features due to downsampling.

[0244] When the collected signature image is affected by factors such as viewing angle, illumination changes, and blur, BiFPN can enhance the model's adaptability to these changes by fusing multi-scale features. That is, no matter how the angle of the signature image changes or how the image's illumination changes, BiFPN can extract sufficient useful information, improve the robustness of the extracted signature image features, and avoid feature extraction failure in complex environments.

[0245] BiFPN uses an efficient feature fusion method and an adaptive weighting mechanism to alleviate the redundant information problem that may be caused by simple additive fusion in traditional FPN. The weighting method of BiFPN allows the key information in the signature image to receive more attention, thereby improving the detection effect.

[0246] BiFPN's feature fusion method is highly adaptable and can automatically adjust the feature fusion strategy according to the size and position of the signature image in the first image and the complexity of the background; thus ensuring that no matter how the size and clarity of the signature image changes, the model can be flexibly adjusted to obtain the optimal target fused image.

[0247] BiFPN can more accurately fuse features from different layers, enabling the model to not only detect the presence of signature images but also improve positioning accuracy. Accurate positioning is particularly important for detecting signature images, as it is subsequently necessary to precisely define the position and orientation of the signature image for decoding and obtaining the information contained in the signature image.

[0248] According to the image generation method for a photovoltaic system provided in an embodiment of the present application, the target feature image and the adjacent feature images are processed by using the model parameters and the first constant corresponding to the target feature processing module, and the target feature image and the adjacent feature images are effectively fused to obtain a target fused image, thereby improving the detection accuracy and positioning accuracy of the feature code image in the target fused image, reducing the information loss of the obtained target fused image, and enhancing the robustness and adaptability of the target fused image.

[0249] In some embodiments, step 130 may further include:

[0250] Performing feature recognition on at least a portion of the fused feature images of the plurality of different scales to obtain pixel position information of each feature code image in at least a portion of the fused feature image and identification information corresponding to each feature code image;

[0251] A target layout map is generated based on the pixel position information and the identification information.

[0252] In this embodiment, the pixel position information is the pixel coordinate information of the feature code image in the fused feature image.

[0253] The pixel position information may include coordinate information of each vertex of the feature code image, one vertex of the feature code image, and information such as the width and height of the feature code image.

[0254] Of course, the pixel position information may also be other information that can locate the feature code image from the fused feature image, and this application does not limit this.

[0255] The identification information is obtained by identifying the characteristic code image and is used to characterize the identification information of the inverter and the identification information of the photovoltaic components electrically connected to the inverter.

[0256] The identification information of the inverter may be information such as the serial number of the inverter and the serial number of the inverter.

[0257] In actual implementation, multiple identification information may be obtained by respectively decoding each feature code image in at least a portion of the fused feature image.

[0258] During actual execution, a fused feature image with higher definition may be selected from at least part of the fused feature images as an image for decoding the feature code image.

[0259] After determining the fused feature image for decoding, the fused feature image can be cropped to separate the feature code images to more accurately obtain the identification information.

[0260] The inverter is automatically paired with the PV panels it is connected to through the identified identification information.

[0261] After obtaining the position information and identification information of each pixel, the drawing tool is called to automatically generate the target layout diagram.

[0262] For Figure 2 The photovoltaic system shown can generate Figure 10 The target layout diagram is shown.

[0263] For Figure 3 The photovoltaic system shown can generate Figure 11 The target layout diagram is shown.

[0264] like Figure 12 As shown, in the actual execution process, after obtaining at least part of the fused feature image, at least part of the fused feature image is input into the prediction module, and based on the pixel position information corresponding to the feature code image, at least part of the fused feature image is cropped to obtain the original feature code area; the original feature code area is scaled to different degrees, such as reducing it by 0.5 times or enlarging it by 2 times; after obtaining the 0.5 times feature code area, the 0.25 times feature code area and the 2 times feature code area, the feature code area is decoded by the pyzbar library to obtain the identification information.

