Optimization method and device for distributed photovoltaic system

Abstract

The invention discloses a distributed photovoltaic system optimization method and device, and the method comprises the steps: determining a heat island intensity index through the temperature part data of a target city, and calculating the solar radiation intensity and reflected light distribution according to a three-dimensional building model; the photovoltaic panel temperature and the system power generation efficiency of the distributed photovoltaic system are calculated through the temperature data and the solar radiation intensity, a photovoltaic panel installation strategy is determined, a photovoltaic panel heat balance model is constructed according to the installation strategy, key influence factors are determined through the heat balance model to optimize the installation strategy, and the photovoltaic system is adjusted according to the optimized strategy. And re-optimizing the installation strategy according to the system heat island intensity and the reflected light distribution to obtain an optimized photovoltaic system deployment strategy. The system optimization strategy is generated through the optimization method, the power generation efficiency of the photovoltaic system deployment strategy is improved, the photovoltaic panel temperature caused by the heat island effect and building reflection radiation is reduced, and intelligent management and efficient operation of the photovoltaic power generation system are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic system optimization, and in particular to an optimization method and device for a distributed photovoltaic system. Background Art

[0002] When deploying distributed photovoltaic systems in urban environments, the urban heat island effect and light reflected from building walls are two key factors that together affect the power generation efficiency of photovoltaic panels. The urban heat island effect is caused by the absorption and storage of heat by man-made structures such as buildings and roads in cities, resulting in higher temperatures in the central areas of cities than in surrounding rural areas. This increase in temperature will have various effects on photovoltaic systems, including: increased operating temperature of photovoltaic panels, reduced power generation efficiency, increased radiation intensity, or local overheating of photovoltaic panels.

[0003] On the other hand, the walls of high-rise buildings in cities can reflect sunlight, and this reflected light increases the radiation intensity on the surface of photovoltaic panels. In theory, the increased radiation intensity should increase the power generation of photovoltaic panels, but in reality, due to the indirect nature of reflected light, the light received by photovoltaic panels may not be at the best angle, so it may not be effectively converted into electrical energy. At the same time, reflected light may also cause local overheating of photovoltaic panels, affecting their long-term stability and power generation efficiency.

[0004] Therefore, in order to solve the problems caused by the above two factors, the existing technology mostly uses photovoltaic power prediction technology to reduce the uncertainty of photovoltaic output power. However, as the prediction area increases and the number of photovoltaic power stations increases, the prediction calculation efficiency of the traditional prediction model method gradually decreases, and the traditional prediction model does not take into account the influence of meteorological factors in the power station area, and the meteorological characteristics of photovoltaic power stations may vary greatly, which will also have a certain impact on the accuracy of regional power prediction. Summary of the invention

[0005] The present invention provides a distributed photovoltaic system optimization method and device to achieve the technical effect of improving the efficiency and stability of the deployed photovoltaic system, while also ensuring the efficient operation and long-term operation stability of the photovoltaic system.

[0006] In order to solve the above technical problems, an embodiment of the present invention provides an optimization method for a distributed photovoltaic system, comprising the following steps:

[0007] Acquire temperature distribution data of the target city, and calculate the temperature distribution data by an interpolation analysis method to obtain a heat island intensity index;

[0008] Acquire a three-dimensional building model of the target city, and calculate the solar radiation intensity and reflected light distribution according to the three-dimensional building model;

[0009] Determine the photovoltaic panel temperature and system power generation efficiency of the distributed photovoltaic system according to the temperature distribution data and the solar radiation intensity, and output a first photovoltaic panel installation strategy;

[0010] Calculating and determining the radiation intensity of the photovoltaic panels based on the first photovoltaic panel installation strategy and the three-dimensional building model, and constructing a photovoltaic panel thermal balance model based on the radiation intensity of the photovoltaic panels;

[0011] Determine key influencing factors according to the photovoltaic panel thermal balance model, optimize and adjust the first photovoltaic panel installation strategy according to the key influencing factors, output and optimize and adjust the distributed photovoltaic system according to the second photovoltaic panel installation strategy;

[0012] The actual operation data of the distributed photovoltaic system after optimization and adjustment is obtained, a multivariate regression model is constructed according to the actual operation data, the multivariate regression model is optimized and adjusted according to the heat island intensity index and the reflected light distribution, and the optimized distributed photovoltaic system deployment strategy is output.

