Photovoltaic recommendation method, device and storage medium based on photovoltaic identification and clustering

Abstract

The present application provides a photovoltaic recommendation method, device, and storage medium based on photovoltaic identification and clustering, which relates to the field of new energy planning technology, including: obtaining target building parameters and location information corresponding to each building in the target area; for each building, inputting the target building parameters and location information corresponding to the building as dependent variables into the GWR model to obtain the photovoltaic installation area of ​​the building, and the GWR model is used to reflect the correspondence between the target building parameters and location information and the photovoltaic installation area of ​​the building; and determining the target buildings recommended for photovoltaic installation in the target area based on the photovoltaic installation area of ​​each building in the target area. The present application predicts the photovoltaic installation area of ​​a building based on the target building parameters and location information of the building, and predicts the photovoltaic installation potential of the building using the target building parameters as regression coefficients. By using model analysis to replace the manual recommendation process, the recommendation results are more objective and reasonable, significantly improving the accuracy and intelligence of the recommendation.

Description

Technical Field

[0001] The present application relates to the technical field of new energy planning, and in particular to a photovoltaic recommendation method, device and storage medium based on photovoltaic identification and clustering. Background Art

[0002] Distributed photovoltaic power generation refers to photovoltaic power generation facilities built near user sites, operating in a manner characterized by self-consumption by users, with excess electricity being connected to the grid, and balanced by the distribution system. In cities, distributed photovoltaic power generation is typically installed on rooftops. Therefore, the scale (installed capacity) of urban distributed photovoltaic power generation is considered to be primarily determined by the roof area of ​​urban buildings.

[0003] In related technologies, personnel use visual maps to identify buildings without distributed photovoltaics installed. Based on the building's roof area, they manually determine whether the building is suitable for distributed photovoltaics and make recommendations for users who are suitable. This method suffers from insufficient accuracy and low recommendation efficiency. Summary of the Invention

[0004] The present application provides a photovoltaic recommendation method, device and storage medium based on photovoltaic identification and clustering, which are used to solve the problems of inaccurate and low efficiency of manual recommendation.

[0005] In the first aspect, the present application provides a photovoltaic recommendation method based on photovoltaic identification and clustering, including: obtaining target building parameters and location information corresponding to each building in the target area, where the target building parameters are building parameters that affect photovoltaic installation; for each building, inputting the target building parameters and location information corresponding to the building as dependent variables into a geographically weighted regression (GWR) model to obtain the photovoltaic installation area of ​​the building, wherein the GWR model is used to reflect the correspondence between the target building parameters and location information and the photovoltaic installation area of ​​the building; and determining the target buildings recommended for photovoltaic installation in the target area based on the photovoltaic installation area of ​​each building in the target area.

[0006] Optionally, the target building parameters and location information corresponding to the building are input into the GWR model as dependent variables to obtain the photovoltaic installation area of ​​the building, including: determining the photovoltaic installation area of ​​the building according to the following formula:

[0007]

[0008] Among them, y i represents the photovoltaic installation area of ​​the i-th building, (u i , v i ) represents the location information of the i-th building, β0(u i , v i ) represents the intercept term of the i-th building, β k (u i, v i ) represents the kth regression coefficient of the i-th building, x ij represents the jth target building parameter of the i-th building, the value range of j is [1, k], k is the total number of parameter types of the target building parameter, ε i represents the regression residual.

[0009] Optionally, target buildings recommended for photovoltaic installation in the target area are determined based on the photovoltaic installation area of ​​each building in the target area, including: obtaining the building density, building height and value attribute information of the building; determining the building density level, building height level and value attribute level of the building based on the building density, building height and value attribute information respectively; combining different levels of building density, building height and value attribute information to form a clustering matrix of buildings in the target area; and determining the buildings corresponding to the combinations whose average photovoltaic installation area meets the requirements in the clustering matrix as target buildings recommended for photovoltaic installation in the target area.

[0010] Optionally, before obtaining the target building parameters and location information corresponding to each building in the target area, it also includes: obtaining the location information and photovoltaic installation area of ​​sample buildings that have installed photovoltaics in the specified area; obtaining the building parameters of the sample buildings, the types of building parameters include building density, building height, building outline area, building area, building type and value attribute information; inputting the combination of different types of building parameters into the GWR model as independent variables one by one, and taking the photovoltaic installation area as the dependent variable for fitting processing; calculating the goodness of fit R-square value corresponding to the combination of different types of building parameters based on the fitting processing results; and determining the type of building parameters contained in the combination of building parameters with the largest R-square value as the parameter type of the target building parameters.

[0011] Optionally, the location information and photovoltaic installation area of ​​sample buildings with installed photovoltaics in the specified area are obtained, including: based on the map satellite image of the specified area, using the deep high-resolution representation learning HRNet algorithm model to identify the photovoltaic equipment installed on the sample buildings; marking the identified installed photovoltaic equipment to obtain a marked area; determining the marked area as the photovoltaic installation area, and determining the location information of the building corresponding to the marked area as the location information of the building with installed photovoltaics.

[0012] Optionally, after marking the identified installed photovoltaic equipment and obtaining the marked area, the method further includes: comparing the marked area with building outline area data of the corresponding location;

[0013] The marked area that is not within the building outline area is eliminated to obtain a new marked area; correspondingly, the marked area is determined as the photovoltaic installation area, and the location information of the building corresponding to the marked area is determined as the location information of the building where photovoltaics have been installed, including: determining the new marked area as the photovoltaic installation area, and determining the location information of the building corresponding to the new marked area as the location information of the building where photovoltaics have been installed.

[0014] Optionally, the HRNet algorithm model is trained through the following process: obtaining a map satellite image of the sample area and marking information contained in the map satellite image, where the marking information marks the area of ​​photovoltaic equipment installed on the buildings in the sample area; using the map satellite image and the marking information to train the HRNet algorithm model to obtain a trained HRNet algorithm model.

[0015] In the second aspect, the present application provides a photovoltaic recommendation device based on photovoltaic identification and clustering, including: an acquisition module for acquiring target building parameters and location information corresponding to each building in the target area, where the target building parameters are building parameters that affect photovoltaic installation; a first determination module for inputting the target building parameters and location information corresponding to each building as dependent variables into a geographically weighted regression (GWR) model to obtain the photovoltaic installation area of ​​the building, wherein the GWR model is used to reflect the correspondence between the target building parameters and location information and the photovoltaic installation area of ​​the building; a second determination module for determining the target buildings recommended for photovoltaic installation in the target area based on the photovoltaic installation area of ​​each building in the target area.

[0016] In a third aspect, the present application provides an electronic device comprising: a memory, a processor; the memory is used to store program instructions; the processor is used to call the program instructions to execute any one of the photovoltaic recommendation methods provided in the first aspect above.

[0017] In a fourth aspect, the present application provides a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement any photovoltaic recommendation method provided in the first aspect.

[0018] In a fifth aspect, the present application provides a computer program product, comprising a computer program; when the computer program is executed, the photovoltaic recommendation method provided in the first aspect above is implemented.