[0265] In the case of decoding failure, binarization, sharpening and Gaussian noise processing may be performed on the original feature code region and the scaled feature code region, and the processed feature code region may be decoded.

[0266] It should be noted that binarization is to process the feature code area into an image of black and white pixels, which is conducive to decoding. Sharpening and Gaussian noise processing are used to blur the feature code area, which is also conducive to decoding.

[0267] According to the image generation method of the photovoltaic system provided in the embodiment of the present application, by performing feature recognition on at least a portion of the fused feature image, pixel position information and identification information corresponding to the feature code image are obtained, the photovoltaic components connected to the inverter are effectively determined based on the identification information, and the photovoltaic components and the inverter are automatically paired. There is no need for the user to manually scan the feature codes one by one for pairing, which reduces manual operations, improves the efficiency of generating the target layout diagram, and shortens the delivery time of the target layout diagram.

[0268] In some embodiments, performing feature recognition on at least a portion of the fused feature images of a plurality of fused feature images of different scales to obtain pixel position information of each feature code image in at least a portion of the fused feature images may further include:

[0269] Calculating a rotation angle corresponding to each feature code image based on at least a portion of the positioning points corresponding to each feature code image in the first fused feature image in at least a portion of the fused feature image and an affine transformation matrix;

[0270] Based on each rotation angle, pixel position information of each feature code image in the first fused feature image is updated to obtain new pixel position information.

[0271] In this embodiment, the first fused feature image is an image with higher image quality among at least a portion of the fused feature images.

[0272] During actual execution, a clarity threshold may be set, and the clarity of each fused feature image may be compared with the clarity threshold to screen out the first fused feature image that meets the clarity threshold.

[0273] The clarity of multiple fused feature images of different scales may also be sorted, and the fused feature image with the greatest clarity may be selected as the first fused feature image.

[0274] Of course, in actual implementation, the first fused feature image may be determined from a plurality of fused feature images of different scales in any other feasible manner.

[0275] At least some of the positioning points are multiple vertices of the feature code image in the first fused image.

[0276] The affine transformation matrix is a matrix that combines rotation, scaling, and translation to achieve coordinate transformation.

[0277] The rotation angle is the rotation angle of the feature code image in the horizontal direction or the vertical direction.

[0278] The new pixel position information includes rotation angle information and is used to locate the pixel position information of the feature code image in the first fused feature image.

[0279] It should be noted that, unlike tabular images, feature code images in non-tabular images may have various angles in addition to horizontal and vertical, and therefore the rotation angle of the feature code images needs to be calculated.

[0280] For Figure 13 The characteristic code image shown can be selected as follows Figure 14 At least some of the positioning points A, B and C shown are used to calculate the rotation angle corresponding to the feature code image through the affine transformation matrix.

[0281] Continue with Figure 14 Taking at least some of the positioning points shown as an example, the angles of the triangle determined by the positioning points A, B, and C can be obtained through vectors. Considering the gradient distortion caused by the shooting angle of the actual scene, the triangle composed of at least some of the positioning points is not a regular right triangle, but the angle size still satisfies that the angle corresponding to the maximum angle is a right angle of the feature code image, and the maximum angle is the angle A corresponding to the positioning point A.

[0282] For the two positioning points other than the maximum angle, namely positioning point B and positioning point C, based on the center point of the three positioning points, it can be deduced that the first positioning point in the counterclockwise direction is C and the second positioning point in the counterclockwise direction is B.

[0283] The center point is the angle with the anchor point A being 0.

[0284] In actual execution, the angles of the triangle can be calculated using the following formulas:

[0285]

[0286]

[0287]

[0288] in, It is a directed line segment from anchor point A to anchor point B; is a directed line segment from anchor point A to anchor point C; is a directed line segment from anchor point B to anchor point A; is a directed line segment from anchor point B to anchor point C; is a directed line segment from anchor point C to anchor point B; is a directed line segment from the anchor point C to the anchor point A; is angle A; is angle B; It is angle C.