[0013] The optimization method provided by the present invention first performs spatial interpolation analysis on the temperature distribution data obtained from the target city through an interpolation analysis method to obtain the city's heat island intensity index, thereby determining the heat island intensity change trend of various areas of the target city at different times and seasons, thereby determining the degree of heat island effect in various areas of the city, providing a decision-making basis for urban thermal environment management, and also providing reference data for the subsequent output city's photovoltaic panel installation strategy.

[0014] After determining the city's heat island intensity index, the system will obtain the city's three-dimensional building model and use it to calculate the city's solar radiation intensity and reflected light distribution, providing reference data for subsequent optimization and adjustment of the first photovoltaic panel installation strategy.

[0015] Through the temperature distribution data and solar radiation intensity of the target city, the system can calculate the photovoltaic panel temperature and system power generation efficiency of the distributed photovoltaic system in the city, and determine the first photovoltaic panel installation strategy through the temperature distribution data and solar radiation intensity, that is, the first photovoltaic panel installation strategy obtained after considering various factors such as system power generation efficiency and photovoltaic panel installation environment, providing a data basis for subsequent optimization.

[0016] After obtaining the first photovoltaic panel installation strategy, the system adjusts the installation position of the photovoltaic panels in the three-dimensional building model according to the strategy, and determines the radiation intensity of the photovoltaic panels based on the adjusted model calculation. That is, the direct solar radiation and the reflected radiation on the building surface at different solar altitudes and azimuths are determined through model simulation, so as to determine the radiation intensity on the photovoltaic panels, and construct a photovoltaic panel thermal balance model based on the determined photovoltaic panel radiation intensity, thereby improving the accuracy and reliability of the model in predicting the temperature changes and power generation performance of the photovoltaic panels.

[0017] The constructed photovoltaic panel thermal balance model can further determine the key factors affecting the photovoltaic panel temperature and photovoltaic power generation efficiency, namely the key influencing factors, and readjust the weight parameters of the first photovoltaic panel installation strategy based on the determined key influencing factors. The strategy is readjusted with the photovoltaic panel temperature change and power generation efficiency as the optimization goals to obtain the second photovoltaic panel installation strategy.

[0018] After obtaining the second photovoltaic panel installation strategy, the distributed photovoltaic system can be adjusted through the strategy, and the strategy can be further optimized according to the actual operation data of the photovoltaic system, including the heat island intensity index and the distribution of reflected light, to obtain the optimized photovoltaic system optimization strategy, which improves the power generation efficiency of the optimized photovoltaic system deployment strategy, and also reduces the temperature of the photovoltaic panels caused by the urban heat island effect and building reflected radiation, thereby improving the overall efficiency of the photovoltaic power generation system. At the same time, by identifying the power generation mode and efficiency characteristics of the photovoltaic system under different environmental conditions, and determining the correlation between the actual power generation efficiency and environmental factors, the deployment strategy of the photovoltaic system is optimized, and the intelligent management and efficient operation of the photovoltaic power generation system are also realized.

[0019] As a preferred example, the step of obtaining the temperature distribution data of the target city and calculating the temperature distribution data by an interpolation analysis method to obtain the heat island intensity index includes:

[0020] Acquire initial temperature distribution data of the target city, perform outlier elimination processing and missing value supplement processing on the initial temperature distribution data, and obtain the temperature distribution data;

[0021] Performing spatial interpolation analysis on the temperature distribution data by using the Kriging interpolation method to obtain a temperature distribution surface of the target city;

[0022] The heat island intensity index is calculated based on the temperature distribution surface, and the target city is classified into a regional heat island effect level based on the heat island intensity index to obtain a regional heat island level.

[0023] By identifying outliers and supplementing missing values ​​in the initial temperature distribution data, the quality and integrity of the temperature distribution data are improved. By spatially interpolating the temperature distribution data using the Kriging interpolation method, the temperature distribution surface in different seasons in the city, i.e., the temperature distribution map, can be obtained. For example, the temperature distribution comparison map in winter and summer more intuitively shows the heat island effect of various regions in the target city. By calculating the heat island intensity index of each region in the target city through the temperature distribution map and classifying the heat island effect level of each region, the degree to which each region in the target city is affected by the heat island effect can be determined. Users can also be prompted with areas that are severely affected, providing a decision-making basis for urban thermal environment management.

