Building integrated photovoltaic power generation potential prediction method and system
By building a three-dimensional digital twin model and using deep learning algorithms to identify land use properties, combining solar resource characteristics and optimization algorithms, the problem of difficulty in accurately evaluating the potential of building photovoltaic power generation in the existing technology is solved, and high-precision photovoltaic layout planning and power generation potential prediction are achieved.
Patent Information
- Application Number
- CN202510610246.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-13
- Publication Date
- 2025-06-10
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The prior art is difficult to accurately evaluate the photovoltaic installation area and power generation potential of building surfaces, especially in terms of reflecting the three-dimensional spatial characteristics and dynamic environmental conditions of the building complex.
By integrating the spatial morphological prediction data, remote sensing images and GIS data of the building, a three-dimensional digital twin model is built, and a deep learning algorithm is used to identify the properties of land and divide functional partitions. Combining solar energy resource characteristics and optimization algorithms, we determine the optimal installation angle and arrangement of photovoltaic modules, and build a timing prediction model to predict photovoltaic power generation potential.
It has achieved high-precision evaluation of the potential of building photovoltaic power generation, solved the problems of single data sources of three-dimensional model and low accuracy of land use properties in the existing technology, significantly improved the accuracy of photovoltaic layout planning, and achieved high-precision prediction of photovoltaic power generation potential in the long-term evolution cycle.
Smart Images

Figure CN120125068A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of building photovoltaic power generation assessment, and more specifically, relates to a method and system for predicting the potential of building-integrated photovoltaic power generation. Background Art
[0002] In the technical field of building-integrated photovoltaic power generation assessment, accurately assessing the photovoltaic installable area and power generation potential on the building surface is the key to achieving efficient energy utilization.
[0003] In the prior art, the measurement methods of photovoltaic installable area usually rely on two-dimensional building drawings or limited three-dimensional modeling data, which are difficult to comprehensively reflect the three-dimensional spatial characteristics and dynamic environmental conditions of the building complex. For example, the existing modeling methods analyze the sunlight conditions on the building surface mostly relying on static parameters, ignoring the time-frequency variation characteristics of solar radiation and the dynamic shadow effects of surrounding obstacles. In addition, the existing technology lacks sufficient analysis on the matching between building layout and photovoltaic module installation methods. For example, it does not fully quantify the restrictive effects of building density gradient and orientation distribution differences on the installable area. In terms of power generation potential prediction, existing models mostly rely on a single data source or simplified assumptions, resulting in a significant deviation between the prediction results and the actual power generation capacity.
[0004] These technical bottlenecks make it difficult for the existing methods to meet the needs of fine assessment of photovoltaic resources for large-scale building complexes. Summary of the Invention
[0005] To solve the deficiencies in the prior art, the present invention provides a method and system for predicting the potential of building-integrated photovoltaic power generation.
[0006] The present invention adopts the following technical solutions.
[0007] The first aspect of the present invention provides a method for predicting the potential of building-integrated photovoltaic power generation, including the following steps: Integrate the spatial form prediction data, remote sensing images, and GIS of the buildings in the set area to construct a three-dimensional digital twin model of the set area, and use deep learning algorithms to identify the land use nature and divide functional areas including building density gradient and building orientation distribution characteristics; Based on the three-dimensional digital twin model of the set area, determine the total building area of the set area by region; based on the total building area of the buildings in the set area, and in combination with multi-dimensional influencing factors at the set area level, determine the building-integrated photovoltaic available area and building-integrated photovoltaic installation density by region; based on the characteristics of solar energy resources, use optimization algorithms to determine the optimal installation angle and arrangement of building-integrated photovoltaic modules, and respectively determine the radiation amount and effective surface area of the photovoltaic module surface in different regions; Based on the conversion efficiency of the photovoltaic module, the radiation amount on the surface of the photovoltaic module, and the effective surface area, a time series prediction model integrating the decomposition strategy and the swarm intelligence optimization algorithm is constructed to predict the photovoltaic developable installed capacity and the annual power generation potential in different zones during the spatial form adjustment period of the building.
[0008] Optionally, the construction of the 3D digital twin model of the set area includes: Collect the spatial form prediction data of the buildings in the set area and perform preprocessing; Obtain high-resolution remote sensing images of the set area through satellite remote sensing or aerial remote sensing, and perform radiometric correction on the original remote sensing images through sensor gain parameters and offset parameters; Collect the basic geographic information data of the set area and integrate them, including terrain data, road network data, and water system data, and import these data into the GIS platform for data format conversion and projection unification; Import the preprocessed spatial form prediction data, remote sensing images, and GIS data of the buildings in the set area into 3D modeling software, and use data fusion algorithms to integrate data from different sources to construct a 3D digital twin model of the set area.
[0009] Optionally, the use of deep learning algorithms to identify land use types and divide functional zones includes: Train a CNN model for land use type identification based on the 3D digital twin model of the set area, input the preprocessed 3D digital twin model of the set area into the trained CNN model, and obtain the predicted results of the land use type of each area; According to the land use type identification results, combined with the functional distribution characteristics in the spatial form prediction data of the buildings in the set area, use clustering algorithms to perform clustering analysis on the set area, and divide the areas with the same land use type and function into the same functional zone, including commercial areas, residential areas, and industrial areas.