[0289] After determining the three vertices, the rotation angle of the feature code image is calculated using the affine transformation matrix.

[0290] Any one of at least some of the positioning points is selected as the coordinate origin, for example, the positioning point C is selected, and then the numerical direction is used as the y-axis and the horizontal direction is used as the x-axis.

[0291] In actual implementation, the affine transformation matrix can be determined based on the following formula:

[0292]

[0293] in, is the scaling factor of the feature code image; The times of counterclockwise rotation around the origin based on the scaled feature code image; is the horizontal translation vector; is the vertical translation vector; is the horizontal direction; y is the vertical direction.

[0294] The above formula can be transformed into:

[0295] Among them, a, b, c, d, e and f are unknown quantities.

[0296] Since at least some of the positioning points of the signature image are not collinear, and the above-mentioned affine transformation matrix has 6 degrees of freedom, a set of linear equations can be obtained from the three points, and a unique solution can be obtained.

[0297] The linear equation can be determined based on the following formula:

[0298]

[0299]

[0300]

[0301]

[0302]

[0303]

[0304] Solve to get a, b, c, d, e and f; to get Figure 14 The triangle BAC shown is rotated around the origin (C) to Figure 15 The angle required for the triangle B'A'C' shown is the rotation angle.

[0305] The rotation angle can be calculated based on the following formula:

[0306]

[0307] in, is the rotation angle.

[0308] According to the image generation method of the photovoltaic system provided in the embodiment of the present application, by respectively determining multiple positioning points in each feature code image in the first fused image, the rotation angle of the feature code image is accurately calculated based on the positioning points and the affine transformation matrix, thereby updating the pixel position information, making the new pixel position information more accurate, improving the positioning accuracy of the feature code image, making the subsequently generated target layout diagram more consistent with the actual setting situation, and improving the accuracy of the generated target layout diagram.

[0309] In some embodiments, after step 130, the method may further include:

[0310] Obtain electrical information collected by each inverter separately;

[0311] The electrical information is added to the target layout diagram according to the pixel position information corresponding to each feature code image in the target layout diagram.

[0312] In this embodiment, the electrical information is data collected by the inverter related to the power generation status of the photovoltaic components.

[0313] The electrical information may include electrical data of the DC side and the AC side.

[0314] The electrical data on the DC side may include voltage, current, power and other data.

[0315] The electrical data measured by AC can include: grid voltage and grid connection point power, etc.

[0316] The specific content of the electrical information can be determined based on actual conditions and is not limited in this application.

[0317] The target layout diagram generated is as follows Figure 11 For example, based on the connection relationship between the inverter and the photovoltaic module, the electrical information collected by the inverter is added to the corresponding position in the target layout diagram.

[0318] For example, if the PV modules connected to the inverter with serial number (i.e., identification information) 012345678932 are the first vertical row of PV modules, the electrical information collected by the inverter will be displayed at the position corresponding to the first vertical row of PV modules.

[0319] According to the image generation method of the photovoltaic system provided in the embodiment of the present application, by obtaining the electrical information collected by the inverter, the information collected by the inverter is displayed corresponding to the pixel position information of the photovoltaic components connected to the inverter at the target layout diagram, so that the user can intuitively understand the power generation status of each photovoltaic component, thereby improving the user experience; it is also convenient to quickly locate the faulty photovoltaic component or inverter in the event of a photovoltaic system failure, thereby reducing the losses caused by the photovoltaic system failure.

[0320] In some embodiments, step 130 may further include:

[0321] Inputting at least part of the fused feature images of the multiple fused feature images of different scales into a target prediction model to obtain a target layout map output by the target prediction model;

[0322] The target prediction model is trained based on at least one of a classification loss function, a confidence loss function, and an intersection-over-union loss function.

[0323] In this embodiment, the target prediction model is a model that predicts the fused feature image and generates a target layout map.

[0324] In such Figure 8 In the prediction module shown, the target prediction model can be encapsulated.