[0024] As a preferred example, the obtaining of the three-dimensional building model of the target city and calculating the solar radiation intensity and the distribution of reflected light according to the three-dimensional building model include:

[0025] Acquire building point cloud data of the target city, and sequentially perform denoising, point cloud classification and three-dimensional reconstruction on the building point cloud data to obtain the three-dimensional building model;

[0026] The solar radiation data of the target city and the reflectivity data of the building surface materials are obtained, the solar radiation data and the reflectivity data are input into the three-dimensional building model, and the solar radiation intensity and the reflected light distribution of the building surface are calculated by a Monte Carlo simulation algorithm.

[0027] A three-dimensional building model of the city is constructed based on the city’s building point cloud data, which provides a data basis for the subsequent acquisition of the city’s solar radiation intensity and the distribution of reflected light on the surface of urban buildings. By obtaining solar radiation data and building surface reflectivity data and constructing a three-dimensional building model, the system can use the Monte Carlo simulation algorithm to simulate the reflection and scattering process of sunlight on the surface of urban buildings, thereby determining the radiation brightness and radiation flux distribution of each building, and then determining the distribution of reflected light on the surface of each building, as well as the solar radiation intensity obtained by calculating various radiation intensities, providing reference data for the subsequent calculation of photovoltaic panel installation strategies.

[0028] As a preferred example, the photovoltaic panel temperature and system power generation efficiency of the distributed photovoltaic system are determined based on the temperature distribution data and the solar radiation intensity, and a first photovoltaic panel installation strategy is output, specifically:

[0029] Inputting the temperature distribution data and the solar radiation intensity into a preset Sandia array performance model for calculation and processing to obtain the photovoltaic panel temperature and the system power generation efficiency;

[0030] Acquire building density data and green coverage data of the target city, and construct a photovoltaic array multi-objective optimization model based on the building density data, the green coverage data, the photovoltaic panel temperature, and the system power generation efficiency;

[0031] The photovoltaic array multi-objective optimization model is optimized and solved by a genetic algorithm, and the first photovoltaic panel installation strategy is output.

[0032] The temperature distribution data and solar radiation intensity are processed through the Sandia array performance model. While ensuring the spatiotemporal continuity and consistency of the temperature data and providing ambient temperature boundary conditions for subsequent photovoltaic array performance simulations, the solar radiation intensity data is used to comprehensively consider factors such as the material properties, electrical parameters and installation conditions of the photovoltaic panels, thereby obtaining the photovoltaic panel temperature and the power generation efficiency of the photovoltaic system, thereby improving the accuracy and reliability of the obtained data.

[0033] After obtaining the above data, the system will also obtain the city's building density data and green coverage data to ensure that the subsequently generated photovoltaic panel installation strategy fully considers the shading effect of greening on photovoltaic panel radiation. The photovoltaic array multi-objective optimization model constructed based on the above data not only comprehensively considers various factors such as system power generation, green building shading and landscape coordination, but also further optimizes it through genetic algorithms, thereby improving the practicality, reliability and accuracy of the output photovoltaic panel installation strategy.

[0034] As a preferred example, the calculating and determining the photovoltaic panel radiation intensity based on the first photovoltaic panel installation strategy and the three-dimensional building model, and constructing a photovoltaic panel thermal balance model based on the photovoltaic panel radiation intensity, includes:

[0035] Adjusting parameters of the three-dimensional building model according to the first photovoltaic panel installation strategy, and simulating the reflection and shielding of sunlight by Monte Carlo ray tracing method to obtain the radiation flux per unit area of ​​the photovoltaic panel;

[0036] Constructing a corresponding building surface radiation transmission mathematical model according to the radiation flux, and calculating the photovoltaic panel radiation intensity of the photovoltaic panel through the building surface radiation transmission mathematical model;

[0037] The thermal balance model of the photovoltaic panel is constructed according to the radiation intensity of the photovoltaic panel and the heat exchange process of the photovoltaic panel.