[0019] The photovoltaic recommendation method, device, and storage medium based on photovoltaic identification and clustering provided in this application include: obtaining target building parameters and location information corresponding to each building in the target area, where the target building parameters are building parameters that affect photovoltaic installation; for each building, inputting the target building parameters and location information corresponding to the building as dependent variables into the GWR model to obtain the photovoltaic installation area of ​​the building, wherein the GWR model is used to reflect the correspondence between the target building parameters and location information and the photovoltaic installation area of ​​the building; and determining the target buildings recommended for photovoltaic installation in the target area based on the photovoltaic installation area of ​​each building in the target area. This application adopts the GWR model to predict the photovoltaic installation area of ​​the building based on the target building parameters and location information of the building, and uses the target building parameters as regression coefficients to predict the photovoltaic installation potential of the building. By using model analysis to replace the manual recommendation process, the recommendation results are more objective and reasonable, significantly improving the accuracy and intelligence of the recommendation. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0021] Figure 1 Schematic diagram of application scenarios provided by embodiments of the present application;

[0022] Figure 2 Schematic diagram of the process of photovoltaic recommendation method based on photovoltaic identification and clustering provided in the embodiment of this application Figure 1 ;

[0023] Figure 3 A schematic diagram of the potential distribution of rooftop distributed photovoltaic installations provided in an embodiment of the present application;

[0024] Figure 4 A schematic diagram of building density levels provided for an embodiment of this application;

[0025] Figure 5 A schematic diagram of building height levels provided in an embodiment of the present application;

[0026] Figure 6 A schematic diagram of the building value attribute levels provided in the embodiment of this application;

[0027] Figure 7 A schematic diagram of the cluster analysis results provided in the embodiment of the present application;

[0028] Figure 8 Schematic diagram of the process of photovoltaic recommendation method based on photovoltaic identification and clustering provided in the embodiment of this application Figure 2 ;

[0029] Figure 9Schematic diagram of the network structure of the HRNet algorithm model provided in the embodiment of this application;

[0030] Figure 10 Schematic diagram of the process of photovoltaic recommendation method based on photovoltaic identification and clustering provided in the embodiment of this application Figure 3 ;

[0031] Figure 11 A schematic diagram of the structure of a photovoltaic recommendation device based on photovoltaic identification and clustering provided in an embodiment of the present application;

[0032] Figure 12 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application.

[0033] The above drawings illustrate specific embodiments of the present application, which will be described in more detail below. These drawings and the textual description are not intended to limit the scope of the present application in any way, but rather to illustrate the concepts of the present application to those skilled in the art by reference to specific embodiments. DETAILED DESCRIPTION

[0034] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.

[0035] Figure 1 This is a schematic diagram of an application scenario provided by an embodiment of the present application. Figure 1 As shown, the application scenario involves a building 110 that has photovoltaics installed and a target building 120 that is recommended to install photovoltaics.

[0036] The photovoltaic building 110 includes a building 111 and a photovoltaic device 112. The building 111 can be a residential house, a residential building, an office building, a factory platform or a teaching building, etc., and has a platform or position on the top for placing the photovoltaic device 112.

[0037] In cities, photovoltaic devices 112 mainly include rooftop photovoltaic devices. Photovoltaic devices 112 generally include solar panels and other components that convert solar energy into electrical energy and are installed on open, less obstructed rooftop locations.

[0038] This application aims to analyze the primary factors influencing photovoltaic installation in buildings 110 already equipped with photovoltaics. Based on these primary factors, the potential / possibility of photovoltaic installation in buildings not already installed is determined, thereby identifying buildings with higher installation potential as target buildings 120 for recommended photovoltaic installation. The photovoltaic recommendation method proposed in this application will be explained in detail below, using specific examples.

[0039] Figure 2 Schematic diagram of the process of photovoltaic recommendation method based on photovoltaic identification and clustering provided in the embodiment of this application Figure 1 .like Figure 2 As shown, the recommended photovoltaic method includes:

[0040] S201: Obtain target building parameters and location information corresponding to each building in the target area, where the target building parameters are building parameters that affect photovoltaic installation.

[0041] Building parameters are parameters related to building properties, and may include building density, building height, building area, and building age. These factors affect the installed capacity of photovoltaic installations to varying degrees.

[0042] For example, higher building density increases the impact of shadows cast by surrounding buildings, reducing the available area for distributed rooftop PV. On the other hand, distributed PV adoption has a peer effect: higher building density means greater influence from rooftop PV adoption by friends and neighbors, making it more conducive to distributed PV deployment. Therefore, the layout of distributed rooftop PV is subject to both the negative impact of high density and the positive impact of high density and peer effects.

[0043] Higher building heights increase the difficulty of constructing rooftop distributed photovoltaics, negatively impacting their adoption. However, higher building heights also reduce the shadowing effects of surrounding buildings on the rooftop. This suggests that building height may have a dual impact on the layout of rooftop distributed photovoltaics.

[0044] Target building parameters refer to those with a significant impact on PV installations. While the types of target building parameters are generally consistent over the same period or region, they may differ across different regions or timeframes. For example, determining which building parameters are target building parameters involves using computational models to determine the impact of each building parameter. Target building parameters are then identified as those with an impact exceeding a threshold or a set number of building parameters with the highest impact.

[0045] The location information may include the location coordinates of the building and the city where it is located, etc. In actual application, the location information of the building can be obtained through map software (such as Baidu Maps, Google Maps or Amap, etc.) or the Global Positioning System (GPS).

[0046] S202: For each building, the target building parameters and location information corresponding to the building are input as dependent variables into a Geographically Weighted Regression (GWR) model to obtain the building's photovoltaic installation area. The GWR model is used to reflect the correspondence between the target building parameters and location information and the building's photovoltaic installation area.

[0047] The GWR model is an improved spatial linear regression method. In global regression models such as Ordinary Least Squares (OLS), when two or more variables exhibit multicollinearity, the results are often unreliable. Geographically Weighted Regression (GWR) overcomes this problem through local regression. In this embodiment, a GWR model is constructed using target building parameters and location information as independent variables and PV installation area as the dependent variable. The GWR model is then used to fit and calculate the PV installation area of ​​each building.

[0048] In some embodiments, specifically, inputting target building parameters and location information corresponding to a building as dependent variables into a GWR model to obtain the photovoltaic installation area of ​​the building includes: determining the photovoltaic installation area of ​​the building according to the following formula:

[0049]

[0050] Among them, y i represents the photovoltaic installation area of ​​the i-th building, (u i , v i ) represents the location information of the i-th building, β0(u i , v i ) represents the intercept term of the i-th building, β k (u i , v i ) represents the kth regression coefficient of the i-th building, x ij represents the jth target building parameter of the i-th building, the value range of j is [1, k], k is the total number of parameter types of the target building parameter, ε i represents the regression residual.

[0051] According to Formula 1, the location information and target building parameters of each building in the target area are sequentially entered and accumulated to obtain the photovoltaic installation area of ​​each building. In the embodiment of the present application, rooftop distributed photovoltaic is used as an example for explanation.

[0052] S203: Determine target buildings recommended for photovoltaic installation in the target area based on the photovoltaic installation area of ​​each building in the target area.

[0053] Alternatively, buildings with a PV installation area greater than a set threshold can be identified as target buildings for PV installation recommendations. Alternatively, the resulting PV installation areas can be ranked, with buildings ranked highest in PV installation area as recommended target buildings. During the recommendation process, the number or range of target buildings can be selected based on actual needs.