[0325] The classification loss function is a function that measures the classification accuracy of the recognition results of the target prediction model for the fusion feature image.

[0326] The classification loss function can be a cross entropy loss function, such as the focal loss function and the label smoothing loss function.

[0327] The classification loss function can be obtained by fusing different cross entropy loss functions, for example, fusing the focus loss function and the label smoothing loss function to obtain the classification loss function.

[0328] In actual implementation, the classification loss function can be determined based on the following formula:

[0329]

[0330] in, is the classification loss value; is the number of pixels of the fused feature image in the target prediction model, and C is the number of categories of the target prediction model; If it is category c, it is 1, otherwise it is 0; is the predicted probability.

[0331] The categories of the target prediction model may include: the presence of a feature code image and the absence of a feature code image.

[0332] The confidence loss function is a function that measures the difference between the confidence of the prediction result of the target prediction model output for the fused feature image and the actual label.

[0333] The confidence loss functions include: confidence loss function in the form of mean square error (MSE) and confidence loss function in the form of cross entropy.

[0334] The actual choice of the confidence loss function can be determined based on actual conditions and is not limited in this application.

[0335] In actual implementation, the confidence loss function can be determined based on the following formula:

[0336]

[0337] in, is the confidence loss value; If the target actually exists in the pixel area of the first image corresponding to the fused feature map, it is 1, otherwise it is 0; is the probability of predicting the existence of the target.

[0338] The intersection-over-union loss function is a function that measures the degree of overlap between the feature code images in the fused feature image.

[0339] like Figure 16 The figure shows the definition of the loss function for the predicted position, where c is the diagonal of the bounding box between the true value box and the predicted box, and d is the distance between the center points of the true value box and the predicted box.

[0340] In actual implementation, the intersection-over-union loss function can be determined based on the following formula:

[0341]

[0342] in, is the intersection-over-union loss value; N is the predicted N boxes; is the intersection and union fraction; is the Euclidean distance of the bright spot; b is the coordinate of the center point of the prediction box, is the coordinate of the center point of the real frame; is the width of the outer frame, The height of the bounding box.

[0343] After determining the classification loss function, confidence loss function and intersection-over-union loss function, one or more of the classification loss function, confidence loss function and intersection-over-union loss function can be fused to obtain the target loss function, and the target prediction model can be optimized through the target loss function.

[0344] The target loss function is a function obtained by fusing one or more of the classification loss function, the confidence loss function, and the intersection-over-union loss function.

[0345] For example, the target loss function may be any one of a classification loss function, a confidence loss function, and an intersection-over-union loss function.

[0346] For example, the target loss function can also be obtained by fusing the classification loss function and the confidence loss function.

[0347] In the actual implementation process, the target loss function can also be determined based on the following formula:

[0348]

[0349] in, 、 as well as is the weight coefficient.

[0350] In the actual implementation process, multiple samples of different scales can be collected in advance to fuse feature images, and the machine learning model or neural network model can be trained with the output weight coefficient as the goal to obtain the weight coefficient.

[0351] The specific value of the weight coefficient can also be customized by the user or determined based on actual conditions, and this application does not limit it.

[0352] According to the image generation method of the photovoltaic system provided in the embodiment of the present application, the prediction accuracy of the target prediction model is optimized and the accuracy of the target layout map output by the target prediction model is improved by training the target prediction model with one or more of the classification loss function, the confidence loss function and the intersection-over-union loss function.

[0353] The photovoltaic system image generation method provided in the embodiment of the present application can be executed by an image generation device of the photovoltaic system. In the embodiment of the present application, the photovoltaic system image generation method is executed by an image generation device of the photovoltaic system as an example to illustrate the image generation device of the photovoltaic system provided in the embodiment of the present application.

[0354] An embodiment of the present application also provides an image generating device for a photovoltaic system.

[0355] A photovoltaic system consists of multiple inverters.

[0356] Each inverter is electrically connected to at least one photovoltaic module.