[0038] After adjusting the parameters of the three-dimensional building model through the output of the first photovoltaic panel installation strategy, the system will once again simulate the reflection and shielding of sunlight through the Monte Carlo ray tracing method to determine the solar radiation intensity per unit area of ​​each photovoltaic panel after the placement of the photovoltaic panels is adjusted, that is, the radiation flux, and construct the corresponding mathematical model of radiation transmission on the building surface through the determined radiation flux to further clarify the propagation and distribution process of radiation energy on the building surface.

[0039] The radiation intensity of the photovoltaic panel can be calculated by constructing a transmission mathematical model. According to the radiation intensity of the photovoltaic panel and the heat exchange process of the photovoltaic panel, a thermal balance model of the photovoltaic panel can be constructed. The heat transfer mechanisms including solar radiation absorption, photoelectric conversion, convection heat transfer and thermal radiation are considered, and the mathematical relationship between the photovoltaic panel temperature and environmental conditions, material properties, structural parameters and other factors is established. In this way, a thermal balance model of the photovoltaic panel is constructed, which enables the model to accurately predict the temperature changes and power generation performance of the photovoltaic panel.

[0040] As a preferred example, determining key influencing factors according to the photovoltaic panel thermal balance model, and optimizing and adjusting the first photovoltaic panel installation strategy according to the key influencing factors, includes:

[0041] The photovoltaic panel temperature and the power generation efficiency of the system are simulated and analyzed by the photovoltaic panel thermal balance model to obtain a corresponding change curve;

[0042] Screening the influencing factors of the change curve by using a sensitivity analysis algorithm to obtain the key influencing factors;

[0043] Then, the first photovoltaic panel installation strategy is optimized and adjusted in parameters based on the key influencing factors through a genetic algorithm.

[0044] By simulating the temperature and power generation efficiency of photovoltaic panels through the constructed thermal balance model, the curves of the change of photovoltaic panel temperature and power generation efficiency over time can be obtained. By analyzing the above curves, various factors that may affect the temperature and power generation efficiency of photovoltaic panels can be screened out. The most significant factors affecting the temperature and efficiency of photovoltaic panels, such as ambient temperature, wind speed or conversion efficiency, can be screened out through the sensitivity analysis algorithm. After determining the key factors, the system will make secondary optimization adjustments to the installation strategy of photovoltaic panels based on the key factors, further improving the power generation efficiency of photovoltaic panels while reducing their temperature impact.

[0045] As a preferred example, the obtaining of actual operation data of the distributed photovoltaic system after optimization and adjustment, and constructing a multivariate regression model according to the actual operation data, includes:

[0046] Acquiring the actual operation data of the distributed photovoltaic system, and performing dimensionality reduction and feature extraction processing on the actual operation data to obtain a photovoltaic feature data set;

[0047] A cluster analysis is performed on the photovoltaic characteristic data set, the photovoltaic characteristic data set is divided into different clusters according to preset power generation indicators, and photovoltaic power generation influencing factors are determined according to the clusters, and then the multivariate regression model is constructed according to the photovoltaic power generation influencing factors and power generation efficiency.

[0048] After the secondary adjustment of the photovoltaic panel installation strategy is determined, the system can adjust the distributed photovoltaic system according to it, so as to obtain the actual operation data of the adjusted photovoltaic system, and obtain the photovoltaic feature data set by performing data dimension reduction and feature extraction on the actual operation data of the photovoltaic system. Then, the photovoltaic feature data set is clustered to explore the power generation law and efficiency characteristics of photovoltaic panels under different environmental conditions, and to construct a corresponding multivariate regression model to represent the quantitative relationship between power generation efficiency and environmental factors.

[0049] As a preferred example, the multivariate regression model is optimized and adjusted according to the heat island intensity index and the reflected light distribution, and the optimized distributed photovoltaic system optimization strategy is output, including:

[0050] Adjusting the parameters of the multivariate regression model according to the heat island intensity index and the reflected light distribution, predicting photovoltaic power generation efficiency through the adjusted model, and outputting photovoltaic prediction data;

[0051] The photovoltaic prediction data is compared with the actual photovoltaic data, and the model prediction error is evaluated on the comparison result by a deviation analysis method, and the model parameters are adjusted by feedback based on the evaluation result;

[0052] The second photovoltaic panel installation strategy is optimized through the feedback-adjusted model to obtain the distributed photovoltaic system deployment strategy.