[0054] In an embodiment of the present application, target building parameters and location information corresponding to each building in the target area are obtained. The target building parameters are building parameters that affect photovoltaic installation. For each building, the target building parameters and location information corresponding to the building are input into the GWR model as dependent variables to obtain the photovoltaic installation area of ​​the building. The GWR model is used to reflect the correspondence between the target building parameters and location information and the photovoltaic installation area of ​​the building. Based on the photovoltaic installation area of ​​each building in the target area, the target building recommended for photovoltaic installation in the target area is determined. The embodiment of the present application adopts the GWR model to predict the photovoltaic installation area of ​​the building based on the target building parameters and location information of the building, and uses the target building parameters as regression coefficients to predict the photovoltaic installation potential of the building. By using model analysis to replace the manual recommendation process, the recommendation results are more objective and reasonable, significantly improving the accuracy and intelligence of the recommendation.

[0055] Based on the above embodiment, target buildings recommended for photovoltaic installation in the target area are determined according to the photovoltaic installation area of ​​each building in the target area. The photovoltaic installation area can be quantile clustered to obtain the distribution of photovoltaic installation potential. Figure 3 This is a schematic diagram of the potential distribution of rooftop distributed photovoltaic installations in an embodiment of the present application. A circle in the figure indicates high potential, representing a high photovoltaic installation area in the area and a high potential for building-installed photovoltaics within the area; a triangle indicates medium potential, representing a medium photovoltaic installation area and a medium potential for building-installed photovoltaics within the area; and a square indicates low potential, representing a low photovoltaic installation area and a low potential for building-installed photovoltaics within the area. The potential for building-installed photovoltaics may vary within different target areas or over different timescales.

[0056] Optionally, target buildings recommended for photovoltaic installation in the target area are determined based on the photovoltaic installation area of ​​each building in the target area, including: obtaining the building density, building height and value attribute information of the building; determining the building density level, building height level and value attribute level of the building based on the building density, building height and value attribute information respectively; combining different levels of building density, building height and value attributes to form a clustering matrix of buildings in the target area; and determining the buildings corresponding to the combinations whose average photovoltaic installation area meets the requirements in the clustering matrix as target buildings recommended for photovoltaic installation in the target area.

[0057] This embodiment of the present application provides another specific method for determining target buildings. The construction of rooftop distributed photovoltaics is positively correlated with the amount of solar radiation received by the roof. Solar radiation is related to solar resources on the one hand, and negatively correlated with the degree to which the current building is blocked by other buildings on the other. Building density and building height can be used to jointly measure this indicator. These data can be directly obtained from the "Spatial Distribution Data of Building Bottom Outlines and Building Heights in Major Chinese Cities" published by the Institute of Geographic Sciences and Natural Resources, Chinese Academy of Sciences.

[0058] Optionally, you can define building area as the product of building height and building outline area, and introduce the kernel density parameter to calculate building density. The building height and building outline area of ​​each building in the target area can be obtained from the "Spatial Distribution Data of Building Base Outlines and Building Heights in Major Chinese Cities" published by the Institute of Geographic Sciences and Natural Resources, Chinese Academy of Sciences.

[0059] Optionally, the embodiment of the present application uses the kernel density analysis tool in the user desktop component ArcMap to determine the kernel density of each building in the target area. The search radius of the kernel density can be determined according to the following formula:

[0060]

[0061] Among them, SearchRadius represents the search radius, D m Represents the (weighted) median distance to the (weighted) mean center, where n is the number of points if the population field is not used. If the population field is provided, n is the sum of the population field values, and SD is the standard distance.

[0062] The building density level of the building is determined according to the building density. Specifically, after the kernel density calculation is completed, the numpy quantile method in the computer programming language Python is used to perform building kernel density grading. Figure 4 The schematic diagram of building density level provided in the embodiment of this application is as follows: Figure 4As shown, the density levels are divided into three levels: high density, medium density and low density. It means high density, which means the building density in this area is high; It means medium density, which means the building density in this area is medium; Indicates low density, indicating that the building density in the area is low. Building density and building density levels may vary in different target areas or over different time dimensions.

[0063] The building height classification is determined based on building height. The national standard "Residential Design Code" (GB 50096-2011) issued by the Ministry of Housing and Urban-Rural Development of the People's Republic of China can be used as a reference to classify building heights into four categories: 1-3, 4-6, 7-10, and 11 or more stories. One- to three-story buildings are defined as low-rise buildings, with a height range of 0-9 meters; four- to six-story buildings are defined as medium-rise buildings, with a height range of 9-18 meters; seven- to nine-story buildings are defined as mid- to high-rise buildings, with a height range of 18-27 meters; and buildings with ten stories or more are defined as high-rise buildings, with a height range of 27 meters or more.

[0064] Figure 5 The schematic diagram of building height levels provided in the embodiment of this application is as follows: Figure 5 As shown, number 1 represents low-rise, which means the building in the circle where number 1 is located is low-rise; number 2 represents middle-rise, which means the building in the circle where number 2 is located is middle-rise; number 3 represents middle-high-rise, which means the building in the circle where number 3 is located is middle-high-rise; number 4 represents high-rise, which means the building in the circle where number 4 is located is high-rise. Figure 5 It can be seen that high-rise buildings are mainly concentrated in the regional center, and the building height gradually decreases from the regional center to the periphery, but some high-rise, mid-high-rise or mid-rise buildings will also appear in the outer circle.

[0065] Building value attribute information is related to building housing prices. To obtain value attribute information for each building within a target area, housing prices can be obtained from "Digital City News" or other summary articles, magazines, and newspapers in that area to determine the building's value attribute information. For buildings with missing housing price data, kriging is used to interpolate housing price data. This allows the housing price at the missing data point to be inferred from the surrounding housing prices, thereby assigning value attribute information to all buildings.

[0066] Among them, Kriging is an interpolation method that considers the statistical relationship between measurement points. Compared with interpolation methods such as spline function method and natural neighbor method, it considers the spatial correlation of the described objects, making the interpolation results more scientific and closer to the actual situation. Specifically, Kriging interpolation can be implemented according to the following formula:

[0067]

[0068] in, represents the predicted value of the building at the predicted location (i.e., predicted housing price), Z(s i ) represents the measurement value at the i-th position, λ i represents the weight of the measurement value at the i-th position, s0 represents the predicted position, and N is the number of measurements. For example, this embodiment of the application uses Arcmap 10.4 software to perform kriging interpolation on the 2018 housing price data. Similar to the above-mentioned kernel density processing method, the housing price data is graded using Python's numpy quantile method to obtain three levels of value attribute information for the corresponding buildings. Figure 6 The schematic diagram of the building value attribute level provided in the embodiment of this application is as follows: Figure 6 As shown in the figure, ● represents high housing prices, indicating that the housing prices of buildings in this area are high and the building value attributes are high; ▲ represents medium housing prices, indicating that the housing prices of buildings in this area are in the middle and the building value attributes are in the middle; ■ represents low housing prices, indicating that the housing prices of buildings in this area are low and the building value attributes are low. The value attributes of buildings may vary in different target areas or over different time dimensions.

