Centralized photovoltaic power station capacity prediction method and system based on satellite image and topographic data

Abstract

The invention belongs to the technical field of satellites, and discloses a centralized photovoltaic power station capacity prediction method and system based on a satellite image and topographic data, and the method comprises the steps: carrying out the recognition of a Sentinel remote sensing image, obtaining a photovoltaic distribution position after photovoltaic automatic recognition according to the spectral characteristics, including the mean value of each spectral band, NDVI, NDWI and other indexes, and carrying out the prediction of the capacity of a photovoltaic power station. Meanwhile, elevation data provided by DEM images are integrated to calculate a topographic slope value, then influence factors such as photovoltaic image area, inclination angle, direction angle and installation spacing are calculated by using parameters such as image resolution and geographic latitude, and finally, parameters such as known photovoltaic scale and photovoltaic materials are combined to obtain a photovoltaic power generation system. And a prediction model of the capacity of the centralized photovoltaic power station is constructed with the installation capacity of a small number of known photovoltaic power stations, so that accurate prediction from the image area of the photovoltaic power station identified by the satellite image to the actual installation capacity is realized. The method comprehensively considers the influence of the direction angle, the inclination angle, the installation spacing and the like of the photovoltaic power station on the actual installation capacity, is suitable for mountainous and hilly areas with complex terrains, improves the problem of inaccuracy of a method for estimating the capacity based on the projection area, and provides a scientific basis for the capacity evaluation of the photovoltaic power station.

Description

Technical Field

[0001] The present invention belongs to but is not limited to the field of satellite technology, and in particular relates to a centralized photovoltaic power station capacity prediction method and system based on satellite images and terrain data. Background Art

[0002] Under the background of dual carbon, photovoltaic power generation is a way to promote the transformation of energy structure. Due to the low cost of photovoltaic systems, it is widely adopted in my country, especially centralized photovoltaic power stations. But at the same time, due to the surge in number, the identification and capacity assessment of photovoltaic devices has also become a problem. Obtaining specific capacity data of photovoltaics is of great significance to many parties. Electric power utilities rely on photovoltaic databases for long-term capacity planning of power grids, which not only supports the daily operation of the power grid, but also helps energy trading activities. For the government, mastering the latest and reliable database information in the field of photovoltaic applications is the key to formulating effective energy policies. At the same time, researchers also benefit from these databases and can promote and evaluate innovative solutions for smart grids, covering explorations in areas such as electric vehicle integration and local microgrid management.

[0003] Existing photovoltaic power station capacity prediction methods mostly rely on ground-based field measurements or projected area identified based on satellite images. However, due to the existence of three-dimensional parameters such as tilt angle, azimuth angle, and installation spacing, the projected area presented in the image is different from the area of ​​the actual installed panels. Therefore, the area directly identified through satellite images is difficult to accurately reflect the actual installed capacity of the photovoltaic power station.

[0004] To this end, a centralized photovoltaic power station capacity prediction method based on satellite images and terrain data is proposed. The present invention collects the mean of each spectral band of the Sentinel remote sensing image, index features such as NDVI and NDWI, and calculates the slope value of each pixel point and the geographical latitude of the installation site based on the elevation data provided by the DEM image, establishes a fitting relationship between the above parameters of a small number of known photovoltaics and the known capacity, and establishes a calculation method for predicting the installation capacity based on the identified area. This method not only has a high efficiency in photovoltaic power station capacity identification, but also takes into account the three-dimensional direction of photovoltaic installation, and the capacity prediction of photovoltaic power stations is more accurate. Summary of the invention

[0005] In view of the problems existing in the prior art, the present invention provides a centralized photovoltaic power station capacity prediction method and system based on satellite images and terrain data.

[0006] The present invention is implemented as follows: a centralized photovoltaic power station capacity prediction method based on satellite images and terrain data, the method comprising:

[0007] S1: Obtain remote sensing images of designated centralized photovoltaic power station areas through satellites, process the images, and calculate relevant index features based on the image data;

[0008] S2: Based on the PV power station remote sensing image dataset obtained in S1, the classification model is trained and tested to identify PV in the designated area, eliminating interference factors such as vegetation, waters, and roads, and obtaining the distribution location of PV;

[0009] S3: using the photovoltaic distribution position obtained in S2 to obtain the number of photovoltaic array pixels, and combining the image resolution of the satellite image to calculate the projected area of ​​the identified centralized photovoltaic;

[0010] S4: Calculate the terrain slope dataset of the specified area based on the DEM image digital elevation data;

[0011] S5: Combine the photovoltaic distribution location, local solar radiation meteorological parameters, geographic latitude, terrain slope, etc. to calculate and identify the optimal photovoltaic installation tilt angle, direction angle and other parameters to display the three-dimensional position of the photovoltaic array;

[0012] S6: Using the photovoltaic data set obtained above, including the optimal photovoltaic installation tilt angle, direction angle, terrain slope, etc., and combining the known photovoltaic scale, photovoltaic materials and other influencing factors to form a multidimensional vector, a regression model is constructed between the photovoltaic projection area and the known local photovoltaic capacity to achieve accurate prediction of the photovoltaic power station capacity;

[0013] S7: After processing the satellite remote sensing images of the photovoltaic power station to be predicted, photovoltaic identification is performed to obtain the photovoltaic image area, and then the known optimal tilt angle, direction angle, photovoltaic material and other parameters are brought into the fitted relationship to obtain the local photovoltaic capacity data.

[0014] Furthermore, in S1, the image needs to be processed by color correction, image segmentation, etc., and the spectral band mean, NDVI, NDWI and other index features need to be calculated as the data set for training the model.