[0053] After obtaining the multivariate regression model, the system will adjust the model parameters through the heat island intensity index and the distribution of reflected light, so that the output prediction data takes into account factors such as the city's heat island intensity and reflected light distribution, and predict the power generation efficiency of the photovoltaic system through the adjusted model. The prediction data and the actual power generation efficiency data of the photovoltaic system, that is, the actual photovoltaic data, are compared to determine the error between the model and the actual photovoltaic system, and the error is analyzed and evaluated to determine the source of the error. Then, the model is feedback adjusted according to the source of the error to further improve the practicality, reliability and accuracy of the model.

[0054] As a preferred example, the method of evaluating the model prediction error of the comparison result by using the deviation analysis method, and performing feedback adjustment on the model parameters according to the evaluation result, includes:

[0055] Attributing and decomposing the comparison results by the deviation analysis method to obtain error source analysis, and performing parameter correction on the model according to the error source analysis;

[0056] An uncertainty analysis is performed on the comparison results to obtain corresponding sources of uncertainty, and then parameters of the model are adjusted according to the sources of uncertainty.

[0057] The error sources are determined by the deviation analysis method, and the model parameters are corrected according to the error sources, that is, the factors that have a greater impact on the prediction error are determined, so as to achieve dynamic parameter adjustment of the model, so that the model can adapt to the actual operating status and environmental conditions of the photovoltaic system. Through the uncertainty analysis of the error, the uncertainty sources in the model parameters, input data and model structure are clarified, the reliability and risk of the model prediction are evaluated, and the model and algorithm are further optimized, which improves the robustness and adaptability of the model prediction, realizes the continuous iteration and upgrading of the model, and also improves the scientificity and practicality of the simulation prediction of photovoltaic power generation efficiency.

[0058] Accordingly, this embodiment further provides an optimization device for a distributed photovoltaic system, the optimization device comprising a heat island intensity determination module, a building model calculation module, a first strategy output module, a radiation intensity calculation module, a second strategy adjustment module and a photovoltaic system optimization module; wherein:

[0059] The heat island intensity determination module is used to obtain temperature distribution data of the target city, and calculate the temperature distribution data through an interpolation analysis method to obtain a heat island intensity index;

[0060] The building model calculation module is used to obtain the three-dimensional building model of the target city, and calculate the solar radiation intensity and the reflected light distribution according to the three-dimensional building model;

[0061] The first strategy output module is used to calculate and determine the photovoltaic panel temperature and system power generation efficiency of the distributed photovoltaic system according to the temperature distribution data and the solar radiation intensity, and output a first photovoltaic panel installation strategy;

[0062] The radiation intensity calculation module is used to calculate and determine the radiation intensity of the photovoltaic panel based on the first photovoltaic panel installation strategy and the three-dimensional building model, and to construct a photovoltaic panel thermal balance model based on the photovoltaic panel radiation intensity;

[0063] The second strategy adjustment module is used to determine the key influencing factors according to the photovoltaic panel thermal balance model, optimize and adjust the first photovoltaic panel installation strategy according to the key influencing factors, and output and optimize and adjust the distributed photovoltaic system according to the second photovoltaic panel installation strategy;

[0064] The photovoltaic system optimization module is used to obtain the actual operation data of the distributed photovoltaic system after optimization and adjustment, construct a multivariate regression model based on the actual operation data, optimize and adjust the multivariate regression model according to the heat island intensity index and the reflected light distribution, and output the optimized distributed photovoltaic system optimization strategy. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1:A flow chart of an embodiment of the distributed photovoltaic system optimization method provided by the present invention;

[0066] Figure 2 : A structural diagram of an embodiment of an optimization device for a distributed photovoltaic system provided by the present invention. DETAILED DESCRIPTION

[0067] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0068] Embodiment 1

[0069] Please refer to Figure 1 , which is a flow chart of an embodiment of the distributed photovoltaic system optimization method provided by the present invention, including steps 101 to 106, each step is specifically as follows:

[0070] Step 101: Acquire temperature distribution data of a target city, and calculate the temperature distribution data by an interpolation analysis method to obtain a heat island intensity index.