[0069] Different levels of building density, building height and value attributes are combined separately to form a clustering matrix of buildings in the target area. Specifically, existing buildings are divided into several groups through cluster analysis to achieve the effect of similar building characteristics within the group but different characteristics between groups. First, the physical properties and value attributes of the buildings need to be preliminarily classified, and all possible classification situations are arranged and combined according to the classification results to form a clustering matrix.

[0070] For example, in an embodiment of the present application, the building samples in the target area are arranged and combined according to three types of building density (low density A, medium density B, high density C), four types of building height (low-rise 1, medium-rise 2, medium-high-rise 3, high-rise 4) and three types of housing prices (low housing price a, medium housing price b, high housing price c), and a total of 36 clustering types are obtained, as shown in Table 1 below. The rooftop distributed photovoltaic area is the value of the matrix.

[0071] Table 1 Clustering matrix of buildings adopting rooftop distributed photovoltaics in the target area

[0072]

[0073] The distributed photovoltaic area (installed capacity) on the roof of each building is classified according to the above clustering rules. Figure 7 The schematic diagram of the cluster analysis results provided in the embodiment of this application is as follows: Figure 7As shown in the figure, the horizontal axis represents the different cluster combinations of building density, height and value attributes, and the vertical axis is the average photovoltaic installation area of ​​rooftop distributed photovoltaic under this combination. Figure 7 Analysis shows that:

[0074] (1) The distribution of rooftop distributed photovoltaics in the central old city (high building density, high housing prices) is weaker than that in the secondary new areas (mixed high and medium building density, medium housing prices). When the floor heights 1-4 are fixed and the area of ​​rooftop distributed photovoltaics corresponding to AC and ac is observed, it can be found that the heights of C1c, C2c, C3c, and C4c are generally lower than those of B1b, B2b, B3b, B4b, C1b, C2b, C3b, and C4b. The former represents high building density and high housing prices at all building heights, while the latter refers to high-density and medium-density areas with medium housing prices at all building heights. The possible reason is that the houses in the central old city are older and more old, while the secondary new areas such as Fengtai District and Shijingshan District were developed later, the house age is relatively low, and the remaining utilization years are long. Generally, the life cycle of rooftop distributed photovoltaics is more than 25 years. When the expected building age is insufficient, people tend not to consider installing distributed photovoltaics. Therefore, the rooftop distributed photovoltaics in the central old city are less than those in the secondary new areas.

[0075] (2) As building density increases, the negative impact of building height on photovoltaic installations becomes increasingly apparent. Figure 7 In the figure, the building density represented by the first capital letter and the housing price represented by the third lowercase letter are fixed, and the vertical axis photovoltaic area changes with the 1-4 representing the floor height. For example, C4c is lower than B4c and lower than A4c, and the area difference is large, while the difference between C3c, B3c and A3c is slightly smaller, indicating that the higher the floor height, the faster the rooftop distributed photovoltaic area decreases with the increase of building density.

[0076] The possible reasons can be divided into two categories. On the one hand, as the building density increases, the shadow areas produced by surrounding buildings under sunlight are more, the mutual shading increases, the effective solar radiation decreases, and the photovoltaic installed capacity is less. When the building density is low, the shadows of surrounding buildings are less likely to fall on the building, and the photovoltaic installed capacity is larger at this time; on the other hand, the higher the building height, the greater the wind speed, and the higher the wind resistance and installation cost requirements for photovoltaic panels. Therefore, the higher the building height, the fewer photovoltaic installed capacity.

[0077] (3) In areas with low building density, the higher the value attribute, the larger the photovoltaic installed capacity; while in areas with high building density, the higher the value attribute, the smaller the photovoltaic installed capacity. Figure 7 It can be seen that in area A, no matter how the heights of floors 1-4 change, the rooftop distributed photovoltaic area increases from a to c; in area C, no matter how the heights of floors 1-4 change, the rooftop distributed photovoltaic area shows a decreasing trend from a to c.

[0078] A possible reason for this is that low-density areas are often located in suburbs, where per capita income is lower than in densely populated urban areas. Therefore, rooftop distributed PV projects in low-density areas have a greater investment value proposition. Investors with higher housing prices and better economic conditions are more likely to invest in PV projects, leading to larger installed PV capacity. In contrast, in densely populated urban areas, rooftops have greater non-economic benefits, such as recreational activities and green spaces. Investors with higher housing prices and better economic conditions may have a greater need for leisure time or face the significant time cost of installing PV. Consequently, in densely populated areas, the higher the value proposition, the smaller the installed PV capacity.

[0079] Figure 8 Schematic diagram of the process of photovoltaic recommendation method based on photovoltaic identification and clustering provided in the embodiment of this application Figure 2 , is executed before obtaining the target building parameters and location information corresponding to each building in the target area. Figure 8 As shown, the recommended photovoltaic method includes:

[0080] S801: Obtain location information and photovoltaic installation area of ​​sample buildings that have installed photovoltaics in a specified area.

[0081] The designated area refers to a sample region containing sample buildings with installed photovoltaic systems. Analysis of these sample buildings will be conducted to determine factors influencing photovoltaic installation. The location information of the sample buildings can be obtained using satellite maps and will not be further detailed here. The photovoltaic installation area of ​​the sample buildings can also be obtained using satellite maps.

[0082] In a specific embodiment, the location information and photovoltaic installation area of ​​sample buildings with photovoltaic installations in a specified area are obtained, including: based on a map satellite image of the specified area, using a Deep High-Resolution Representation Learning (HRNet) algorithm model to identify photovoltaic equipment installed on sample buildings; marking the identified installed photovoltaic equipment to obtain a marked area; determining the marked area as the photovoltaic installation area, and determining the location information of the building corresponding to the marked area as the location information of the building with photovoltaic installations.

[0083] In the embodiment of the present application, buildings with photovoltaics installed in a designated area are used as sample buildings, and the photovoltaics are identified using satellite map data and a trained HRNet algorithm model. The designated area can be exemplarily selected from Beijing, Jiangsu Province, or Chongqing Municipality. When selecting satellite maps, compared to satellite remote sensing image databases such as Baidu Satellite Maps and Amap Satellite Maps, Google Maps has the characteristics of fast data updates and more comprehensive regional coverage. In the embodiment of the present application, Google Satellite Maps data is selected as a method for obtaining map satellite images.

[0084] When using satellite map data, the lowest processing level that can clearly distinguish rooftop photovoltaics is preferred, because the clearer the image, the more storage space it takes up, and an image that is too large will make it impossible for ordinary computers to process it. On the other hand, there are precision constraints when shooting satellites. The higher the photo resolution, the more difficult it is for the satellite to collect image data. A processing level that is too low may cover more area, but it may not be able to clearly identify the location of the photovoltaics. In order to balance clarity, data size, and total data volume, the embodiment of this application adopts satellite map data with a resolution of 0.28 meters.

[0085] Based on the satellite map data, an HRNet algorithm model is trained. Optionally, the HRNet algorithm model is trained by obtaining a map satellite image of the sample area and labeling information contained in the map satellite image, where the labeling information identifies the area of ​​photovoltaic equipment installed on buildings in the sample area; and training the HRNet algorithm model using the map satellite image and the labeling information to obtain a trained HRNet algorithm model.