[0357] like Figure 17 The image generation device of the photovoltaic system includes: a first processing module 1710 , a second processing module 1720 and a third processing module 1730 .

[0358] A first processing module 1710 is configured to obtain a first image corresponding to the photovoltaic system; the first image includes a signature image corresponding to each inverter, and the relative position information of pixels of each signature image in the first image matches the actual relative position information of the plurality of inverters;

[0359] The second processing module 1720 is configured to input the first image into a target fusion model, and the target fusion model performs multi-layer feature extraction and feature fusion processing on the first image to obtain a plurality of fused feature images of different scales;

[0360] The third processing module 1730 is configured to generate a target layout image based on at least a portion of the fused feature images of the multiple fused feature images of different scales.

[0361] According to the image generation device of the photovoltaic system provided in the embodiment of the present application, multi-layer feature extraction and feature fusion are performed on the first image that intuitively reflects the actual setting of the photovoltaic system and the connection status of the inverter and the photovoltaic module through the feature code image through the target fusion model, so as to obtain multiple fused feature images of different scales, obtain the rich features and details included in the first image, and automatically generate the target layout diagram based on the fused feature image without manual operation, thereby reducing the time to obtain the target layout diagram, improving the efficiency of generating the target layout diagram, reducing the delivery time, and improving the user experience.

[0362] In some embodiments, the second processing module 1720 may also be configured to:

[0363] The first image is input into the target fusion model, and the target fusion model performs feature extraction and feature fusion processing of at least four modules on the first image to obtain at least four fused feature images of different scales.

[0364] In some embodiments, the second processing module 1720 may also be configured to:

[0365] The target feature image and the adjacent feature image input to the target feature processing module are processed by the model parameters and the first constant corresponding to the target feature processing module in the target fusion model to obtain a target fusion image; the target feature image is output by the target feature processing module, and the adjacent feature image is output by the module adjacent to the target feature processing module; the model parameters are obtained by training based on multiple sample feature code images.

[0366] In some embodiments, the third processing module 1730 may also be configured to:

[0367] Performing feature recognition on at least a portion of the fused feature images of the multiple different scales to obtain pixel position information of each feature code image in the at least a portion of the fused feature images and identification information corresponding to each feature code image; the identification information is obtained by recognizing the feature code images and is used to characterize the inverter and the photovoltaic modules electrically connected to the inverter;

[0368] A target layout map is generated based on the pixel position information and the identification information.

[0369] In some embodiments, the third processing module 1730 may also be configured to:

[0370] Calculating a rotation angle corresponding to each feature code image based on at least a portion of the positioning points corresponding to each feature code image in the first fused feature image in at least a portion of the fused feature image and an affine transformation matrix;

[0371] Based on each rotation angle, pixel position information of each feature code image in the first fused feature image is updated to obtain new pixel position information.

[0372] In some embodiments, the apparatus may further include a fourth processing module configured to:

[0373] Obtain electrical information collected by each inverter separately;

[0374] The electrical information is added to the target layout diagram according to the pixel position information corresponding to each feature code image in the target layout diagram.

[0375] In some embodiments, the third processing module 1730 may also be configured to:

[0376] Inputting at least part of the fused feature images of the multiple fused feature images of different scales into a target prediction model to obtain a target layout map output by the target prediction model;

[0377] The target prediction model is trained based on at least one of a classification loss function, a confidence loss function, and an intersection-over-union loss function.

[0378] The image generation device of the photovoltaic system in the embodiments of the present application can be a photovoltaic system, or can be an electronic device that is communicatively connected to the photovoltaic system, or can be a component of the photovoltaic system or electronic device, such as an integrated circuit or chip. The electronic device can be a terminal or other device other than a terminal. For example, the electronic device can be a mobile phone, tablet computer, laptop computer, PDA, in-vehicle electronic device, mobile internet device (MID), augmented reality (AR) / virtual reality (VR) device, robot, wearable device, ultra-mobile personal computer (UMPC), netbook, or personal digital assistant (PDA), etc. It can also be a server, network attached storage (NAS), personal computer (PC), television (TV), ATM, or self-service machine, etc., and the embodiments of the present application are not specifically limited.