[0010] Optionally, the determination of the building integrated photovoltaic available area and the building integrated photovoltaic installation density for each zone includes: Collect relevant data at the set area level, including building layout, orientation, and distribution information of surrounding obstacles, obtain the requirements for sunlight conditions for the installation of photovoltaic modules on the building surface, including the minimum sunlight duration and sunlight angle, and collect the construction control indicators of the set area, including building density, floor area ratio, and greening requirements; For different set area levels, analyze the influence degree of each dimensional influencing factor on the building integrated photovoltaic available area and installation density; According to the sunlight condition requirements for the installation of photovoltaic modules on the building surface, combined with the building density gradient and building orientation distribution characteristics of each zone, calculate the area of photovoltaic modules that can be installed on the buildings meeting the sunlight conditions in each zone. For each zone, calculate the available area of building-integrated photovoltaics through the installable area of building-integrated photovoltaics, the utilization coefficient of building-integrated photovoltaics, and the correction coefficient considering regulatory planning indicators in this zone; For each zone, calculate the installation density of building-integrated photovoltaics based on the available area of building-integrated photovoltaics in this zone and the total power of photovoltaic modules that can actually be installed in this zone.
[0011] Optionally, the method for determining the optimal installation angle and arrangement of building-integrated photovoltaic modules based on solar resource characteristics includes: Obtain the multi-dimensional characteristics of solar resources, including the minute-by-minute change data of total solar radiation, direct radiation, diffuse radiation, sunshine hours, solar altitude angle, and azimuth angle, and process the collected solar radiation data using wavelet analysis; Determine the optimal installation angle of building-integrated photovoltaic modules through a hybrid intelligent optimization algorithm. Taking the maximum solar radiation received on the surface of the photovoltaic modules as the objective, considering the solar radiation characteristics obtained from wavelet analysis, establish an objective function:
[0012] Wherein, At time , solar azimuth angle is , altitude angle is , and photovoltaic installation angle is , the solar radiation intensity received on the surface of the photovoltaic modules; is a weight function based on the wavelet analysis result; Adopt a hybrid intelligent optimization algorithm combining genetic algorithm and simulated annealing algorithm to determine the optimal installation angle ; Considering multi-factor coupling, including the geometric shape of the building, the bearing capacity of the building structure, the shadow occlusion of surrounding buildings and obstacles, the size and electrical connection requirements of photovoltaic modules, use the ant colony algorithm to optimize the arrangement of photovoltaic modules.
[0013] Optionally, the method for processing the collected solar radiation data using wavelet analysis includes: The original solar radiation time series is , select Daubechies wavelet for discrete wavelet transform, and decompose into components of different scales and frequencies:
[0014] Among them, is the detail coefficient, reflecting the high-frequency part of the signal; is the approximation coefficient, reflecting the low-frequency part of the signal; is the wavelet basis function, is the scaling function; By analyzing different scale components, the variation characteristics and laws of solar energy resources are extracted.
[0015] Optionally, the separately determining the radiation amount and the effective surface area of the photovoltaic module surface includes: Based on the acquired solar energy resource data, a radiation transfer model is constructed to calculate the radiation amount on the surface of the photovoltaic module. According to the collected solar energy resource data and atmospheric parameters, each parameter in the radiation transfer model is determined. Considering the reflection, refraction, and absorption characteristics of the photovoltaic module surface, a radiation energy balance equation for the photovoltaic module surface is established:
[0016] Among them, is the total radiation amount received on the surface of the photovoltaic module; is the direct radiation amount; is the solar incidence angle; is the diffuse radiation amount; is the ground albedo; is the installation angle of the photovoltaic module; is the ground-reflected radiation amount; The radiation transfer equation is solved by numerical calculation to obtain the radiation amount on the surface of the photovoltaic module at different times and under different weather conditions; Based on the three-dimensional model of the target partition, combined with the optimal arrangement of the photovoltaic modules in this partition, the shadow detection algorithm is used to calculate the shadow area of the photovoltaic modules at different times. According to the shadow analysis results, the effective surface area of the photovoltaic modules is calculated.
[0017] Optionally, the constructing a time series prediction model integrating a decomposition strategy and a swarm intelligence optimization algorithm includes: The historical photovoltaic power generation data is decomposed by variational mode decomposition to obtain a number of intrinsic mode components , , …, ; A time series prediction model is constructed. Its input layer receives the modal component data after variational mode decomposition. The hidden layer contains multiple time series feature extraction units, and the output layer outputs the prediction result; The hyperparameters of the time series prediction model are optimized by a swarm intelligence optimization algorithm. In each search iteration, the hyperparameters are updated through the performance feedback of the time series prediction network; For each modal component Divide it into a training set and a test set, and use the training set to train the time series prediction network optimized by the swarm intelligence optimization algorithm search. Adjust the parameters of the network through the backpropagation algorithm to minimize the fitness function. After training, use the test set to verify the model to obtain the predicted values of each modal component ; Integrate the predicted values of each modal component to obtain the predicted value of the photovoltaic power generation: .
[0018] Optionally, predicting the photovoltaic developable installed capacity and annual power generation potential of different partitions during the spatial form adjustment period of the building through a time series prediction model that fuses the decomposition strategy and the swarm intelligence optimization algorithm includes: Input the radiation amount and effective surface area on the surface of the photovoltaic module into the trained prediction model, combine with the conversion efficiency of the photovoltaic module, and consider the system loss to obtain the predicted value of the photovoltaic power generation of different partitions ; Predict the annual power generation potential based on the predicted value of the photovoltaic power generation:
[0019] wherein, is the number of time steps in a year; Calculate the photovoltaic developable installed capacity based on the predicted value of the annual power generation potential:
[0020] wherein, is the equivalent number of hours; is the conversion efficiency of the photovoltaic module; is the correction coefficient, and the value range is 0.75 - 0.80.