[0086] Manually screening satellite images showing photovoltaics from satellite map datasets offers the highest accuracy. However, as the number of satellite images increases, it becomes increasingly difficult to allocate sufficient manpower and time to complete photovoltaic image recognition. Therefore, machine learning technology is required for image recognition. The HRNet algorithm model is a novel and highly efficient image recognition algorithm.

[0087] Figure 9 Schematic diagram of the network structure of the HRNet algorithm model provided in the embodiment of this application. The HRNet algorithm model maintains high-resolution representation by connecting high-resolution to low-resolution convolutions in parallel, and enhances high-resolution representation by repeatedly performing multi-scale fusion across parallel convolutions, rather than synthesizing high-resolution information from low resolution as used by other neural network algorithms. Therefore, HRNet can maintain high-resolution representation throughout the entire process, so that spatial sensitivity information is maintained throughout the process, which gives it more advantages in satellite remote sensing imagery / map satellite image recognition.

[0088] Specifically, before training the HRNet algorithm model, the satellite images of the sample area are first sampled and marked. Before using the HRNet algorithm model to identify rooftop distributed photovoltaics, it is necessary to use a supervised learning method to classify the satellite map data into images with photovoltaics and without photovoltaics for training, so that the model can learn to distinguish rooftop distributed photovoltaics. The marking method can be to use the image recognition and marking program developed by the Institute of Automation of the Chinese Academy of Sciences to manually mark the rooftop photovoltaic images divided into tiles, and divide the marked images into three categories: "with photovoltaics", "without photovoltaics", and "uncertain". Optionally, the marking information can also mark the area of ​​​​the photovoltaic equipment installed on the buildings in the sample area.

[0089] These map satellite images and marking information are input into the HRNet algorithm model for training, resulting in a trained HRNet algorithm model. The trained HRNet algorithm model is then used to identify the photovoltaic equipment installed on the sample buildings. The identified installed photovoltaic equipment is marked to obtain a marked area. The marked area is determined as the photovoltaic installation area, and the location information of the building corresponding to the marked area is determined as the location information of the photovoltaic-installed building.

[0090] S802: Acquire building parameters of the sample building, where the types of building parameters include building density, building height, building outline area, building area, building type, and value attribute information.

[0091] Through S801, sample buildings with photovoltaic installations are obtained, along with the location information and photovoltaic installation area of ​​these sample buildings. Through S802, the building parameters of these sample buildings are further obtained. The types of building parameters include building density, building height, building outline area, building area, building type, and value attribute information. Specifically, building density can be determined based on the kernel density provided in the above embodiment, and building height and building outline area can be obtained from the "Spatial Distribution Data of Building Bottom Outlines and Building Heights in Major Cities in China" published by the Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences. The building area is the product of the building height and the building outline area. The building type is obtained by matching the building coordinates with Baidu Maps. Value attribute information is positively correlated with the building housing price. The building housing price can be obtained from "Digital City Affairs" or other summary articles, magazines, and newspapers in the sample area to determine the building's value attribute information. For buildings with missing housing price data, the housing price data can also be interpolated using the Kriging method, so that the housing price at the point without data can be inferred from the surrounding housing prices, thereby assigning value attribute information to all buildings.

[0092] S803: Input different types of building parameter combinations into the GWR model as independent variables one by one, and use the photovoltaic installation area as the dependent variable for fitting.

[0093] First, the building parameters proposed in S802 are randomly combined, and different types of building parameter combinations are input into the GWR model as independent variables one by one. Optionally, different types of building parameter combinations can be used as independent variables one by one, and the photovoltaic installation area can be used as the dependent variable, and input into Formula 1 for fitting processing.

[0094]

[0095] Among them, y i represents the photovoltaic installation area of ​​the i-th sample building, (u i , v i) represents the location information of the i-th sample building, β0(u i , v i ) represents the intercept term of the i-th sample building, β k (u i , v i ) represents the kth regression coefficient of the i-th sample building, x ij represents the jth target building parameter of the i-th sample building, the value range of j is [1, k], k is the total number of parameter types of the sample building parameters, ε i represents the regression residual.

[0096] In these embodiments, each building in the target area is regarded as a spatial point / data point, and the target building parameters of each building are regarded as labels / features of these spatial points / data points.

[0097] The regression coefficient is generally fitted using the weighted least squares method. Specifically, it can be determined according to the following formula:

[0098]

[0099] Among them, W(u i , v i ) represents the spatial weight diagonal matrix, X is the independent variable matrix, Y is the dependent variable vector, The regression coefficient coordinates are (u i , v i The spatial weight is calculated using the bi-square function. Specifically, it can be determined according to the following formula:

[0100]

[0101] Among them, ω ij is the weight when estimating the unknown spatial point i using the known spatial point j, d ij is the Euclidean distance between the unknown spatial point i and the known spatial point j, h represents the bandwidth, and the bandwidth is calculated using the minimum Akaike Information Criterion, corrected (AICc) method. Specifically, it can be determined according to the following formula:

[0102] AICc=-n ln(σ′)+n ln(2π)+n{[n+tr(S)] / [n-2-tr(S)]}, Formula 6

[0103] Where AICc represents the bandwidth h, n is the number of data points, σ′ is the error term, and tr(S) is the trace of the hat matrix S.

[0104] For example, the embodiment of the present application distinguishes the independent variables, takes the rooftop distributed photovoltaic area as the dependent variable, and proposes five regression models: Model 1, all influencing factors (building parameters) are included and input into the GWR model for geographically weighted regression analysis, that is, the independent variables are set as building density, building height, building outline area, building type, and value attribute information as independent variables; Model 2, building height, building outline area, building type, and value attribute information are used as independent variables; Model 3, building height and building outline area are combined, building area is used, and building density, building type, building area, and value attribute information are used as independent variables; Model 4, building type, building area, and value attribute information are used as independent variables; Model 5, building area and value attribute information are used as independent variables of the fifth model.

[0105] S804: Calculate the goodness of fit R-square value corresponding to the combination of different types of building parameters based on the fitting processing result.

[0106] R-squared refers to the goodness of fit, which is the degree of fit of the regression line to the observed values. It is used to evaluate the degree of consistency between the predicted value and the actual value. The larger the R-squared value, the stronger the model's ability to explain the rooftop distributed photovoltaic area. The R-squared value is defined as follows: the ratio of the regression sum of squares of the variation in y caused by variable x to the total sum of squares of the variation in y. The R-squared value can be calculated according to the following formula:

[0107]

[0108] Among them, R 2 That is R square, SSR is the regression sum of squares, SSE is the residual sum of squares, SST represents the total sum of squares, SST = SSR + SSE.

[0109] S805: Determine the type of the building parameters included in the combination of building parameters with the largest R-square value as the parameter type of the target building parameter.

[0110] In this embodiment of the present application, if the building parameter types included in the building parameter combination with the maximum R-squared value are building area and value attribute information, then building area and value attribute information are determined as the parameter types of the target building parameter. In other embodiments, the building parameter types included in the building parameter combination with the maximum R-squared value may be different.

[0111] On the basis of the above embodiment, optionally, the HRNet algorithm model is used to identify the photovoltaic equipment installed on the sample building, and the identified installed photovoltaic equipment is marked. After obtaining the marked area, it also includes: comparing the marked area with the building outline area data of the corresponding position; eliminating the marked area that is not in the building outline area to obtain a new marked area; correspondingly, determining the marked area as the photovoltaic installation area, and determining the location information of the building corresponding to the marked area as the location information of the building where photovoltaics have been installed, including: determining the new marked area as the photovoltaic installation area, and determining the location information of the building corresponding to the new marked area as the location information of the building where photovoltaics have been installed.