[0379] The photovoltaic system image generation device in the embodiment of the present application may be a device having an operating system. The operating system may be an Android operating system, an iOS operating system, or other possible operating systems, which are not specifically limited in the embodiment of the present application.

[0380] The image generation device of the photovoltaic system provided in the embodiment of the present application can achieve Figures 1 to 16 To avoid repetition, the various processes implemented in the method embodiment are not described here.

[0381] An embodiment of the present application also provides a photovoltaic system.

[0382] In some embodiments, the photovoltaic system may include: a plurality of photovoltaic modules and a plurality of inverters.

[0383] In this embodiment, each inverter is electrically connected to at least one photovoltaic module.

[0384] For example, an inverter is connected to a photovoltaic panel.

[0385] For another example, an inverter is connected to two or more photovoltaic modules, that is, the inverter is a one-to-many inverter.

[0386] The photovoltaic system generates a layout diagram based on the photovoltaic system image generation method described in any one of the above embodiments.

[0387] According to the photovoltaic system provided in the embodiment of the present application, multi-layer feature extraction and feature fusion are performed on the first image that intuitively reflects the actual setting of the photovoltaic system and the connection status of the inverter and the photovoltaic module through the feature code image through the target fusion model, so as to obtain multiple fused feature images of different scales, obtain the rich features and details included in the first image, and automatically generate the target layout diagram based on the fused feature image without manual operation, thereby reducing the time to obtain the target layout diagram, improving the efficiency of generating the target layout diagram, reducing the delivery time, and improving the user experience.

[0388] In some embodiments, as Figure 18 As shown, an embodiment of the present application further provides an electronic device 1800, including a processor 1801, a memory 1802, and a computer program stored in the memory 1802 and executable on the processor 1801. When the program is executed by the processor 1801, each process of the above-mentioned photovoltaic system image generation method embodiment is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be described here.

[0389] It should be noted that the electronic devices in the embodiments of the present application include the mobile electronic devices and non-mobile electronic devices mentioned above.

[0390] An embodiment of the present application further provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the various processes of the above-mentioned photovoltaic system image generation method embodiment are implemented, and the same technical effects can be achieved. To avoid repetition, they are not further described here.

[0391] The processor is the processor in the electronic device described in the above embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0392] An embodiment of the present application further provides a computer program product, including a computer program, which implements the above-mentioned method for generating an image of a photovoltaic system when executed by a processor.

[0393] The processor is the processor in the electronic device described in the above embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0394] An embodiment of the present application further provides a chip, which includes a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run programs or instructions to implement the various processes of the above-mentioned photovoltaic system image generation method embodiment, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0395] It should be understood that the chip mentioned in the embodiments of the present application can also be called a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.

[0396] It should be noted that, in this article, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the statement "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element. In addition, it should be noted that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in the opposite order according to the functions involved. For example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0397] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, or the part that contributes to the relevant technology, can be embodied in the form of a computer software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in each embodiment of this application.

[0398] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of this application, ordinary technicians in this field can also make many forms without departing from the purpose of this application and the scope of protection of the claims, all of which are within the protection of this application.

[0399] Throughout this specification, reference to terms such as "one embodiment," "some embodiments," "illustrative embodiments," "examples," "specific examples," or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0400] Although the embodiments of the present application have been shown and described, those skilled in the art will appreciate that various changes, modifications, substitutions, and variations may be made to the embodiments without departing from the principles and intent of the present application, and that the scope of the present application is defined by the claims and their equivalents.