[0021] The second aspect of the present invention provides a building integrated photovoltaic power generation potential prediction system, based on the building integrated photovoltaic power generation potential prediction method described in the first aspect of the present invention. The system includes: A three-dimensional set area modeling and land use zoning module, which is used to integrate the spatial form prediction data, remote sensing images and GIS of the buildings in the set area, construct a three-dimensional set area digital twin model, and automatically identify the land use nature and divide the functional zones of the set area by using deep learning algorithms; A building area calculation module, which is used to calculate the total building area of the set area by region based on the building data in the three-dimensional model; A photovoltaic available area evaluation module, which is used to evaluate the building integrated photovoltaic available area by region and calculate the actual installation density; The photovoltaic optimization configuration module is used to determine the optimal installation angle and arrangement of photovoltaic modules through an optimization algorithm, and calculate the effective radiation amount, effective installation area and power generation efficiency per unit area on the surface of the modules; The photovoltaic power generation prediction module is used to predict the developable installed capacity and annual power generation potential of photovoltaics in different zones during the spatial form adjustment period of the building by using a time series prediction model that combines a decomposition strategy and a swarm intelligence optimization algorithm.
[0022] Compared with the prior art, the beneficial effects of the present invention at least include: 1. Provide a three-dimensional digital twin model of the set area by integrating the predicted data of the spatial form of the building in the set area, remote sensing images and GIS, and use a deep learning algorithm for land use nature identification, solving the problems of single data source of the three-dimensional model and low accuracy of land use nature identification in the prior art, and providing a high-precision spatial benchmark for the evaluation of photovoltaic power generation potential.
[0023] 2. The present invention provides a solution for determining the available area and installation density of building-integrated photovoltaics by comprehensively considering multi-dimensional influencing factors at the set area level. Combining the building density gradient, orientation distribution characteristics and dynamic shadow occlusion conditions, the available area and installation density of photovoltaics are determined by zone, solving the defect of insufficient quantification of multi-dimensional influencing factors (such as dynamic changes in building layout and local climate characteristics) in the prior art, and significantly improving the accuracy of photovoltaic layout planning.
[0024] 3. The present invention provides a time series prediction model that combines a decomposition strategy and a swarm intelligence optimization algorithm, solving the problems of low accuracy of the prior prediction model and inability to adapt to the dynamic adjustment of the building spatial form, and realizing high-precision prediction of photovoltaic power generation potential in a long-term evolution cycle. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] Figure 1 is a schematic flowchart of a method for predicting the potential of building-integrated photovoltaic power generation provided according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0026] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. The described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the spirit of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.
[0027] The present invention provides a method for predicting the potential of building-integrated photovoltaic power generation in Embodiment 1, including: Step 1: Integrate the spatial form prediction data, remote sensing images, and GIS of the buildings in the set area, construct a 3D digital twin model of the set area, use deep learning algorithms to identify the land use nature, and divide the functional areas including the building density gradient and building orientation distribution characteristics.
[0028] The set area is, for example but not limited to, an urban area taking a city as a unit, or a local area of a city, etc.
[0029] Preferably, Step 1 includes: Step 1.1: Collect the spatial form prediction data of the buildings in the set area; The spatial form prediction data of the buildings in the set area comes from, for example but not limited to, planning data.
[0030] Step 1.2: Collect and process remote sensing images. Obtain high-resolution remote sensing images of the set area through satellite remote sensing or aerial remote sensing, and perform radiometric correction on the remote sensing images of the set area. The radiometric correction formula is:
[0031] In the formula, is the radiance value after correction, and are the gain and offset parameters of the sensor respectively, is the pixel gray value of the original remote sensing image; Then, through the geometric correction method, using ground control points (GCPs), perform geometric fine correction on the corrected remote sensing image through the polynomial transformation formula. The polynomial transformation formula is:
[0032] In the formula, is the coordinate of the image after correction, is the coordinate of the original image, and are the polynomial coefficients, is the polynomial degree (generally taken as 2 - 3 times); Step 1.3: Integrate geographic information system (GIS) data. Collect the basic geographic information data of the set area, including terrain data (DEM), road network data, water system data, etc., import these data into the GIS platform, perform data format conversion and projection unification to ensure that all data is in the same geographic coordinate system for subsequent data fusion operations; Step 1.4: Import the preprocessed spatial form prediction data, remote sensing images, and GIS data of the buildings in the set area into 3D modeling software, use the data fusion algorithm to integrate data from different sources, and construct a 3D digital twin model of the set area; Step 1.5: Train a CNN model for land use type recognition based on the digital twin model of the three-dimensional set area. Input the preprocessed digital twin model of the three-dimensional set area into the trained CNN model to obtain the predicted results of the land use type for each area. Further preferably, the training of a CNN model for land use type recognition based on the digital twin model of the three-dimensional set area specifically includes: Construct a CNN model including multiple convolutional layers, pooling layers, and fully connected layers. The model structure includes: An input layer for receiving the feature map after fusing the preprocessed remote sensing images and GIS data; a convolutional layer using a 3x3 convolutional kernel, a stride of 1, a padding of 1, 16 convolutional kernels, and a ReLU activation function; a pooling layer using 2x2 max pooling with a stride of 2; a fully connected layer with the number of neurons equal to the number of land use type categories and a Softmax activation function. The formula is:
[0033] In the formula, is the input vector, is the number of categories, is the predicted probability of the category; Extract sample data from the constructed digital twin model of the three-dimensional set area, including regional images of different land use types and corresponding GIS attribute information. The land use types include residential land, commercial land, industrial land, public service land, etc. Label the sample data and divide it into a training set, a validation set, and a test set according to a ratio of 8:1:1. Use the training set to train the constructed CNN model, adopting the cross-entropy loss function:
[0034] Among them, is the number of samples, is the number of categories, is the sample belonging to the category true label (0 or 1), is the probability that the model predicts that the sample belongs to the category ; Select the Adam optimizer, and its parameters are, , , ; During the training process, continuously adjust the parameters of the model to minimize the loss function on the validation set and avoid overfitting.