[0071] The optimization method provided in the embodiment of the present invention first performs spatial interpolation analysis on the temperature distribution data obtained from the target city through an interpolation analysis method to obtain the city's heat island intensity index, so as to determine the heat island intensity change trend of various regions of the target city at different times and seasons, so as to determine the degree of heat island effect in various regions of the city, provide a decision-making basis for urban thermal environment management, and provide reference data for the subsequent output city's photovoltaic panel installation strategy.

[0072] In this embodiment, the step of acquiring the temperature distribution data of the target city and calculating the temperature distribution data by an interpolation analysis method to obtain the heat island intensity index includes:

[0073] Acquire initial temperature distribution data of the target city, perform outlier elimination processing and missing value supplement processing on the initial temperature distribution data, and obtain the temperature distribution data;

[0074] Performing spatial interpolation analysis on the temperature distribution data by using the Kriging interpolation method to obtain a temperature distribution surface of the target city;

[0075] The heat island intensity index is calculated based on the temperature distribution surface, and the target city is classified into a regional heat island effect level based on the heat island intensity index to obtain a regional heat island level.

[0076] By identifying outliers and supplementing missing values ​​in the initial temperature distribution data, the quality and integrity of the temperature distribution data are improved. By spatially interpolating the temperature distribution data using the Kriging interpolation method, the temperature distribution surface in different seasons in the city, i.e., the temperature distribution map, can be obtained. For example, the temperature distribution comparison map in winter and summer more intuitively shows the heat island effect of various regions in the target city. By calculating the heat island intensity index of each region in the target city through the temperature distribution map and classifying the heat island effect level of each region, the degree to which each region in the target city is affected by the heat island effect can be determined. Users can also be prompted with areas that are severely affected, providing a decision-making basis for urban thermal environment management.

[0077] In this embodiment, the initial temperature distribution data acquired by the system includes temperature data of different time periods and seasons, and the acquired data needs to cover a sufficiently large area and time span to facilitate subsequent comprehensive spatial interpolation analysis. After the initial temperature distribution data is obtained, the acquired temperature data will also be preprocessed, including checking the integrity and accuracy of the data, removing outliers and supplementing missing values, and preferably interpolating and smoothing the data to improve data quality.

[0078] In this embodiment, when obtaining the temperature data of the target city, temperature data of different time and space scales can be collected through a variety of observation methods such as meteorological stations, automatic weather stations or mobile observation stations. For example: 100 automatic weather stations are deployed in the central area of ​​the target city, and each station records temperature data every 10 minutes. After continuous observation for one year, temperature distribution data with high time and space resolution is obtained. When preprocessing the acquired temperature data, the box plot method can be used to identify outliers, and data points that exceed 1.5 times the interquartile range of the upper and lower quartiles are regarded as outliers and removed. For missing values, time series interpolation methods such as linear interpolation, spline interpolation, etc. can be used to estimate the missing values ​​based on the temperature data before and after the missing values.

[0079] When the system provided in this embodiment uses the Kriging interpolation method to perform spatial interpolation analysis on the preprocessed temperature data, the system will establish the best linear unbiased estimation model based on the spatial distribution and temperature value of the data points, specifically: select a suitable variogram model, such as an exponential model, a Gaussian model or a spherical model, determine the model parameters through variogram fitting, calculate the spatial correlation and weight coefficient between the data points, and finally obtain a continuous temperature distribution surface in the central area of ​​the city. Based on the interpolation analysis results, the system can also generate temperature distribution maps of the central area of ​​the city in different time periods and seasons, and intuitively present the spatiotemporal distribution characteristics of the urban heat island effect in a visual way.

[0080] When using the Kriging interpolation method for spatial interpolation analysis, you can choose an exponential variogram model, fit the variogram using the least squares method, and determine parameters such as the nugget value, sill value, and range. Based on the fitted variogram, calculate the covariance matrix between the data points, and then calculate the Kriging weight coefficient to obtain the temperature estimate at any location. For the interpolation analysis results, you can generate a temperature distribution map of the city center area with a 1km×1km grid, and draw temperature distribution maps for different seasons, such as a temperature distribution comparison map for winter and summer.