[0015] Furthermore, in S2, as a preferred embodiment, the recognition model may adopt a relatively mature model such as a random forest model or a deep learning model.

[0016] Furthermore, in S3, the area represented by a single pixel is calculated according to the unit resolution of the remote sensing image. For example, if the unit resolution is set to M and the unit resolution is 1m, then the area represented by a single pixel is 1m. 2 The number of photovoltaic pixels identified is extracted and represented by D, and the photovoltaic projection area is further calculated and represented by S:

[0017] S=M*D.

[0018] Furthermore, the terrain slope in S4 will affect the installation spacing, direction angle and east-west tilt angle of the photovoltaic array. Generally, the north-south tilt angle is relatively stable. After calculating the terrain slope, its influence needs to be considered from multiple perspectives.

[0019] Furthermore, the installation tilt angle in S5 refers to the angle between the photovoltaic array and the horizontal plane. Generally, the higher the latitude, the larger the tilt angle. The most ideal tilt angle is to maximize the annual power generation of the solar cell and minimize the difference in power generation between winter and summer. The optimal tilt angle of the photovoltaic can be obtained by combining the geographical latitude and the local meteorological annual statistical parameters.

[0020] The direction angle refers to the east-west, south-north orientation of the photovoltaic array. my country is located in the northern hemisphere, and the best direction angle is due south, i.e. 0°. However, for mountainous and hilly areas with uneven terrain, the direction angle will vary due to the slope, and needs to be analyzed based on specific circumstances.

[0021] Further, in S6, firstly, the prediction of the photovoltaic power station capacity based on the photovoltaic area is linear, but the installation tilt angle itself is nonlinear for the installation capacity, so it is necessary to perform a trigonometric function transformation on the installation tilt angle. Let the installation tilt angle be α, and calculate the value of 1 / cos(α) as one of the parameters of the fitting formula to ensure its linear relationship. Secondly, it is known that the area of ​​the photovoltaic projection obtained for each photovoltaic array is deviated from the actual area due to the influence of the three-dimensional position, but the deviation caused is not the same. At the same time, the scale of the photovoltaic array itself and the photovoltaic material are different, and the power generation efficiency of the photovoltaic cell is also different. In order to make the fitted regression model more accurate, each photovoltaic array in the regression model has its own unique regression parameter. At the same time, the effect of the fitted relationship is evaluated to ensure the feasibility and reliability of this model in predicting photovoltaic capacity.

[0022] Furthermore, in S6, c represents the photovoltaic capacity, s represents the photovoltaic projection area, γ represents the capacity per square meter of the photovoltaic array, which is an estimated parameter, and the subscript i represents the i-th photovoltaic array. Assuming that the relationship between the photovoltaic capacity and the photovoltaic area is linear, the regression fitting model is:

[0023] c i =γ i s i +γ0

[0024] Among them, γ0 represents the bias coefficient, which represents the existing deviation; because the photovoltaic array is affected by the terrain, the installation inclination angle, direction angle, and installation spacing are not the same, and the projected area is different from the actual installation area. At the same time, there are factors such as the size of the photovoltaic array and photovoltaic materials that affect the power generation capacity. In order to achieve accurate prediction of capacity, each photovoltaic array has a unique regression coefficient γ i . Let γi The expression is:

[0025] γ i =β T v i +β0

[0026] where v i represents the relevant predictive variables that affect capacity prediction, such as the optimal photovoltaic installation tilt angle, direction angle, terrain slope, photovoltaic array size and photovoltaic materials, etc., and β T The regression weight vectors representing the various prediction variables common to all PV arrays reflect the degree of influence of each variable on the PV capacity prediction. The bold symbols represent vectors, and β0 is the bias coefficient here;

[0027] In order to maintain the linear relationship of the fitting equation, the inclination angle needs to be processed. Assuming the installation inclination angle is α, the value of 1 / cos(α) is calculated as the composition v i One of the elements of

[0028] Transform the formula:

[0029] c i =(β T v i +β0)s i +γ0 (4)

[0030]

[0031] The final fitting expression is obtained:

[0032] c i =β′ T v′ i +β′0 (6)

[0033] where v′ i =[s i v i ,s i ],β′=[β T , β0] T ,β′0 represents the bias coefficient here. v′ i It is a multidimensional statistical vector composed of the known photovoltaic area and the installation tilt angle, direction angle, geographical slope, photovoltaic array size and photovoltaic material parameters. β′ and β′0 are the fitted regression coefficients, which can be obtained by regression estimation;

[0034] After obtaining the regression model, MSE (mean square error) can be used as an evaluation indicator of the regression fitting effect. The smaller the MSE, the better the fitting effect, and the more it can illustrate the feasibility of the model in predicting photovoltaic capacity based on photovoltaic area.

[0035] Another object of the present invention is to provide a centralized photovoltaic power station capacity prediction system based on satellite images and terrain data based on the centralized photovoltaic power station capacity prediction method based on satellite images and terrain data, the system specifically comprising:

[0036] The feature calculation module obtains remote sensing images of the designated centralized photovoltaic power station area through satellites, processes the images, and calculates relevant index features based on the image data;

[0037] The photovoltaic identification module is connected to the feature calculation module. Based on the acquired photovoltaic power station remote sensing image data set, the classification model is trained and tested to identify photovoltaics in the designated area, eliminating interference factors such as vegetation, waters, and roads, and obtaining the distribution location of photovoltaics.