[0035] Step 1.6, according to the land use nature recognition results, combined with the functional distribution characteristics in the spatial form prediction data of the buildings in the set area, use the clustering algorithm to perform clustering analysis on the set area, and divide the areas with the same or similar land use nature and functions into the same functional partition, including commercial areas, residential areas, industrial areas, etc. The functional partition includes the building density gradient and the building orientation distribution characteristics.
[0036] It should be noted that considering the problem that the multi-dimensional influencing factors at the set area level are not fully considered in the prior art, resulting in a deviation between the photovoltaic layout and the actual demand, the present invention divides the set area into partitions with different characteristics by dividing the functional partition including the building density gradient and the building orientation distribution characteristics, providing a differential calculation benchmark for the subsequent steps.
[0037] Step 2, based on the three-dimensional digital twin model of the set area constructed in Step 1, determine the total building area of the set area by partition.
[0038] Step 3, based on the total building area of the set area determined in Step 2, combined with the multi-dimensional influencing factors at the set area level, determine the available area of building integrated photovoltaic and the installation density of building integrated photovoltaic by partition.
[0039] Preferably, the Step 3 includes: Step 3.1, collect the relevant data at the set area level, including but not limited to information such as building layout, orientation, and distribution of surrounding obstacles; obtain the requirements for the sunlight exposure conditions for the installation of photovoltaic modules on the building surface, such as the minimum sunlight exposure duration, sunlight exposure angle, etc.; collect the construction control indicators of the set area, such as building density, floor area ratio, greening requirements, etc.; More preferably, the set area level includes the set area, the block, and the building monomer; Step 3.2, for different set area levels, analyze the influence degree of each dimension influencing factor on the available area of building integrated photovoltaic and the installation density; Exemplarily, at the set area level, mainly consider the influence of the overall topography, climate conditions, and spatial structure characteristics of the set area on photovoltaic utilization; at the block level, pay attention to the building layout, road orientation, and surrounding shading conditions within the block; at the building monomer level, focus on analyzing the orientation, height, and shape of the building; Step 3.3, according to the requirements for the sunlight exposure conditions for the installation of photovoltaic modules on the building surface, combined with the building density gradient and the building orientation distribution characteristics of each partition, calculate the area of the building that can install photovoltaic modules meeting the sunlight exposure conditions in each partition. For each partition, the following formula can be used to calculate the available installation area of building integrated photovoltaic:
[0040] Wherein, is the installable area for building - integrated photovoltaics, is the total area of building - integrated in this sub - area, is the correction factor considering sunlight conditions, and its value is determined according to the proportion of building - integrated that meets sunlight conditions in this sub - area, and can be obtained through sunlight simulation software or on - site measurement and statistics; Step 3.4: Adjust the installable area of building - integrated photovoltaics in combination with the construction control indicators of the set area to ensure that the photovoltaic installation complies with the technical specifications and actual space conditions of the set area.
[0041] Further preferably, the specific content of Step 3.4 includes: For each sub - area, the available area of building - integrated photovoltaics is calculated by the following formula:
[0042] where, is the available area of building - integrated photovoltaics; is the available coefficient of building - integrated photovoltaics. Considering that in addition to sunlight conditions, there may be other limiting factors, such as the load - bearing capacity of building - integrated structures and the need for maintenance channels, etc., this coefficient is determined according to the actual situations of different regions and different types of land use, and the value range is , and for each sub - area, it can be determined by methods such as expert evaluation, historical data statistics or experimental analysis the value of; is the correction factor considering regulatory planning indicators, and its value is determined according to the degree of restriction of regulatory planning indicators on photovoltaic installation in this sub - area, and can be obtained through simulation calculation.
[0043] It should be noted that when calculating the available area of building - integrated photovoltaics for different sub - areas, the available coefficient of building - integrated photovoltaics will be adjusted according to the building density gradient characteristics of different sub - areas. In high - density areas, the available coefficient of building - integrated photovoltaics will decrease, and vice versa.
[0044] Step 3.5: Determine the installation density of building - integrated photovoltaics for each sub - area. For each sub - area, the installation density of building - integrated photovoltaics is calculated by the following formula:
[0045] where, is the installation density of building - integrated photovoltaics, is the total power of photovoltaic modules that can actually be installed in this sub - area, is the available area of building - integrated photovoltaics in this sub - area.
[0046] Step 4: Based on the characteristics of solar resources, use an optimization algorithm to determine the optimal installation angle and arrangement of building-integrated photovoltaic modules, and respectively determine the radiation amount and effective surface area of the photovoltaic module surfaces in different zones.
[0047] Preferably, the said Step 4 includes: Step 4.1: Obtain the multi-dimensional characteristics of solar resources, and use wavelet analysis to process the collected solar radiation data; More preferably, the multi-dimensional characteristics of the solar resources can be obtained by comprehensively using ground meteorological stations, satellite remote sensing, and distributed solar monitoring sensor networks to obtain long-term solar resource data of the target area.
[0048] More preferably, the solar resource data includes minute-by-minute change data of the total solar radiation amount, direct radiation amount, diffuse radiation amount, sunshine hours, solar altitude angle, azimuth angle, as well as relevant environmental parameters such as atmospheric transparency, aerosol optical thickness, and ground albedo.