Claims

1. A method for generating an image of a photovoltaic system, characterized in that: The photovoltaic system includes a plurality of inverters, each of which is electrically connected to at least one photovoltaic module; the method includes: Acquire a first image corresponding to the photovoltaic system; the first image includes a feature code image corresponding to each of the inverters, and the relative position information of pixels of each of the feature code images in the first image matches the actual relative position information of the multiple inverters; Inputting the first image into a target fusion model, and performing multi-layer feature extraction and feature fusion processing on the first image by the target fusion model to obtain a plurality of fused feature images of different scales; generating a target layout map based on at least some of the fused feature images in the plurality of fused feature images at different scales; The generating of the target layout map based on at least some of the fused feature images of the multiple fused feature images of different scales includes: performing feature recognition on at least a portion of the fused feature images of the multiple fused feature images of different scales to obtain pixel position information of each feature code image in the at least a portion of the fused feature image and identification information corresponding to each feature code image; the identification information is obtained by recognizing the feature code image and is used to characterize identification information of the inverter and identification information of the photovoltaic module electrically connected to the inverter; generating the target layout map based on the pixel position information and the identification information; The performing feature recognition on at least a portion of the fused feature images in the plurality of fused feature images of different scales to obtain pixel position information of each feature code image in the at least a portion of the fused feature images includes: Calculating a rotation angle corresponding to each of the feature code images based on at least a portion of the positioning points corresponding to each of the feature code images in the first fused feature image in the at least partially fused feature image and an affine transformation matrix; Based on each of the rotation angles, the pixel position information of each of the feature code images in the first fused feature image is updated to obtain new pixel position information.

2. The method for generating an image of a photovoltaic system according to claim 1, wherein: The first image is input into a target fusion model, and the target fusion model performs multi-layer feature extraction and feature fusion processing on the first image to obtain a plurality of fused feature images of different scales, including: The first image is input into the target fusion model, and the target fusion model performs feature extraction and feature fusion processing of at least four modules on the first image to obtain at least four fused feature images of different scales.

3. The method for generating an image of a photovoltaic system according to claim 2, wherein: The target fusion model includes: a first feature processing module, a second feature processing module, a third feature processing module and a fourth feature processing module; The first feature processing module includes: a first submodule, a second submodule, and a third submodule; the second feature processing module includes: a fourth submodule, a fifth submodule, and a sixth submodule; the third feature processing module includes: a seventh submodule, an eighth submodule, and a ninth submodule; the fourth feature processing module includes: a tenth submodule, an eleventh submodule, and a twelfth submodule; The first input port corresponding to the first submodule is used to input the first image; The first sub-input port corresponding to the second sub-module is connected to the first sub-output port corresponding to the first sub-module; The third input port corresponding to the third submodule is connected to the first suboutput port corresponding to the first submodule and the second output port corresponding to the second submodule respectively; The fourth input port corresponding to the fourth submodule is connected to the second suboutput port corresponding to the first submodule; The first sub-input port corresponding to the fifth sub-module is connected to the first sub-output port corresponding to the fourth sub-module, and the second sub-output port corresponding to the fifth sub-module is connected to the second sub-input port corresponding to the second sub-module; The first sub-input port corresponding to the sixth sub-module is connected to the first sub-output port corresponding to the fifth sub-module and the first sub-output port corresponding to the fourth sub-module respectively; the second sub-input port corresponding to the sixth sub-module is connected to the second sub-output port corresponding to the third sub-module; The seventh input port corresponding to the seventh submodule is connected to the second sub-output port corresponding to the fourth submodule; The first sub-input port corresponding to the eighth sub-module is connected to the first sub-output port corresponding to the seventh sub-module, and the second sub-output port corresponding to the eighth sub-module is connected to the second sub-input port corresponding to the fifth sub-module; The first sub-input port corresponding to the ninth sub-module is connected to the first sub-output port corresponding to the eighth sub-module and the first sub-output port corresponding to the seventh sub-module, and the second sub-input port corresponding to the ninth sub-module is connected to the second sub-output port corresponding to the sixth sub-module; The tenth input port corresponding to the tenth submodule is connected to the second sub-output port corresponding to the seventh submodule; The eleventh input port corresponding to the eleventh submodule is connected to the tenth output port corresponding to the tenth submodule, and the second sub-output port corresponding to the eleventh submodule is connected to the second sub-input port corresponding to the eighth submodule; The first sub-input port corresponding to the twelfth sub-module is connected to the first sub-output port corresponding to the eleventh sub-module and the tenth output port corresponding to the tenth sub-module, and the second sub-input port corresponding to the twelfth sub-module is connected to the second sub-output port corresponding to the ninth sub-module; The first sub-output port corresponding to the third sub-module, the first sub-output port corresponding to the sixth sub-module, the first sub-output port corresponding to the ninth sub-module, and the twelfth output port corresponding to the twelfth sub-module are respectively used to output the fused feature images of different scales.