[0038] The projection area calculation module is connected to the photovoltaic identification module, and uses the obtained photovoltaic distribution position to obtain the number of photovoltaic array pixels, and combines the image resolution of the satellite image to calculate the projection area of ​​the identified centralized photovoltaic;

[0039] The data set calculation module calculates the terrain slope data set of the specified area based on the DEM image digital elevation data;

[0040] The parameter identification module is connected to the photovoltaic identification module, and calculates the optimal photovoltaic installation tilt angle, direction angle and other parameters to display the three-dimensional position of the photovoltaic array in combination with the photovoltaic distribution location, local solar radiation meteorological parameters, geographic latitude, terrain slope, etc.;

[0041] The regression model building module is connected with the projection area calculation module, the data set calculation module and the parameter identification module. It uses the photovoltaic data set obtained above, including the optimal photovoltaic installation tilt angle, direction angle, terrain slope, etc., and combines the known photovoltaic scale, photovoltaic materials and other influencing factors to form a multidimensional vector, and builds a regression model between the photovoltaic projection area and the known local photovoltaic capacity to achieve accurate prediction of the photovoltaic power station capacity.

[0042] The data acquisition module is connected to the regression model building module. After processing the satellite remote sensing images of the photovoltaic power station to be predicted, photovoltaic identification is performed to obtain the photovoltaic image area. Then, the known optimal tilt angle, direction angle, photovoltaic material and other parameters are brought into the fitted relationship to obtain the local photovoltaic capacity data.

[0043] Another object of the present invention is to provide a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the centralized photovoltaic power station capacity prediction method based on satellite images and terrain data.

[0044] In combination with the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solutions to be protected by the present invention are as follows:

[0045] First, compared with the existing technology, the present invention additionally considers multiple factors such as direction angle, photovoltaic installation inclination angle and installation spacing, photovoltaic scale, photovoltaic materials, etc., which significantly improves the prediction accuracy of photovoltaic power station installation capacity and provides a basis for the reasonable operation and planning of photovoltaic power stations.

[0046] Compared with the existing technology, the present invention additionally considers multiple factors such as direction angle, photovoltaic installation inclination angle and installation spacing, photovoltaic scale, photovoltaic material, etc., and considers the three-dimensional orientation of photovoltaics and complex terrain through satellite images and DEM images in combination with image recognition models, solves the problem of deviation between the projection area of ​​photovoltaic images and the actual installation area, and proposes a capacity prediction model method with wider applicability and higher accuracy, which significantly improves the prediction accuracy of the installed capacity of photovoltaic power stations, and provides a basis for the reasonable operation and planning of photovoltaic power stations, the formulation of energy policies, etc.

[0047] The present invention automatically identifies photovoltaics through satellite images to obtain a reliable photovoltaic database without the need for large-scale ground field measurements, and reduces financial and time costs through efficient data processing and technical analysis.

[0048] The present invention takes into account the terrain complexity of photovoltaic installations, constructs a centralized photovoltaic power station capacity prediction model that is reliable and applicable to a variety of terrain conditions, and has strong practicality.

[0049] Second, the expected benefits and commercial value of the technical solution of the present invention after transformation are as follows: The efficient and high-precision capacity prediction of this technology helps photovoltaic power stations to reasonably arrange production plans, optimize resource allocation, reduce power generation losses caused by forecast deviations, better adapt to changes in power market demand, improve power supply reliability and flexibility, enhance market competitiveness, and thus improve the economic benefits of power stations. At the same time, as an important part of clean energy, the efficient operation of photovoltaic power stations is of great significance to achieving the global carbon neutrality goal, helping to optimize the energy structure, reduce carbon emissions, and promote sustainable development.

[0050] The technical solution of the present invention fills the technical gap in the industry at home and abroad: some previous studies have also proposed methods to determine the spatial position of photovoltaic systems, such as measuring the time series data of power generation, and approximately calculating the azimuth and tilt angle from the shape of the daily power generation curve, but such methods require accurate knowledge of the distribution location of the photovoltaic system, and most photovoltaic system power generation time series cannot be obtained, thus excluding the large-scale application of these methods. In contrast, this technology uses satellite images and terrain data that are easy to obtain and have a wider range of use to calculate the 3D direction of photovoltaics, which can be applied on a large scale.

[0051] The technical solution of the present invention solves a technical problem that people have been eager to solve but have never succeeded in solving: for a long time, the capacity prediction of photovoltaic power stations has faced problems such as incomplete data, poor adaptability of prediction models, and complex changes in the external environment, which makes it difficult to improve the prediction accuracy, thereby affecting the economic benefits of the power station and the stability of the power grid. The technical solution of the present invention accurately captures various factors that affect power generation by deeply mining data associations, optimizing the structure of prediction models, and introducing advanced machine learning or artificial intelligence algorithms, thereby significantly improving the prediction accuracy. This solution not only meets the industry's urgent need for high-precision predictions, but also lays a solid foundation for the efficient operation of photovoltaic power stations and the flexible scheduling of the power market.

[0052] Third, the technical solution of the present invention solves the following technical problems in the current capacity prediction of centralized photovoltaic power stations in industrial applications and achieves significant technical progress:

[0053] ###1. Low accuracy of existing photovoltaic power station capacity estimation

[0054] The existing technology roughly estimates the photovoltaic capacity by the area of ​​the photovoltaic power station, ignoring many key factors, such as the installation angle, direction, terrain slope, etc. of the photovoltaic panels, which leads to large deviations in the prediction results. The present invention introduces a combination of satellite images and terrain data, taking into account multi-dimensional factors such as the three-dimensional position information, optimal tilt angle, direction angle, terrain slope, etc. of the photovoltaic power station, and uses this information to construct a multi-dimensional regression model, which greatly improves the accuracy of photovoltaic power station capacity prediction. By considering the multi-factor influence of the actual installation environment, especially the processing of nonlinear characteristics, the actual photovoltaic power station power generation capacity can be calculated more accurately.