[0049] At the same time, in combination with Geographic Information System (GIS) data, obtain information such as the topography and landforms, distribution and height of surrounding buildings in the target area.
[0050] More preferably, the use of wavelet analysis to process the collected solar radiation data includes: the original solar radiation time series is , select Daubechies wavelet for discrete wavelet transform, and decompose into components of different scales and frequencies:
[0051] wherein, is the detail coefficient, reflecting the high-frequency part of the signal and embodying the short-term fluctuations of solar radiation; is the approximation coefficient, reflecting the low-frequency part of the signal and embodying the long-term trend of solar radiation; is the wavelet basis function, is the scaling function; By analyzing the components of different scales, extract the change characteristics and laws of solar resources.
[0052] Step 4.2: Determine the optimal installation angle of the building-integrated photovoltaic modules through a hybrid intelligent optimization algorithm. Taking the maximum solar radiation amount received by the photovoltaic module surface as the goal, considering the solar radiation characteristics obtained by wavelet analysis, establish an objective function:
[0053] wherein, is at time , and the solar azimuth angle is , the altitude angle is , the PV installation angle is , the solar radiation intensity received on the surface of the PV module; is a weight function based on the wavelet analysis result, which is used to highlight the time periods with significant changes in solar radiation.
[0054] Step 4.3, adopt a hybrid intelligent optimization algorithm combining the genetic algorithm (GA) and the simulated annealing algorithm (SA) to determine the optimal installation angle of the PV module .
[0055] Specifically, the genetic algorithm is used to search for the potential optimal solution space globally, and the simulated annealing algorithm is used to finely optimize the solution obtained by the genetic algorithm locally to avoid falling into the local optimal solution. During the algorithm execution process, the algorithm parameters are dynamically adjusted according to the spatio-temporal variation characteristics of solar energy resources, such as the crossover probability and mutation probability of the genetic algorithm, the initial temperature and cooling rate of the simulated annealing algorithm, etc. Through multiple rounds of iterative calculations, the optimal installation angle of the PV module is determined .
[0056] Step 4.4, considering multi-factor coupling, use the ant colony algorithm to differentially optimize the arrangement of PV modules in different zones.
[0057] Further preferably, in the step 4.4, considering multi-factor coupling includes: multi-factors such as the geometric shape of the building, the bearing capacity of the building structure, the shadow occlusion of surrounding buildings and obstacles, the size of the PV module and the electrical connection requirements, etc.
[0058] Further preferably, in the step 4.4, using the ant colony algorithm to optimize the arrangement of PV modules includes: Abstract the arrangement problem of PV modules into a path search problem of ants on a graph structure. Ants select the placement positions of the modules according to the pheromone concentration and heuristic information during the search process, and at the same time consider the limitations of multi-factor coupling. Through the collaborative search of multiple ants and the update iteration of pheromone, gradually find the optimal arrangement method to achieve the efficient utilization of the building facade space and the minimization of shadow occlusion.
[0059] It should be noted that when using the ant colony algorithm to optimize the arrangement of PV modules for the multi-factor coupling limitations in different zones, it also includes the differentiation of pheromone rules. Exemplarily, for the commercial area, the pheromone deposition strategy is to reduce pheromone deposition in high-obstacle areas to guide avoiding the shadow area and set a more stringent component spacing threshold; for the residential area, the pheromone deposition strategy is a uniform deposition strategy to promote regular arrangement and relax the spacing constraint; for the industrial area, the pheromone deposition strategy is to enhance deposition along the main beam structure to increase the installation density.
[0060] Step 4.5: Based on the solar resource data obtained in Step 4.1, construct a radiative transfer model to calculate the radiation amount on the surface of the photovoltaic module.
[0061] Further preferably, Step 4.5 includes: According to the collected solar resource data and atmospheric parameters, determine the parameters in the radiative transfer model. Adopt a radiative transfer model considering multiple atmospheric scattering and absorption, such as the discrete ordinate method (DISORT) or the two-stream approximation model, and combine with the atmospheric parameters (such as atmospheric composition, aerosol distribution, etc.) of the target area to calculate the propagation process of solar radiation in the atmosphere; Considering the reflection, refraction, and absorption characteristics of the surface of the photovoltaic module, establish a radiation energy balance equation for the surface of the photovoltaic module:
[0062] Wherein, is the total radiation amount received on the surface of the photovoltaic module; is the direct radiation amount; is the solar incidence angle; is the scattered radiation amount; is the ground albedo; is the installation angle of the photovoltaic module; is the ground-reflected radiation amount.
[0063] Use numerical calculation methods to solve the radiative transfer equation to obtain the radiation amount on the surface of the photovoltaic module at different times and under different weather conditions.
[0064] Step 4.6: Calculate the effective surface area of the surface of the photovoltaic module based on 3D modeling and shadow analysis; Further preferably, Step 4.6 specifically includes: Based on the 3D model of the target area, combined with the optimal arrangement of the photovoltaic modules, use a shadow detection algorithm to calculate the shadow area of the photovoltaic modules at different times. According to the shadow analysis results, calculate the ineffective surface area of the photovoltaic modules, and obtain the effective surface area by subtracting the ineffective surface area from the total area of the photovoltaic modules.
[0065] Step 5: According to the relevant parameters of the photovoltaic modules, construct a time series prediction model integrating the decomposition strategy and the swarm intelligence optimization algorithm to predict the photovoltaic developable installed capacity and annual power generation potential in different partitions during the spatial form adjustment period of the building.