4. The method for generating an image of a photovoltaic system according to claim 3, wherein: The feature fusion is performed based on the following steps: The target feature image and the adjacent feature image input to the target feature processing module in the target fusion model are processed using the model parameters and the first constant corresponding to the target feature processing module to obtain a target fusion image; the target feature image is output by the target feature processing module, and the adjacent feature image is output by a module adjacent to the target feature processing module; the model parameters are obtained by training based on multiple sample feature code images.

5. The method for generating an image of a photovoltaic system according to any one of claims 1 to 4, characterized in that: After generating the target layout map based on at least some of the fused feature images of the multiple fused feature images at different scales, the method further includes: respectively obtaining electrical information collected by each of the inverters; The electrical information is added to the target layout diagram according to pixel position information corresponding to each feature code image in the target layout diagram.

6. The method for generating an image of a photovoltaic system according to any one of claims 1 to 4, characterized in that: The generating of the target layout map based on at least some of the fused feature images of the multiple fused feature images of different scales further includes: Inputting at least part of the fused feature images of the plurality of fused feature images of different scales into a target prediction model to obtain the target layout map output by the target prediction model; The target prediction model is trained based on at least one of a classification loss function, a confidence loss function, and an intersection-over-union loss function.

7. An image generation device for a photovoltaic system, characterized in that: The photovoltaic system includes a plurality of inverters, each of which is electrically connected to at least one photovoltaic module; the device includes: a first processing module configured to obtain a first image corresponding to the photovoltaic system; the first image comprising a feature code image corresponding to each of the inverters, wherein pixel relative position information of each of the feature code images in the first image matches actual relative position information of the plurality of inverters; a second processing module, configured to input the first image into a target fusion model, and have the target fusion model perform multi-layer feature extraction and feature fusion processing on the first image to obtain a plurality of fused feature images of different scales; a third processing module, configured to generate a target layout map based on at least some of the fused feature images of the plurality of fused feature images of different scales; The third processing module is further configured to: performing feature recognition on at least a portion of the fused feature images of the multiple fused feature images of different scales to obtain pixel position information of each feature code image in the at least a portion of the fused feature image and identification information corresponding to each feature code image; the identification information is obtained by recognizing the feature code image and is used to characterize identification information of the inverter and identification information of the photovoltaic module electrically connected to the inverter; generating the target layout map based on the pixel position information and the identification information; The third processing module is further configured to: Calculating a rotation angle corresponding to each of the feature code images based on at least a portion of the positioning points corresponding to each of the feature code images in the first fused feature image in the at least partially fused feature image and an affine transformation matrix; Based on each of the rotation angles, the pixel position information of each of the feature code images in the first fused feature image is updated to obtain new pixel position information.

8. A photovoltaic system, characterized in that: include: Multiple photovoltaic panels; a plurality of inverters, each of the inverters being electrically connected to at least one of the photovoltaic modules; The photovoltaic system generates a layout diagram based on the photovoltaic system image generation method according to any one of claims 1 to 6.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the image generation method for the photovoltaic system according to any one of claims 1 to 6 is implemented.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for generating an image of a photovoltaic system according to any one of claims 1 to 6 is implemented.

11. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the method for generating an image of a photovoltaic system according to any one of claims 1 to 6 is implemented.

Images