[0055] ###2. Lack of comprehensive analysis of terrain, meteorological and other data

[0056] Traditional photovoltaic power station capacity prediction methods are often based only on area, without considering the impact of terrain slope and meteorological parameters on photovoltaic efficiency. To address this problem, the present invention obtains terrain slope through satellite remote sensing images combined with DEM image digital elevation data, and calculates the optimal installation tilt angle and direction angle of the photovoltaic array in combination with local meteorological parameters such as solar radiation and latitude. This improved method can fully consider the impact of different terrain and geographical conditions on the power generation efficiency of photovoltaic panels in photovoltaic capacity prediction, making the prediction results more reliable and operational.

[0057] ###3. Insufficient model simplification

[0058] Existing capacity prediction methods usually use simple linear models, which are difficult to handle nonlinear factors such as the inclination angle of photovoltaic panels and terrain slope. The present invention linearizes the nonlinear factors by performing trigonometric function transformation on the installation inclination angle of photovoltaic panels, and then constructs a regression model to solve the nonlinear effect of the inclination angle on capacity prediction. In addition, by introducing multidimensional variables related to photovoltaic capacity (such as photovoltaic materials, array size, etc.) through the regression model, the prediction accuracy of photovoltaic capacity in complex installation environments is improved.

[0059] ###4. Improvements in data acquisition and automated processing

[0060] The present invention solves the problem of much manual intervention in the prior art based on the automated processing flow of satellite images. Photovoltaic distribution and terrain data are obtained through satellite remote sensing images, and combined with automated image processing and model training, the capacity of photovoltaic power stations can be efficiently identified and predicted. By introducing advanced machine learning algorithms such as random forests and deep learning, the automated processing capability of the model is significantly improved, the errors of manual operation are reduced, and the operating efficiency and scalability of the system are improved.

[0061] ###Significant technological advancements:

[0062] - Improved accuracy: Through the fusion of multi-dimensional data and the optimization of regression models, the prediction accuracy is significantly improved, especially in areas with complex terrain and large changes in meteorological conditions, more accurate photovoltaic power station capacity prediction results can be obtained.

[0063] - Strong applicability: The capacity prediction method of the present invention is applicable to different geographical environments, terrain conditions and photovoltaic installation layouts, and can be widely used in the prediction and evaluation of photovoltaic power station capacity across the country and even the world.

[0064] - High degree of automation: Automated processing and image recognition based on satellite images greatly improves data processing efficiency, making it suitable for real-time monitoring and capacity prediction of large-scale photovoltaic power plants, reducing the need for manual intervention.

[0065] - Broad industry application prospects: This technology can be applied to government agencies, photovoltaic development companies and other industries. It has important application value in the planning, operation and management of photovoltaic power stations, and helps to optimize the layout of photovoltaic resources and improve the operating efficiency of photovoltaic power stations.

[0066] Through this invention, the capacity of existing photovoltaic power stations can be accurately evaluated, and a reasonable capacity prediction basis can be provided for new photovoltaic projects, promoting the efficient development of the photovoltaic industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1is a flow chart of a centralized photovoltaic power station capacity prediction method based on satellite images and terrain data provided by an embodiment of the present invention;

[0068] Figure 2 is a flow chart of predicting photovoltaic capacity based on satellite images and fitting relationship provided by an embodiment of the present invention;

[0069] Figure 3 It is a structural diagram of a centralized photovoltaic power station capacity prediction system based on satellite images and terrain data provided by an embodiment of the present invention;

[0070] In the figure: 1. Feature calculation module; 2. Photovoltaic identification module; 3. Projection area calculation module; 4. Data set calculation module; 5. Parameter identification module; 6. Regression model construction module; 7. Data acquisition module. DETAILED DESCRIPTION

[0071] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0072] The following are two industrial application embodiments related to the present invention:

[0073] ###Example 1: Planning and design of large photovoltaic power stations

[0074] In the national new energy development projects, the construction site selection and planning of photovoltaic power stations are crucial. In order to accurately predict the power generation capacity of the photovoltaic power station during the design stage of the photovoltaic power station, the present invention can be applied to site selection and capacity planning. The candidate sites of the photovoltaic power station in the specified area are obtained through satellite remote sensing images, and the random forest algorithm or deep learning model of the present invention is used to analyze the area in combination with terrain slope, solar radiation, local meteorological data, etc. The system can automatically calculate the projected area, optimal tilt angle and direction angle of the photovoltaic array, and generate accurate capacity prediction data. The capacity prediction data provides an important basis for design engineers, which helps to determine the optimal photovoltaic array installation method for each plot, avoid underestimation or overestimation of power generation capacity, and maximize the utilization efficiency of solar energy. This method can significantly improve the design accuracy and project rate of return of photovoltaic power stations.

[0075] ###Example 2: Performance evaluation and expansion planning of existing photovoltaic power plants

[0076] For large-scale photovoltaic power stations that have been put into operation, companies need to regularly evaluate their operating performance to determine whether expansion or upgrading is needed. The present invention can be applied to the capacity assessment and optimization of photovoltaic power stations. By obtaining the latest satellite image data, combined with the layout and terrain information of the photovoltaic power station, the projected area and installation parameters of the existing photovoltaic power station are automatically identified and calculated. Combining historical meteorological data and the actual operation data of the power station, a regression model is used to accurately predict the power generation capacity and capacity deviation of the photovoltaic power station. This information can help companies make expansion decisions, such as which plots to add photovoltaic arrays or replace more efficient photovoltaic materials to increase the overall power generation capacity. This technology will greatly improve the management efficiency and profitability of power stations, and provide a scientific basis for future expansion projects.

[0077] These two embodiments demonstrate the application value of the present invention in the planning, design and management of photovoltaic power stations, which can not only provide capacity prediction for new power stations, but also provide data support for performance evaluation and expansion of existing power stations.