[0112] To further improve the accuracy of the marked areas, the marked areas obtained by the HRNet algorithm model were further screened to obtain more accurate locations of rooftop distributed photovoltaic panels. The longitude and latitude of each distributed photovoltaic panel identified by the HRNet algorithm model and its corresponding area data were combined with the corresponding building outline data. In other words, the area (installed capacity) of photovoltaic panels whose longitude and latitude fell on the same building was summed, and recorded as the total installed photovoltaic capacity of the building. At the same time, photovoltaic panels whose longitude and latitude were not within the building were excluded. This distributed photovoltaic identification error was considered and the data was discarded.

[0113] Figure 10 Schematic diagram of the process of photovoltaic recommendation method based on photovoltaic identification and clustering provided in the embodiment of this application Figure 3 .like Figure 10 As shown, the recommended photovoltaic method includes:

[0114] S1001: Based on the satellite image of the designated area, the HRNet algorithm model is used to identify the photovoltaic equipment installed on the sample building;

[0115] S1002: Marking the identified installed photovoltaic equipment to obtain a marked area;

[0116] S1003: Determine the marked area as the photovoltaic installation area, and determine the location information of the building corresponding to the marked area as the location information of the building where photovoltaics have been installed;

[0117] S1004: Obtaining location information and photovoltaic installation areas of sample buildings that have installed photovoltaics in a specified area;

[0118] S1005: Acquire building parameters of the sample building, where the building parameters include building density, building height, building outline area, building area, building type, and value attribute information;

[0119] S1006: Input different types of building parameter combinations into the GWR model as independent variables one by one, and use the photovoltaic installation area as the dependent variable for fitting;

[0120] S1007: Calculate R-squared values ​​corresponding to different types of building parameter combinations based on the fitting processing results;

[0121] S1008: Determine the type of the building parameters included in the combination of building parameters with the largest R-square value as the parameter type of the target building parameter;

[0122] S1009: Obtain target building parameters and location information corresponding to each building in the target area, where the target building parameters are building parameters that affect photovoltaic installation;

[0123] S1010: For each building, input the target building parameters and location information corresponding to the building as dependent variables into the GWR model to obtain the photovoltaic installation area of ​​the building, wherein the GWR model is used to reflect the corresponding relationship between the target building parameters and location information and the photovoltaic installation area of ​​the building;

[0124] S1011: Obtain building density, building height and value attribute information of the building;

[0125] S1012: Determine the building density level, building height level, and value attribute level of the building based on the building density, building height, and value attribute information, respectively;

[0126] S1013: combining building density, building height, and housing price data of different levels to form a clustering matrix of buildings in the target area;

[0127] S1014: Determine the buildings corresponding to the combinations whose average photovoltaic installation areas meet the requirements in the clustering matrix as target buildings recommended for photovoltaic installation in the target area.

[0128] The building with photovoltaics installed in S1003 is the sample building referred to in S1004, and what is obtained in S1004 is the sample building in the specified area.

[0129] In this embodiment, Beijing is selected as the designated area, and Google Satellite Map is used to collect map satellite images with a resolution of 0.28 meters in tile format during the period from September to October 2020. There are a total of 4,941,573 tile-format map satellite images of Beijing, including 500,316 images of Changping District, 185,328 images of Chaoyang District, 422,415 images of Daxing District, 12,100 images of Dongcheng District, 735,672 images of Fangshan District, 119,196 images of Fengtai District, 201,600 images of Haidian District, 584,168 images of Mentougou District, 1,045,656 images of Miyun District, 352,365 images of Pinggu District, 37,584 images of Shijingshan District, 334,719 images of Shunyi District, 397,458 images of Tongzhou District, and 12,996 images of Xicheng District, with a total size of 1.2TB.

[0130] The HRNet algorithm model is used to identify the photovoltaic equipment installed on the sample buildings contained in these map satellite images. First, some map satellite images are marked with photovoltaics and then input into the HRNet algorithm model for learning and training. A total of 19,421 map satellite images are marked in this embodiment of the application, of which 15,537 are used to train and fit the model, and 2,884 are used to verify the performance of the model. The current accuracy is 89.5%. Overall, the HRNet algorithm model is used for the first time to realize the identification of rooftop distributed photovoltaics with high resolution, and the accuracy can meet the standards of spatial analysis, achieving a high degree of accuracy.

[0131] Using the trained HRNet algorithm model to identify all map satellite images, this embodiment of the application identified a total of 15,412 photovoltaic panels with an installed capacity of approximately 348,700 kilowatts. Next, based on the coordinates of the upper left and lower right corners of the image, the resolution, and the total number of pixels included when downloading the data, the center latitude and longitude of each white area and the area were calculated using the number of pixels, ultimately forming a coordinate-area database for distributed photovoltaics.

[0132] Furthermore, the identified photovoltaic panels were combined with existing Beijing building outline data to accurately locate rooftop distributed photovoltaic panels. The longitude and latitude of each identified distributed photovoltaic panel and its corresponding area data were combined with Beijing building outline data. That is, the area (installed capacity) of photovoltaic panels whose longitude and latitude fell on the same building was summed up and recorded as the total installed photovoltaic capacity of the building. At the same time, photovoltaic panels whose longitude and latitude were not in the building were excluded. This was considered an error in the identification of distributed photovoltaic panels and the data was discarded. Finally, 1,654 buildings that adopted rooftop distributed photovoltaics were identified, namely the sample buildings, with an average photovoltaic installed capacity of 88 kilowatts per building.

[0133] For the sample buildings obtained, we performed S1004-S1007 and used ArcMap software to perform a geographically weighted regression fit. We used the five models from S803 and substituted them into Equation 7 to calculate the R-squared value. ArcMap is a visual geographic information system application that can be used to compile and edit geographic information data and has data processing capabilities. The specific results are shown in Table 2 below:

[0134] Table 2 R-square values ​​of five models

[0135]

[0136] Table 2 shows that Model 5 has the highest R-squared value of 0.648237, meaning that 64.82% of the 1,654 sample buildings are well explained, giving it the strongest explanatory power. Therefore, Model 5 was selected as the model for predicting the area of ​​distributed rooftop photovoltaics in Beijing. The building area and value attribute information included in Model 5 are the most influential factors and are determined as the parameter types of the target building parameters.

[0137] Using the two target building parameters, building area and value attribute information, S1010 is executed to obtain the building's photovoltaic installation area, which is the predicted photovoltaic installation area of ​​the buildings within the target area. Furthermore, S1010-S1014 are executed to obtain target buildings with high photovoltaic installation potential, and photovoltaic installation is recommended for these target buildings.