[0081] As for calculating the heat island intensity index, this embodiment quantifies the intensity of the heat island effect by calculating the average temperature difference between the central area of ​​the city and the suburbs; or uses the calculation method of the heat island intensity index (Heat Island Intensity Index, HII) to quantify it, that is, HII = (Tu-Tr) / Tr, where Tu is the average temperature of the urban area and Tr is the average temperature of the suburbs. The above method can be used to analyze the changing trend of the heat island intensity at different times and seasons. After determining the heat island intensity index of each area in the city, the system will also divide the degree of heat island effect of each area in the city according to the classification standard of the heat island intensity index. This embodiment preferably divides the heat island intensity index into four levels: weak, medium, strong and extremely strong, so as to determine the high temperature area seriously affected by the heat island effect and provide a decision-making basis for urban thermal environment governance.

[0082] When calculating the heat island intensity index, 10 representative areas in the city center can be selected, their average temperatures can be calculated respectively, and then compared with the average temperatures of 10 reference areas in the suburbs to calculate the corresponding HII index. For example, when the average temperature in the city center is 35.5°C and the average temperature in the suburbs is 32.1°C, then HII = (35.5-32.1) / 32.1 = 0.106, indicating that the city has a strong heat island effect. According to the classification standard of the heat island intensity index, the area with HII>0.10 is classified as an extremely strong heat island area, which will prompt that the area urgently needs to take cooling measures, so as to improve the efficiency of the city's cooling decision-making.

[0083] Step 102: Acquire a three-dimensional building model of the target city, and calculate the solar radiation intensity and reflected light distribution according to the three-dimensional building model.

[0084] After determining the city's heat island intensity index, the system will obtain the city's three-dimensional building model and use it to calculate the city's solar radiation intensity and reflected light distribution, providing reference data for subsequent optimization and adjustment of the first photovoltaic panel installation strategy.

[0085] Specifically, the step of acquiring the three-dimensional building model of the target city and calculating the solar radiation intensity and the distribution of reflected light according to the three-dimensional building model in this embodiment includes:

[0086] Acquire building point cloud data of the target city, and sequentially perform denoising, point cloud classification and three-dimensional reconstruction on the building point cloud data to obtain the three-dimensional building model;

[0087] The solar radiation data of the target city and the reflectivity data of the building surface materials are obtained, the solar radiation data and the reflectivity data are input into the three-dimensional building model, and the solar radiation intensity and the reflected light distribution of the building surface are calculated by a Monte Carlo simulation algorithm.

[0088] A three-dimensional building model of the city is constructed based on the city's building point cloud data, which provides a data basis for the subsequent acquisition of the city's solar radiation intensity and the distribution of reflected light on the surface of urban buildings. By obtaining solar radiation data and building surface reflectivity data and constructing a three-dimensional building model, the system can simulate and track the reflection and scattering process of sunlight on the surface of urban buildings through the Monte Carlo simulation algorithm, so as to determine the radiation brightness and radiation flux distribution of each building, and then determine the distribution of reflected light on the surface of each building, as well as the solar radiation intensity obtained by calculating various radiation intensities, providing reference data for the subsequent calculation of photovoltaic panel installation strategies.

[0089] In this embodiment, the building point cloud data of the city can be obtained by using laser radar scanning, drone photogrammetry and other technologies to collect high-precision building point cloud data, and noise removal, point cloud classification and three-dimensional reconstruction are performed through point cloud data processing software to generate a three-dimensional model of the city building, that is, the three-dimensional building model described in this embodiment, including the geometric shape, spatial position and attribute information of the building. The system can extract key parameters such as the height, orientation and spacing of the building from the three-dimensional building model, and the absolute height and relative height of the building can be calculated by using the three-dimensional space analysis method, and the orientation of the building can be determined according to the normal vector of the building's outer wall. The spacing between adjacent buildings can be calculated by the shortest distance algorithm between buildings to form a building parameter database.

[0090] When acquiring urban 3D building model data, high-precision laser radar scanners such as Leica ScanStation P50 can be used to perform dense point cloud scanning of urban buildings, with a point cloud density of more than 1,000 points per square meter. Then, point cloud processing software such as Autodesk ReCap can be used to remove noise points through statistical filtering algorithms, extract building point clouds using segmentation algorithms based on geometric features, and then generate 3D grid models of buildings using Poisson surface reconstruction algorithms.