[0078] like Figure 1 , 2 As shown, an embodiment of the present invention provides a centralized photovoltaic power station capacity prediction method based on satellite images and terrain data, the method comprising:

[0079] S1: Obtain remote sensing images of designated centralized photovoltaic power station areas through satellites, process the images, and calculate relevant index features based on the image data;

[0080] S2: Based on the PV power station remote sensing image dataset obtained in S1, the classification model is trained and tested to identify PV in the designated area, eliminating interference factors such as vegetation, waters, and roads, and obtaining the distribution location of PV;

[0081] S3: using the photovoltaic distribution position obtained in S2 to obtain the number of photovoltaic array pixels, and combining the image resolution of the satellite image to calculate the projected area of ​​the identified centralized photovoltaic;

[0082] S4: Calculate the terrain slope dataset of the specified area based on the DEM image digital elevation data;

[0083] S5: Combine the photovoltaic distribution location, local solar radiation meteorological parameters, geographic latitude, terrain slope, etc. to calculate and identify the optimal photovoltaic installation tilt angle, direction angle and other parameters to display the three-dimensional position of the photovoltaic array;

[0084] S6: Using the photovoltaic data set obtained above, including the optimal photovoltaic installation tilt angle, direction angle, terrain slope, etc., and combining the known photovoltaic scale, photovoltaic materials and other influencing factors to form a multidimensional vector, a regression model is constructed between the photovoltaic projection area and the known local photovoltaic capacity to achieve accurate prediction of the photovoltaic power station capacity;

[0085] S7: After processing the satellite remote sensing images of the photovoltaic power station to be predicted, photovoltaic identification is performed to obtain the photovoltaic image area, and then the known optimal tilt angle, direction angle, photovoltaic material and other parameters are brought into the fitted relationship to obtain the local photovoltaic capacity data.

[0086] In the S1, the image needs to be processed by color correction, image segmentation, etc., and the spectral band mean, NDVI, NDWI and other index features need to be calculated as the data set for training the model.

[0087] In S2, as a preferred embodiment, the recognition model may adopt a relatively mature model such as a random forest model or a deep learning model.

[0088] In S3, the area represented by a single pixel is calculated according to the unit resolution of the remote sensing image. For example, if the unit resolution is set to M and the unit resolution is 1m, then the area represented by a single pixel is 1m. 2 The number of photovoltaic pixels identified is extracted and represented by D, and the photovoltaic projection area is further calculated and represented by S:

[0089] S=M*D.

[0090] The terrain slope in S4 will affect the installation spacing, direction angle and east-west tilt angle of the photovoltaic array. Generally, the north-south tilt angle is relatively stable. After calculating the terrain slope, its impact needs to be considered from multiple perspectives.

[0091] The installation tilt angle in S5 refers to the angle between the photovoltaic array and the horizontal plane. Generally, the higher the latitude, the larger the tilt angle. The most ideal tilt angle is to maximize the annual power generation of the solar cell and minimize the difference in power generation between winter and summer. The optimal tilt angle of the photovoltaic can be obtained by combining the geographical latitude and the local meteorological annual statistical parameters.

[0092] The direction angle refers to the east-west, south-north orientation of the photovoltaic array. my country is located in the northern hemisphere, and the best direction angle is due south, i.e. 0°. However, for mountainous and hilly areas with uneven terrain, the direction angle will vary due to the slope, and needs to be analyzed based on specific circumstances.

[0093] In S6, firstly, the prediction of the photovoltaic power station capacity based on the photovoltaic area is linear, but the installation tilt angle itself is nonlinear for the installation capacity, so it is necessary to perform a trigonometric function transformation on the installation tilt angle. Let the installation tilt angle be α, and calculate the value of 1 / cos(α) as one of the parameters of the fitting formula to ensure its linear relationship. Secondly, it is known that the area of ​​the photovoltaic projection obtained for each photovoltaic array is deviated from the actual area due to the influence of the three-dimensional position, but the deviation caused is not the same. At the same time, the scale of the photovoltaic array itself and the photovoltaic material are different, and the power generation efficiency of the photovoltaic cells is also different. In order to make the fitted regression model more accurate, each photovoltaic array in the regression model has its own unique regression parameter. At the same time, the effect of the fitted relationship is evaluated to ensure the feasibility and reliability of this model in predicting photovoltaic capacity.

[0094] In S6, c represents the photovoltaic capacity, s represents the photovoltaic projection area, γ represents the capacity per square meter of the photovoltaic array, which is an estimated parameter, and the subscript i represents the i-th photovoltaic array. Assuming that the relationship between photovoltaic capacity and photovoltaic area is linear, the regression fitting model is:

[0095] c i =γ i s i +γ0

[0096] Among them, γ0 represents the bias coefficient, which represents the existing deviation; because the photovoltaic array is affected by the terrain, the installation inclination angle, direction angle, and installation spacing are not the same, and the projected area is different from the actual installation area. At the same time, there are factors such as the size of the photovoltaic array and photovoltaic materials that affect the power generation capacity. In order to achieve accurate prediction of capacity, each photovoltaic array has a unique regression coefficient γ i . Let γ i The expression is:

[0097] γ i =β T v i +β0

[0098] where v i represents the relevant predictive variables that affect capacity prediction, such as the optimal photovoltaic installation tilt angle, direction angle, terrain slope, photovoltaic array size and photovoltaic materials, etc., and β T The regression weight vectors representing the various prediction variables common to all PV arrays reflect the degree of influence of each variable on the PV capacity prediction. The bold symbols represent vectors, and β0 is the bias coefficient here;