[0066] Preferably, Step 5 specifically includes: Step 5.1: Decompose the photovoltaic power generation data by variational mode decomposition (VMD) Further preferably, Step 5.1 includes: (1)VMD Parameter Selection a. Determine the bandwidth constraint parameter, which is used to control the bandwidth of each mode; the noise tolerance, which is used to process the noise in the data; the number of modes, that is, the number of intrinsic mode components into which the data is to be decomposed; the initialization method can be selected as random initialization, etc.; the convergence tolerance, which is used to judge whether the algorithm converges.
[0067] b. Data Acquisition and Preprocessing Obtain the historical power generation power time series data of the photovoltaic module , and this time series data also contains the effective area of the photovoltaic module, the radiation amount on the surface of the photovoltaic module, and the operating efficiency of the photovoltaic system corresponding to the moment. Preprocess the data, including operations such as removing outliers and filling missing values, and then perform normalization processing to map the data to the interval:
[0068] where and are the minimum and maximum values of the original data respectively.
[0069] c. VMD Decomposition Input the preprocessed data into the VMD algorithm; Specifically, the goal of VMD is to find intrinsic mode functions (IMFs) such that the following objective function is minimized:
[0070] where is the central frequency of each mode, is the Dirac function. By iteratively solving this objective function, each mode , that is, the intrinsic mode component is obtained.
[0071] d. Result Visualization Plot the original photovoltaic power generation data and each decomposed mode to visually display the data decomposition situation, which is convenient for analysis and understanding.
[0072] (2)VMD Decomposition Use the VMD with the determined parameters to decompose the photovoltaic power generation data, and decompose the non-stationary and non-linear photovoltaic power generation data into several intrinsic mode components , , …, .
[0073] Step 5.2, construct a time series prediction model integrating swarm intelligence optimization algorithm; Further preferably, the said Step 5.2 includes: Exemplarily, the sparrow search algorithm is adopted as the swarm intelligence optimization algorithm, and the LSTM network is used as the time series prediction model framework; a. Sparrow search algorithm (SSA) Initialize the sparrow population, and set the population size as , and the position of each sparrow represents a set of parameters of the LSTM network, such as weights and biases; The fitness function of the sparrow is defined as the mean square error between the predicted value and the true value, that is:
[0074] where, is the number of samples, is the predicted value, is the true value.
[0075] Specifically, sparrows are divided into producers and followers during the search process. Producers are responsible for searching for food globally (i.e., finding better parameters), and followers follow producers to obtain food. At the same time, a vigilant mechanism is introduced. When sparrows perceive danger, they will adjust their positions. By iteratively updating the positions of sparrows, the parameters of the LSTM network are continuously optimized.
[0076] b. Construct an LSTM neural network Construct an LSTM neural network, which consists of an input layer, a hidden layer, and an output layer. The input layer receives the modal component data after VMD decomposition. The hidden layer contains multiple LSTM units, which can learn the long-term dependencies of the data, and the output layer outputs the prediction result; For each modal component , it is divided into a training set and a test set. The training set is used to train the LSTM network optimized by SSA, and the parameters of the network are adjusted through the backpropagation algorithm to minimize the fitness function. After training, the test set is used to verify the model to obtain the predicted values of each modal component.
[0077] c. Integration of prediction results Integrate the predicted values of each modal component to obtain the predicted value of the photovoltaic power generation:
[0078] Step 5.3: Input the radiation amount and effective surface area on the surface of the photovoltaic module determined in Step 4 into the trained time series prediction model that integrates swarm intelligence optimization algorithms, and considering the conversion efficiency of the photovoltaic module and system losses, predict the photovoltaic installable capacity and annual power generation potential in different zones during the spatial form adjustment period of the building. Further preferably, Step 5.3 includes: Input the radiation amount and effective surface area on the surface of the photovoltaic module determined in Step 4 into the trained time series prediction model that integrates swarm intelligence optimization algorithms, and combine with the conversion efficiency of the photovoltaic module and the operating efficiency of the photovoltaic system to obtain the predicted value of the photovoltaic power generation. ; Predict the annual power generation potential based on the predicted value of the photovoltaic power generation:
[0079] where is the number of time steps in a year; Calculate the photovoltaic installable capacity based on the predicted value of the annual power generation potential:
[0080] where is the equivalent hour number, which is the same as the annual radiation amount value. Exemplarily, when the annual radiation amount is 1500 , the equivalent hour number is 1500 hours; is the conversion efficiency of the photovoltaic module; is the correction coefficient, which comprehensively considers the influence of various actual factors such as inverter efficiency, cable loss, dust occlusion, and shadow on power generation, and its value range is 0.75 - 0.80.
[0081] In Embodiment 2 of the present invention, a building - integrated photovoltaic power generation potential prediction system is provided. Based on the building - integrated photovoltaic power generation potential prediction method described in Embodiment 1, the system includes: A three - dimensional set - area modeling and land - use zoning module, which is used to integrate the spatial form prediction data, remote - sensing images, and GIS of the buildings in the set area, construct a three - dimensional digital twin model of the set area, and automatically identify the land use nature using deep - learning algorithms to divide the functional zones of the set area; A building - integrated photovoltaic area calculation module, which is used to calculate the total building area of the set area by region (such as functional zones or administrative zones) based on the building data in the three - dimensional model, and support dynamic update and visual display; A photovoltaic available area evaluation module, which is used to evaluate the photovoltaic available area of the building - integrated photovoltaic by zone and calculate the actual installation density; The photovoltaic optimization configuration module is used to determine the optimal installation angle and arrangement of photovoltaic modules by region through an optimization algorithm, and calculate the effective radiation amount, effective installation area, and power generation efficiency per unit area on the surface of the modules. The photovoltaic power generation prediction module is used to predict the developable installed capacity and annual power generation potential of photovoltaics in different regions during the spatial form adjustment period of the building by using a time series prediction model that combines a decomposition strategy and a swarm intelligence optimization algorithm.