[0138] In the embodiment of the present application, deep high-resolution representation learning image recognition technology is first used to identify rooftop distributed photovoltaics on Beijing's 0.28-meter resolution satellite map / map satellite image, and the photovoltaic area (installed capacity) adopted by each building is obtained, with an identification accuracy of 89.5%. On this basis, geographically weighted regression is used to analyze the impact of factors such as building density, building height, and housing prices on the installed capacity of rooftop distributed photovoltaics, and this is used as the regression coefficient to predict the photovoltaic installation potential of the remaining buildings. The research results show that the maximum installed capacity of rooftop distributed photovoltaics in Beijing is approximately 6.9877 million kilowatts, and the power generation is approximately 8.483 billion kilowatt-hours, which can meet 7.44% of Beijing's total social electricity consumption and reduce carbon dioxide emissions by 7.9899 million tons each year, equivalent to 13.33% of Beijing's carbon dioxide emissions in 2017. Rooftop distributed photovoltaics have great development significance.

[0139] Finally, a cluster analysis was conducted on the spatial layout of rooftop distributed photovoltaics in Beijing. The clustering elements included building density, building height, housing prices and photovoltaic area. The results showed that: (1) The distribution of rooftop distributed photovoltaics in the central old city is weaker than that in the secondary new area. (2) As the building density increases, the negative impact of building height on photovoltaic installations becomes increasingly obvious. (3) In areas with low building density, the higher the housing price, the larger the photovoltaic installation, while in areas with high building density, the higher the housing price, the smaller the photovoltaic installation.

[0140] The embodiments of the present application analyze the main factors affecting photovoltaic installation from the architectural properties of buildings that have already installed photovoltaics, and use these main factors to predict the installation potential of buildings that have not installed photovoltaics. The analysis is carried out with the help of multiple models to replace the manual recommendation process, significantly improving the rationality, accuracy and intelligence of the recommendations.

[0141] The above embodiments have described the photovoltaic recommendation method provided by the present application in detail. The photovoltaic recommendation device, electronic device, storage medium and program product provided by the embodiments of the present application will be specifically explained below.

[0142] Figure 11 This is a schematic diagram of the structure of a photovoltaic recommendation device based on photovoltaic identification and clustering provided in an embodiment of the present application. As shown in the figure, the photovoltaic recommendation device 1100 includes:

[0143] An acquisition module 1101 is configured to acquire target building parameters and location information corresponding to each building in a target area, where the target building parameters are building parameters that affect photovoltaic installation;

[0144] A first determination module 1102 is configured to input target building parameters and location information corresponding to each building as dependent variables into a GWR model to obtain the building's photovoltaic installation area, wherein the GWR model is configured to reflect the correspondence between the target building parameters and location information and the building's photovoltaic installation area;

[0145] The second determining module 1103 is configured to determine target buildings recommended for photovoltaic installation within the target area based on the photovoltaic installation area of ​​each building within the target area.

[0146] Optionally, the first determining module 1102 may be configured to determine the photovoltaic installation area of ​​the building according to the following formula:

[0147]

[0148] Among them, y i represents the photovoltaic installation area of ​​the i-th building, (u i , v i ) represents the location information of the i-th building, β0(u i , v i) represents the intercept term of the i-th building, β k (u i , v i ) represents the kth regression coefficient of the i-th building, x ij represents the jth target building parameter of the i-th building, the value range of j is [1, k], k is the total number of parameter types of the target building parameter, ε i represents the regression residual.

[0149] Optionally, the second determination module 1103 can be used to obtain the building density, building height and value attribute information of the building; determine the building density level, building height level and value attribute level of the building based on the building density, building height and value attribute information respectively; combine the building density, building height and value attribute information of different levels to form a clustering matrix of buildings in the target area; and determine the buildings corresponding to the combinations whose average photovoltaic installation area meets the requirements in the clustering matrix as the target buildings recommended for photovoltaic installation in the target area.

[0150] Optionally, the photovoltaic recommendation device 1100 may further include a third determination module, which may be used to obtain the location information and photovoltaic installation area of ​​sample buildings that have installed photovoltaics in the specified area before obtaining the target building parameters and location information corresponding to each building in the target area; obtain the building parameters of the sample buildings, where the types of building parameters include building density, building height, building outline area, building area, building type and value attribute information; input the combination of different types of building parameters into the GWR model as independent variables one by one, and use the photovoltaic installation area as the dependent variable for fitting processing; calculate the R-square value corresponding to the combination of different types of building parameters based on the fitting processing results; determine the type of building parameters contained in the combination of building parameters with the largest R-square value as the parameter type of the target building parameter.

[0151] Optionally, the third determination module can also be used to identify photovoltaic equipment installed on sample buildings based on map satellite images of the specified area using the HRNet algorithm model; mark the identified installed photovoltaic equipment to obtain a marked area; determine the marked area as the photovoltaic installation area, and determine the location information of the building corresponding to the marked area as the location information of the building where photovoltaics have been installed.

[0152] Optionally, the third determination module can also be used to mark the identified installed photovoltaic equipment to obtain the marked area, and then compare the marked area with the building outline area data of the corresponding position; eliminate the marked area that is not in the building outline area to obtain a new marked area; correspondingly, determine the marked area as the photovoltaic installation area, and determine the location information of the building corresponding to the marked area as the location information of the building where photovoltaics have been installed, including: determining the new marked area as the photovoltaic installation area, and determining the location information of the building corresponding to the new marked area as the location information of the building where photovoltaics have been installed.

[0153] Optionally, the HRNet algorithm model is trained through the following process: obtaining a map satellite image of the sample area and marking information contained in the map satellite image, where the marking information marks the area of ​​photovoltaic equipment installed on the buildings in the sample area; using the map satellite image and the marking information to train the HRNet algorithm model to obtain a trained HRNet algorithm model.

[0154] The device provided in the embodiment of the present application can be used to execute the above-mentioned photovoltaic recommendation method. Its implementation method and technical effects are similar and will not be repeated here.

[0155] Figure 12 This is a schematic diagram of the structure of the electronic device provided in the embodiment of the present application. Figure 12 As shown, the electronic device 1200 includes:

[0156] Processor 1201 , memory 1202 , communication interface 1203 and system bus 1204 .

[0157] Among them, the memory 1202 and the communication interface 1203 are connected to the processor 1201 through the system bus 1204 and complete communication with each other. The memory 1202 is used to store computer execution instructions, the communication interface 1203 is used to communicate with other devices, and the processor 1201 is used to execute computer execution instructions to execute the photovoltaic recommendation method as described in the above method embodiment.

[0158] Specifically, the processor 1201 may include one or more processing units. For example, the processor 1201 may be a CPU, a digital signal processing (DSP), an application-specific integrated circuit (ASIC), etc. A general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in the application may be directly implemented by a hardware processor or implemented by a combination of hardware and software modules in the processor.

[0159] The memory 1202 can be used to store program instructions. The memory 1202 may include a program storage area and a data storage area. Among them, the program storage area may store an operating system, an application required for at least one function (such as a sound playback function, etc.), etc. The data storage area may store data created during the use of the electronic device 1200 (such as audio data, etc.), etc. In addition, the memory 1202 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, a universal flash memory (Universal Flash Storage, referred to as UFS), etc. The processor 1201 executes various functional applications and data processing of the electronic device 1200 by running the program instructions stored in the memory 1202.