[0099] In order to maintain the linear relationship of the fitting equation, the inclination angle needs to be processed. Assuming the installation inclination angle is α, the value of 1 / cos(α) is calculated as the composition v iOne of the elements of

[0100] Transform the formula:

[0101] c i =(β T v i +β0)s i +γ0 (4)

[0102]

[0103] The final fitting expression is obtained:

[0104] c i =β′ T v′ i +β′0 (6)

[0105] where ν′ i =[s i ν i ,s i ],β′=[β T , β0] T ,β′0 represents the bias coefficient here. v′ i It is a multidimensional statistical vector composed of the known photovoltaic area and the installation tilt angle, direction angle, geographical slope, photovoltaic array size and photovoltaic material parameters. β′ and β′0 are the fitted regression coefficients, which can be obtained by regression estimation;

[0106] After obtaining the regression model, MSE (mean square error) can be used as an evaluation indicator of the regression fitting effect. The smaller the MSE, the better the fitting effect, and the more it can illustrate the feasibility of the model in predicting photovoltaic capacity based on photovoltaic area.

[0107] like Figure 3 As shown, an embodiment of the present invention provides a centralized photovoltaic power station capacity prediction system based on satellite images and terrain data based on the centralized photovoltaic power station capacity prediction method based on satellite images and terrain data, and the system specifically includes:

[0108] The feature calculation module 1 obtains remote sensing images of the designated centralized photovoltaic power station area through satellites, processes the images, and calculates relevant index features based on the image data;

[0109] The photovoltaic identification module 2 is connected to the feature calculation module 1, and trains and tests the classification model based on the acquired photovoltaic power station remote sensing image data set, performs photovoltaic identification in the designated area, eliminates interference factors such as vegetation, waters, roads, etc., and obtains the distribution location of photovoltaics;

[0110] The projection area calculation module 3 is connected to the photovoltaic identification module 2, and uses the obtained photovoltaic distribution position to obtain the number of photovoltaic array pixels, and combines the image resolution of the satellite image to calculate the projection area of ​​the identified centralized photovoltaic;

[0111] The data set calculation module 4 calculates the terrain slope data set of the specified area according to the DEM image digital elevation data;

[0112] The parameter identification module 5 is connected to the photovoltaic identification module 2, and calculates and identifies the optimal photovoltaic installation inclination angle, direction angle and other parameters to display the three-dimensional position of the photovoltaic array in combination with the photovoltaic distribution position, local solar radiation meteorological parameters, geographical latitude, terrain slope, etc.;

[0113] The regression model building module 6 is connected with the projection area calculation module 3, the data set calculation module 4 and the parameter identification module 5, and uses the photovoltaic data set obtained above, including the photovoltaic optimal installation tilt angle, direction angle, terrain slope, etc., and combines the known photovoltaic scale, photovoltaic materials and other influencing factors to form a multidimensional vector, and builds a regression model between the photovoltaic projection area and the known local photovoltaic capacity to achieve accurate prediction of the photovoltaic power station capacity;

[0114] The data acquisition module 7 is connected to the regression model building module 6, and performs photovoltaic identification on the satellite remote sensing image of the photovoltaic power station to be predicted, and obtains the photovoltaic image area. Then, the known optimal tilt angle, direction angle, photovoltaic material and other parameters are brought into the fitted relationship to obtain the local photovoltaic capacity data.

[0115] An embodiment of the present invention provides a computer device, which includes a memory and a processor. The memory stores a computer program. When the computer program is executed by the processor, the processor executes the steps of the centralized photovoltaic power station capacity prediction method based on satellite images and terrain data.

[0116] The working principle of the embodiment of the present invention is based on the combination of satellite remote sensing images and terrain data. Through the synergy of multiple modules, centralized photovoltaic power stations are automatically identified and their capacity is accurately predicted. First, the feature calculation module obtains remote sensing images of the photovoltaic power station area through satellites, performs necessary image processing, such as color correction, image segmentation, etc., and calculates relevant image features, such as NDVI (normalized vegetation index), NDWI (water body index), etc. These features provide basic data for subsequent photovoltaic identification, ensuring that the features obtained after image processing accurately reflect the regional characteristics of the photovoltaic power station.

[0117] Secondly, the photovoltaic recognition module uses the data provided by the feature calculation module to identify and eliminate interference factors (such as vegetation, waters, roads, etc.) through a trained classification model (such as a random forest or deep learning model), thereby obtaining the distribution location of photovoltaic power stations. The main function of this module is to use satellite image data to accurately identify photovoltaic arrays to ensure that the identification process is efficient and accurate.

[0118] Next, the projection area calculation module calculates the projection area of ​​the photovoltaic array based on the identified photovoltaic array distribution location and the resolution of the image. By calculating the representative area of ​​each pixel, the module calculates the number of identified photovoltaic pixels and further determines the actual projection area of ​​the photovoltaic array. At the same time, the data set calculation module uses digital elevation data (DEM) to calculate the terrain slope data set of the specified area, providing the necessary terrain information for the subsequent calculation of the installation tilt angle and direction angle.

[0119] Finally, the regression model building module combines multi-dimensional data such as photovoltaic projection area, optimal installation tilt angle, direction angle, terrain slope, photovoltaic materials, etc. to predict the capacity of photovoltaic power stations through regression models. The system uses these variables to build a multi-dimensional regression model to ensure the accuracy of capacity prediction. After regression analysis, the system can predict the capacity of new photovoltaic power stations in the area to be tested, and continuously process new satellite images through the data acquisition module to update the capacity prediction results of photovoltaic power stations. This method greatly improves the accuracy and efficiency of photovoltaic power station capacity prediction and is suitable for large-scale, automated photovoltaic power station planning and management.

[0120] Relevant evidence of the technical effects achieved by the embodiments of the present invention.