[0082] Further preferably, the photovoltaic optimization configuration module is built-in with a partition type discriminator, which automatically loads the corresponding algorithm parameter set for different partitions.
[0083] The present disclosure may be a system, a method, and / or a computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions thereon for causing a processor to implement various aspects of the present disclosure.
[0084] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: the specific implementation manners of the present invention can still be modified or equivalently replaced, and any modification or equivalent replacement that does not depart from the spirit and scope of the present invention shall be covered by the protection scope of the claims of the present invention.
Claims
1. A method for predicting building integrated photovoltaic power generation potential, characterized in that: The steps include: Integrate the spatial morphology prediction data, remote sensing images and GIS of the buildings in the set area, build a three-dimensional digital twin model of the set area, use deep learning algorithms to identify the nature of the land use, and divide the functional zones including the building density gradient and building orientation distribution characteristics; Based on the three-dimensional digital twin model of the set area, the total building area of the set area is determined by zoning; based on the total building area of the set area, combined with the multi-dimensional influencing factors of the set area level, the building integrated photovoltaic usable area and building integrated photovoltaic installation density are determined by zoning; Based on the characteristics of solar energy resources, the optimization algorithm is used to determine the best installation angle and arrangement of building-integrated photovoltaic modules, and the radiation amount and effective surface area of photovoltaic modules in different partitions are determined respectively; According to the conversion efficiency of photovoltaic modules and the radiation and effective surface area of the photovoltaic modules, a time series prediction model integrating decomposition strategy and swarm intelligence optimization algorithm is constructed to predict the exploitable photovoltaic installed capacity and annual power generation potential of different partitions during the spatial form adjustment period of the building.
2. A building integrated photovoltaic power generation potential prediction method according to claim 1, characterized in that: The constructing of the three-dimensional setting area digital twin model includes: Collect and pre-process the spatial form prediction data of buildings in the set area; Obtain high-resolution remote sensing images of the set area through satellite remote sensing or aerial remote sensing, and perform radiation correction on the original remote sensing images through sensor gain parameters and offset parameters; Collect and integrate basic geographic information data of the set area, including terrain data, road network data and water system data, and import these data into the GIS platform for data format conversion and projection unification; The pre-processed spatial morphology prediction data, remote sensing images and GIS data of the set area buildings are imported into the 3D modeling software, and the data fusion algorithm is used to integrate the data from different sources to construct a 3D digital twin model of the set area.
3. A building integrated photovoltaic power generation potential prediction method according to claim 2, characterized in that: The use of deep learning algorithms to identify the nature of land use and divide functional zones includes: Based on the three-dimensional set area digital twin model, a CNN model for land use nature identification is trained, and the preprocessed three-dimensional set area digital twin model is input into the trained CNN model to obtain the land use nature prediction results of each area; According to the results of land use nature identification and combined with the functional distribution characteristics in the spatial morphology prediction data of the buildings in the set area, a clustering algorithm is used to perform clustering analysis on the set area, and areas with the same land use nature and function are divided into the same functional zones, including commercial areas, residential areas and industrial areas.
4. A building integrated photovoltaic power generation potential prediction method according to claim 3, characterized in that: The zoning determines the building integrated photovoltaic usable area and building integrated photovoltaic installation density including: Collect relevant data at the set area level, including building layout, orientation and distribution of surrounding obstructions, obtain the sunshine conditions for installing photovoltaic modules on the building surface, including minimum sunshine duration and sunshine angle, and collect construction control indicators for the set area, including building density, volume ratio and greening requirements; Analyze the influence of various factors on the available area and installation density of building-integrated photovoltaics at different set regional levels; According to the sunshine conditions for the installation of photovoltaic modules on the building surface, combined with the building density gradient and building orientation distribution characteristics of each zone, the area of photovoltaic modules that can be installed in each zone that meets the sunshine conditions is calculated. For each zone, the building integrated photovoltaic installation area, building integrated photovoltaic utilization coefficient and the correction coefficient considering the control planning indicators are used to calculate the building integrated photovoltaic utilization area; For each zone, the building-integrated photovoltaic installation density is calculated based on the building-integrated photovoltaic utilization area in the zone and the total power of photovoltaic modules that can actually be installed in the zone.
5. A building integrated photovoltaic power generation potential prediction method according to claim 4, characterized in that: The method of determining the optimal installation angle and arrangement of building-integrated photovoltaic modules by using an optimization algorithm based on solar energy resource characteristics includes: Obtain multi-dimensional characteristics of solar energy resources, including minute-by-minute changes in total solar radiation, direct radiation, diffuse radiation, sunshine hours, solar altitude angle, and azimuth angle, and use wavelet analysis to process the collected solar radiation data; The optimal installation angle of building-integrated photovoltaic modules is determined by a hybrid intelligent optimization algorithm. The maximum amount of solar radiation received by the surface of the photovoltaic modules is taken as the goal. Considering the solar radiation characteristics obtained by wavelet analysis, the objective function is established: in, For at the moment , the solar azimuth is , the altitude angle is , Photovoltaic installation angle is The solar radiation intensity received by the surface of the photovoltaic module at ; It is a weight function based on the wavelet analysis results; A hybrid intelligent optimization algorithm combining genetic algorithm and simulated annealing algorithm is used to determine the optimal installation angle of photovoltaic modules ; Considering the coupling of multiple factors, including the building's geometry, the building's structural bearing capacity, the shadows cast by surrounding buildings and obstacles, the size of the photovoltaic modules, and the electrical connection requirements, the ant colony algorithm is used to optimize the arrangement of the photovoltaic modules.