[0160] The communication interface 1203 can provide solutions for wireless communications, including 2G / 3G / 4G / 112G, applied to the electronic device 1200. The communication interface 1203 can receive electromagnetic waves through the antenna, filter, amplify, and process the received electromagnetic waves, and transmit them to the modem processor for demodulation. The communication interface 1203 can also amplify the signal modulated by the modem processor and convert it into electromagnetic waves for radiation through the antenna. In some embodiments, at least some functional modules of the communication interface 1203 can be set in the processor 1201. In some embodiments, at least some functional modules of the communication interface 1203 can be set in the same device as at least some modules of the processor 1201.

[0161] The system bus 1204 may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The system bus 1204 may be divided into an address bus, a data bus, a control bus, and the like. For ease of illustration, the figure shows only one thick line, but this does not imply that there is only one bus or only one type of bus.

[0162] It should be noted that the number of memories 1202 and processors 1201 is not limited in this embodiment of the present application, and they can be one or more. Figure 12Taking one as an example, the memory 1202 and the processor 1201 can be connected by wire or wireless means in various ways, such as via a bus. In practical applications, the electronic device 1200 can be various forms of computers or mobile terminals. Computers include laptop computers, desktop computers, workstations, servers, blade servers, mainframe computers, etc.; mobile terminals include personal digital assistants, cellular phones, smart phones, wearable devices, and other similar computing devices.

[0163] The electronic device of this embodiment can be used to execute the technical solution in the above method embodiment. Its implementation principle and technical effects are similar and will not be described in detail here.

[0164] An embodiment of the present application further provides a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the photovoltaic recommendation method in the above method embodiment.

[0165] An embodiment of the present application further provides a computer program product, including a computer program; when the computer program is executed, the photovoltaic recommendation method in the above method embodiment is implemented.

[0166] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present application that follow the general principles of the present application and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, and the true scope and spirit of the present application are indicated by the following claims.

[0167] It should be understood that the present application is not limited to the exact structure described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. A photovoltaic recommendation method based on photovoltaic identification and clustering, characterized in that: include: Obtaining target building parameters and location information corresponding to each building in the target area, wherein the target building parameters are building parameters that affect photovoltaic installation; For each of the buildings, input the target building parameters and location information corresponding to the building as dependent variables into a geographically weighted regression (GWR) model to obtain the photovoltaic installation area of ​​the building, wherein the GWR model is used to reflect the corresponding relationship between the target building parameters and location information and the photovoltaic installation area of ​​the building; For each of the buildings, obtaining the building density, building height, and value attribute information of the building; determining the building density level, building height level, and value attribute level of the building based on the building density, building height, and value attribute information; combining the building density, building height, and value attribute information of different levels to form a clustering matrix of buildings in the target area; determining the buildings corresponding to the combinations whose average photovoltaic installation areas meet the requirements in the clustering matrix as target buildings recommended for photovoltaic installation in the target area; The parameter type of the target building parameter is determined based on the following method: obtaining the location information and photovoltaic installation area of ​​sample buildings that have installed photovoltaics in a specified area; obtaining the building parameters of the sample buildings, the types of the building parameters including building density, building height, building outline area, building area, building type and value attribute information; inputting the combination of different types of building parameters into the GWR model as independent variables one by one, and taking the photovoltaic installation area as the dependent variable for fitting processing; calculating the goodness of fit R-square value corresponding to the combination of different types of building parameters based on the fitting processing results; and determining the type of building parameters contained in the combination of building parameters with the largest R-square value as the parameter type of the target building parameter.

2. The photovoltaic recommendation method according to claim 1, characterized in that: The target building parameters and location information corresponding to the building are input into the GWR model as dependent variables to obtain the photovoltaic installation area of ​​the building, including: Determine the photovoltaic installation area of ​​the building according to the following formula: , Formula 1 in, represents the photovoltaic installation area of ​​the i-th building, Represents the location information of the i-th building, represents the intercept term of the i-th building, represents the kth regression coefficient of the i-th building, Represents the jth target building parameter of the i-th building, the value range of j is [1, k], and k is the total number of parameter types of the target building parameter. represents the regression residual.

3. The photovoltaic recommendation method according to claim 1, characterized in that: The step of obtaining the location information and photovoltaic installation area of ​​sample buildings that have photovoltaics installed in a designated area includes: Based on the satellite image of the map of the designated area, the photovoltaic equipment installed on the sample building is identified using the deep high-resolution representation learning HRNet algorithm model; Marking the identified installed photovoltaic equipment to obtain a marked area; The marked area is determined as the photovoltaic installation area, and the location information of the building corresponding to the marked area is determined as the location information of the building where photovoltaics have been installed.

4. The photovoltaic recommendation method according to claim 3, characterized in that: After marking the identified installed photovoltaic equipment to obtain the marked area, the method further includes: Comparing the marked area with the building outline area data of the corresponding position; Eliminate the marked areas that are not within the building outline area to obtain a new marked area; Correspondingly, determining the marked area as the photovoltaic installation area and determining the location information of the building corresponding to the marked area as the location information of the building where photovoltaics have been installed includes: determining the new marked area as the photovoltaic installation area and determining the location information of the building corresponding to the new marked area as the location information of the building where photovoltaics have been installed.

5. The photovoltaic recommendation method according to claim 3, characterized in that: The HRNet algorithm model is trained through the following process: Obtaining a satellite image of a sample area and marking information contained in the satellite image, wherein the marking information marks an area of ​​photovoltaic equipment installed on buildings in the sample area; The map satellite image and the marking information are used to train the HRNet algorithm model to obtain a trained HRNet algorithm model.

6. A photovoltaic recommendation device based on photovoltaic identification and clustering, characterized in that: include: An acquisition module is used to obtain target building parameters and location information corresponding to each building in the target area, wherein the target building parameters are building parameters that affect photovoltaic installation; A first determination module is configured to input, for each of the buildings, target building parameters and location information corresponding to the building as dependent variables into a geographically weighted regression (GWR) model to obtain the photovoltaic installation area of ​​the building, wherein the GWR model is configured to reflect the correspondence between the target building parameters and location information and the photovoltaic installation area of ​​the building; The second determination module is configured to obtain, for each of the buildings, building density, building height, and value attribute information of the building; determine the building density level, building height level, and value attribute level of the building based on the building density, building height, and value attribute information; combine the building density, building height, and value attribute information of different levels to form a clustering matrix of buildings within the target area; and determine the buildings corresponding to the combinations whose average photovoltaic installation areas meet the requirements in the clustering matrix as target buildings recommended for photovoltaic installation within the target area; The parameter type of the target building parameter in the acquisition module is determined based on the following method: obtaining the location information and photovoltaic installation area of ​​sample buildings with photovoltaic installation in the specified area; obtaining the building parameters of the sample buildings, the types of the building parameters including building density, building height, building outline area, building area, building type and value attribute information; inputting the combination of different types of building parameters into the GWR model as independent variables one by one, and taking the photovoltaic installation area as the dependent variable for fitting processing; calculating the goodness of fit R-square value corresponding to the combination of different types of building parameters based on the fitting processing results; and determining the type of building parameters contained in the combination of building parameters with the largest R-square value as the parameter type of the target building parameter.

7. An electronic device, characterized in that: include: Memory, processor; The memory is used to store program instructions; The processor is configured to call the program instructions to execute the photovoltaic recommendation method according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer-executable instructions, which are used to implement the photovoltaic recommendation method according to any one of claims 1 to 5 when executed by a processor.

Images