[0121] 1. Photovoltaic area correction: When the present invention establishes the model, the two-dimensional projection area of ​​the photovoltaic can be obtained through image recognition, and the addition of three-dimensional parameters can correct the projection area. After comparing with the real data, it can be found that it is closer to the installation area of ​​the photovoltaic. At the same time, the known data is compared with the photovoltaic data calculated by the model. The results can prove that the model is more accurate in predicting the photovoltaic capacity.

[0122] 2. One-to-one estimation parameters: Taking into account the installation location, terrain conditions, and even physical characteristics and size of the photovoltaics that affect the capacity prediction, each photovoltaic in this model has corresponding estimation parameters, which is applicable to photovoltaic equipment under various conditions. Compared with general and fixed models, the establishment of variable and targeted models improves the accuracy of capacity prediction and can serve as evidence of the technical effect of the invention example.

[0123] 3. Papers and verification: Related papers conduct statistical analysis on similar mathematical models, and use mean square error (MSE) to compare the predicted capacity value given by each regression model with the true value, proving that the model has a lower MSE value, better fitting effect and prediction performance than the model that only considers the photovoltaic projection area. At the same time, there are documents showing that the inclination angle and direction angle calculated using satellite images and terrain data have a high accuracy rate, further proving that the photovoltaic capacity prediction model that considers the photovoltaic 3D direction has higher performance than the ordinary 2D direction model, especially for photovoltaic arrays with larger inclination angles, the prediction effect is more significantly improved. All of the above can serve as evidence of the technical effect of this invention example.

[0124] It should be noted that the embodiments of the present invention can be implemented by hardware, software, or a combination of software and hardware. The hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. It can be understood by a person of ordinary skill in the art that the above-mentioned devices and methods can be implemented using computer executable instructions and / or contained in a processor control code, such as a carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. Such code is provided on the carrier medium. The device and its modules of the present invention can be implemented by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., can also be implemented by software executed by various types of processors, and can also be implemented by a combination of the above-mentioned hardware circuits and software, such as firmware.

[0125] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principle of the present invention should be covered by the protection scope of the present invention.

Claims

1. A centralized photovoltaic power station capacity prediction method based on satellite images and terrain data, characterized in that: The method includes: S1: Obtain remote sensing images of designated centralized photovoltaic power station areas through satellites, process the images, and calculate relevant index features based on the image data; S2: Based on the PV power station remote sensing image dataset obtained in S1, the classification model is trained and tested to identify PV in the designated area, eliminating interference factors such as vegetation, waters, and roads, and obtaining the distribution location of PV; S3: using the photovoltaic distribution position obtained in S2 to obtain the number of photovoltaic array pixels, and combining the image resolution of the satellite image to calculate the projected area of ​​the identified centralized photovoltaic; S4: Calculate the terrain slope dataset of the specified area based on the DEM image digital elevation data; S5: Combine the photovoltaic distribution location, local solar radiation meteorological parameters, geographic latitude, terrain slope, etc. to calculate and identify the optimal photovoltaic installation tilt angle, direction angle and other parameters to display the three-dimensional position of the photovoltaic array; S6: Using the photovoltaic data set obtained above, including the optimal photovoltaic installation tilt angle, direction angle, terrain slope, etc., and combining the known photovoltaic scale, photovoltaic materials and other influencing factors to form a multidimensional vector, a regression model is constructed between the photovoltaic projection area and the known local photovoltaic capacity to achieve accurate prediction of the photovoltaic power station capacity; S7: After processing the satellite remote sensing images of the photovoltaic power station to be predicted, photovoltaic identification is performed to obtain the photovoltaic image area, and then the known optimal tilt angle, direction angle, photovoltaic material and other parameters are brought into the fitted relationship to obtain the local photovoltaic capacity data.

2. The centralized photovoltaic power station capacity prediction method based on satellite images and terrain data according to claim 1 is characterized in that: In the S1, the image needs to be processed by color correction, image segmentation, etc., and the spectral band mean, NDVI, NDWI and other index features need to be calculated as the data set for training the model.

3. The centralized photovoltaic power station capacity prediction method based on satellite images and terrain data as claimed in claim 1, characterized in that: In S2, as a preferred embodiment, the recognition model may adopt a relatively mature model such as a random forest model or a deep learning model.

4. The centralized photovoltaic power station capacity prediction method based on satellite images and terrain data as claimed in claim 1, characterized in that: In S3, the area represented by a single pixel is calculated according to the unit resolution of the remote sensing image. For example, if the unit resolution is set to M and the unit resolution is 1m, then the area represented by a single pixel is 1m. 2 The number of photovoltaic pixels identified is extracted and represented by D, and the photovoltaic projection area is further calculated and represented by S: S=M*D.

5. The centralized photovoltaic power station capacity prediction method based on satellite images and terrain data as claimed in claim 1, characterized in that: The terrain slope in S4 will affect the installation spacing, direction angle and east-west tilt angle of the photovoltaic array. Generally, the north-south tilt angle is relatively stable. After calculating the terrain slope, its impact needs to be considered from multiple perspectives.

6. The centralized photovoltaic power station capacity prediction method based on satellite images and terrain data as claimed in claim 1, characterized in that: The installation tilt angle in S5 refers to the angle between the photovoltaic array and the horizontal plane. Generally, the higher the latitude, the larger the tilt angle. The most ideal tilt angle is to maximize the annual power generation of the solar cell and minimize the difference in power generation between winter and summer. The optimal tilt angle of the photovoltaic can be obtained by combining the geographical latitude and the local meteorological annual statistical parameters. The direction angle refers to the east-west, south-north orientation of the photovoltaic array. my country is located in the northern hemisphere, and the best direction angle is due south, i.e. 0°. However, for mountainous and hilly areas with uneven terrain, the direction angle will vary due to the slope, and needs to be analyzed based on specific circumstances.