6. A building-integrated photovoltaic power generation potential prediction method according to claim 5, characterized in that: The processing of the collected solar radiation data by wavelet analysis includes: The original solar radiation time series is , select Daubechies wavelet for discrete wavelet transform, Decompose into components of different scales and frequencies: in, is the detail coefficient, reflecting the high-frequency part of the signal; is the approximate coefficient, reflecting the low-frequency part of the signal; is the wavelet basis function, is the scaling function; By analyzing components at different scales, the changing characteristics and laws of solar energy resources can be extracted.
7. A method for predicting building-integrated photovoltaic power generation potential according to claim 6, characterized in that: The step of respectively determining the radiation amount and the effective surface area of the photovoltaic module surface comprises: Based on the acquired solar energy resource data, a radiation transfer model is constructed to calculate the radiation on the surface of the photovoltaic module. According to the collected solar energy resource data and atmospheric parameters, various parameters in the radiation transfer model are determined. Considering the reflection, refraction and absorption characteristics of the photovoltaic module surface, the radiation energy balance equation on the photovoltaic module surface is established: in, is the total radiation received by the surface of the photovoltaic module; is the amount of direct radiation; is the solar incidence angle; is the amount of scattered radiation; is the ground albedo; The installation angle of the photovoltaic module; is the radiation reflected from the ground; The radiation transfer equation is solved by numerical calculation to obtain the radiation amount on the surface of the photovoltaic module at different times and under different weather conditions; Based on the three-dimensional model of the target partition and the optimal arrangement of photovoltaic modules in the partition, the shadow area of the photovoltaic modules at different times is calculated using the shadow detection algorithm. According to the shadow analysis results, the effective surface area of the photovoltaic modules is calculated.
8. A method for predicting building-integrated photovoltaic power generation potential according to claim 7, characterized in that: The time series prediction model for constructing the fusion decomposition strategy and the swarm intelligence optimization algorithm includes: The historical photovoltaic power generation data is decomposed through variational mode decomposition to obtain several intrinsic mode components. , , …, ; Construct a time series prediction model, whose input layer receives the modal component data after variational modal decomposition, the hidden layer contains multiple time series feature extraction units, and the output layer outputs the prediction results; The hyperparameters of the time series prediction model are optimized through the swarm intelligence optimization algorithm. In each search iteration, the hyperparameters are updated through the performance feedback of the time series prediction network. For each modal component , divide it into training set and test set, use the training set to train the time series prediction network after the swarm intelligence optimization algorithm searches and optimizes it, and adjust the parameters of the network through the back propagation algorithm to minimize the fitness function; after the training is completed, use the test set to verify the model and obtain the predicted values of each modal component ; The predicted values of each modal component After integration, the predicted value of photovoltaic power generation is obtained: 。 9. A method for predicting building-integrated photovoltaic power generation potential according to claim 8, characterized in that: The time series prediction model that integrates the decomposition strategy and the swarm intelligence optimization algorithm predicts the photovoltaic developable installed capacity and annual power generation potential of different partitions during the spatial form adjustment period of the building, including: The radiation amount and effective surface area of the photovoltaic module surface are input into the trained prediction model, combined with the conversion efficiency of the photovoltaic module and considering the system loss, to obtain the predicted value of photovoltaic power generation in different partitions. ; The annual power generation potential is predicted based on the predicted value of photovoltaic power generation: in, is the number of time steps in a year; Calculate the developable installed capacity of photovoltaic power based on the predicted annual power generation potential: in, is the equivalent number of hours; is the conversion efficiency of the photovoltaic module; is the correction coefficient, and its value range is 0.75-0.
80.
10. A building-integrated photovoltaic power generation potential prediction system, based on a building-integrated photovoltaic power generation potential prediction method according to any one of claims 1 to 9, characterized in that: The system includes: The three-dimensional set area modeling and land use zoning module is used to integrate the spatial morphology prediction data, remote sensing images and GIS of the buildings in the set area, build a three-dimensional digital twin model of the set area, use deep learning algorithms to automatically identify the nature of the land use, and divide the set area into functional zones; A building area calculation module is used to calculate the total building area of a set area by region based on the building data in the three-dimensional model; The photovoltaic available area assessment module is used to assess the building-integrated photovoltaic available area by zoning and calculate the actual installation density; Photovoltaic optimization configuration module, which is used to determine the best installation angle and arrangement of photovoltaic modules through optimization algorithms, and calculate the effective radiation amount on the module surface, the effective installation area and the power generation efficiency per unit area; The photovoltaic power generation prediction module is used to integrate the time series prediction model of the decomposition strategy and the swarm intelligence optimization algorithm to predict the photovoltaic developable installed capacity and annual power generation potential of different partitions during the spatial form adjustment period of the building.
Citation Information
Patent Citations
Determining method and device of optimal installing inclination angle of photovoltaic module
CN106372346A
Building roof photovoltaic panel arrangement method
CN113221222A
Photovoltaic power generation potential and economy evaluation method based on GF7 remote sensing image
CN114881399A
Photovoltaic recommendation method and device based on photovoltaic identification and clustering, and storage medium
CN116307512A
Photovoltaic power prediction method based on VMD-PSO-LSTM
CN118886449A
Cited By
Photovoltaic integrated energy storage method and system
CN120414691A