7. The centralized photovoltaic power station capacity prediction method based on satellite images and terrain data as claimed in claim 1, characterized in that: In S6, firstly, the prediction of the photovoltaic power station capacity based on the photovoltaic area is linear, but the installation tilt angle itself is nonlinear for the installation capacity, so it is necessary to perform a trigonometric function transformation on the installation tilt angle. Let the installation tilt angle be α, and calculate the value of 1 / cos(α) as one of the parameters of the fitting formula to ensure its linear relationship. Secondly, it is known that the area of ​​the photovoltaic projection obtained for each photovoltaic array is deviated from the actual area due to the influence of the three-dimensional position, but the deviation caused is not the same. At the same time, the scale of the photovoltaic array itself and the photovoltaic material are different, and the power generation efficiency of the photovoltaic cells is also different. In order to make the fitted regression model more accurate, each photovoltaic array in the regression model has its own unique regression parameter. At the same time, the effect of the fitted relationship is evaluated to ensure the feasibility and reliability of this model in predicting photovoltaic capacity.

8. The centralized photovoltaic power station capacity prediction method based on satellite images and terrain data as claimed in claim 1, characterized in that: In S6, c represents the photovoltaic capacity, s represents the photovoltaic projection area, γ represents the capacity per square meter of the photovoltaic array, which is an estimated parameter, and the subscript i represents the i-th photovoltaic array. Assuming that the relationship between photovoltaic capacity and photovoltaic area is linear, the regression fitting model is: c i =c i s i +γ0 Among them, γ0 represents the bias coefficient, which represents the existing deviation; because the photovoltaic array is affected by the terrain, the installation inclination angle, direction angle, and installation spacing are not the same, and the projected area is different from the actual installation area. At the same time, there are factors such as the size of the photovoltaic array and photovoltaic materials that affect the power generation capacity. In order to achieve accurate prediction of capacity, each photovoltaic array has a unique regression coefficient γ i . Let γ i The expression is: c i =b T v i +β0 where v i represents the relevant predictive variables that affect capacity prediction, such as the optimal photovoltaic installation tilt angle, direction angle, terrain slope, photovoltaic array size and photovoltaic materials, etc., and β T The regression weight vector representing each prediction variable common to all PV arrays reflects the influence of each variable on PV capacity prediction. The bold symbols represent vectors, and β0 is the bias coefficient here; In order to maintain the linear relationship of the fitting equation, the inclination angle needs to be processed. Assuming the installation inclination angle is α, the value of 1 / cos(α) is calculated as the composition v i One of the elements of Transform the formula: c i =(β T v i +β0)s i +γ0 (4) The final fitting expression is obtained: c i =β′ T v′ i +β′0 (6) where v′ i =[s i v i ,s i ],β′=[β T ,β0] T ,β′0 represents the bias coefficient here. v′ i It is a multidimensional statistical vector composed of the known photovoltaic area and the installation tilt angle, direction angle, geographical slope, photovoltaic array size and photovoltaic material parameters. β′ and β′0 are the fitted regression coefficients, which can be obtained by regression estimation; After obtaining the regression model, MSE (mean square error) can be used as an evaluation indicator of the regression fitting effect. The smaller the MSE, the better the fitting effect, and the more it can illustrate the feasibility of the model in predicting photovoltaic capacity based on photovoltaic area.

9. A centralized photovoltaic power station capacity prediction system based on satellite images and terrain data based on the centralized photovoltaic power station capacity prediction method based on satellite images and terrain data as claimed in claims 1-8, characterized in that: The system specifically includes: The feature calculation module obtains remote sensing images of the designated centralized photovoltaic power station area through satellites, processes the images, and calculates relevant index features based on the image data; The photovoltaic identification module is connected to the feature calculation module. Based on the acquired photovoltaic power station remote sensing image data set, the classification model is trained and tested to identify photovoltaics in the designated area, eliminating interference factors such as vegetation, waters, and roads, and obtaining the distribution location of photovoltaics. The projection area calculation module is connected to the photovoltaic identification module, and uses the obtained photovoltaic distribution position to obtain the number of photovoltaic array pixels, and combines the image resolution of the satellite image to calculate the projection area of ​​the identified centralized photovoltaic; The data set calculation module calculates the terrain slope data set of the specified area based on the DEM image digital elevation data; The parameter identification module is connected to the photovoltaic identification module, and calculates the optimal photovoltaic installation tilt angle, direction angle and other parameters to display the three-dimensional position of the photovoltaic array in combination with the photovoltaic distribution location, local solar radiation meteorological parameters, geographic latitude, terrain slope, etc.; The regression model building module is connected with the projection area calculation module, the data set calculation module and the parameter identification module. It uses the photovoltaic data set obtained above, including the optimal photovoltaic installation tilt angle, direction angle, terrain slope, etc., and combines the known photovoltaic scale, photovoltaic materials and other influencing factors to form a multidimensional vector, and builds a regression model between the photovoltaic projection area and the known local photovoltaic capacity to achieve accurate prediction of the photovoltaic power station capacity. The data acquisition module is connected to the regression model building module. After processing the satellite remote sensing images of the photovoltaic power station to be predicted, photovoltaic identification is performed to obtain the photovoltaic image area. Then, the known optimal tilt angle, direction angle, photovoltaic material and other parameters are brought into the fitted relationship to obtain the local photovoltaic capacity data.

10. A computer device, characterized in that: The computer device includes a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the centralized photovoltaic power station capacity prediction method based on satellite images and terrain data as described in any one of claims 1-8.

Images