[0091] For the constructed 3D building model, the system can use the rasterization method to calculate the building height, specifically: generate a building height raster layer with a spatial resolution of 1 meter, and use the MATLAB tool parameter ComputerVisionToolbox to calculate the main orientation of the building's exterior wall through the normal vector clustering algorithm to obtain the angular distribution of the building orientation. For the spacing between adjacent buildings, the ArcGIS Near tool can be used to calculate the minimum Euclidean distance between each building and the adjacent building, and then generate a distance matrix for the building spacing.

[0092] After obtaining the three-dimensional building model of the city, the system will further obtain solar radiation data in the city area, including solar altitude angle, azimuth and direct radiation intensity, and use the Perez model to calculate the solar radiation intensity at different times and locations, taking into account the influence of factors such as atmospheric transmittance, cloud cover and sun position, to obtain the distribution of solar radiation intensity at different locations and times above the city.

[0093] When calculating the solar radiation intensity, the system can use the gendaylit module of the RADIANCE software, based on the Perez sky brightness distribution model, set the longitude and latitude coordinates, time zone and time parameters of the city, and input hourly meteorological parameters such as total cloud cover, low cloud cover, etc., to simulate and calculate the sky radiation brightness distribution. Then use the rtrace tool of RADIANCE to generate radiation sampling points above the city with a spatial step of 1 meter, calculate the direct solar radiation intensity and sky scattered radiation intensity received by each sampling point, and obtain the solar radiation intensity distribution map of the city at different locations and times.

[0094] As for the collection of reflectivity data of building surface materials, spectral measurements will be performed on common building materials such as glass, concrete, and paint to obtain the reflectivity curves of materials at different wavelengths, establish a building material reflectivity database, and provide material parameters for subsequent ray tracing simulations.

[0095] When calculating the reflectivity of common building exterior wall materials, you can use a ground object spectrometer such as ASDFieldSpec4 to measure the spectral reflectivity curve of the material in the visible light to near-infrared band (350-2500nm), and use the Beckman-Kirchhoff model to spectrally correct the measured data to obtain the standard reflectivity curve of the material.

[0096] After obtaining the above two data including solar radiation data and reflectivity data of building surface materials, they are combined with the three-dimensional building model and input into the Monte Carlo ray tracing simulation software, that is, the Monte Carlo simulation algorithm is used to calculate them as described in this embodiment. Specifically, through professional software such as Radiance and TracePro, the parameters of ray tracing such as the number of rays, the number of reflections and the energy threshold are set to simulate the propagation and reflection process of sunlight in the urban building environment. By setting the parameters of light emission, crossover and reception, the reflection and scattering process of light on the building surface is simulated, the radiation brightness and radiation flux of the surface are calculated, and the reflected light intensity and directional distribution at different positions on the building surface are obtained. The ray tracing simulation results are statistically analyzed, the reflected radiation data of the building wall are extracted, and the average radiation intensity, maximum radiation intensity and total radiation energy at different times and orientations are calculated to generate the time series curve diagram and spatial distribution diagram of the reflected radiation, that is, the solar radiation intensity and the reflected light distribution described in this embodiment.

[0097] In the Monte Carlo ray tracing simulation, this embodiment preferably uses the rcontrib tool of RADIANCE, sets the number of rays to 10,000,000, the number of reflections to 5, the energy threshold to 0.001, and inputs the reflectivity curve of the building material as the material parameter to simulate the reflection and scattering process of sunlight on the building surface, and obtains the radiant brightness and radiant flux distribution at each position on the building surface. Then, using the StatisticsandMachineLearningToolbox of MATLAB, statistical analysis is performed on the reflected radiation of the building wall, and indicators such as the average radiation intensity, maximum radiation intensity, and total radiation energy of the wall are calculated, and an interpolation algorithm is used to generate a spatiotemporal distribution diagram of the reflected radiation.

[0098] Step 103: Calculate and determine the photovoltaic panel temperature and system power generation efficiency of the distributed photovoltaic system according to the temperature distribution data and the solar radiation intensity, and output a first photovoltaic panel installation strategy.

[0099] Through the temperature distribution data and solar radiation intensity of the target city, the system can calculate the photovoltaic panel temperature and system power generation efficiency of the distributed photovoltaic system in the city, and determine the first photovoltaic panel installation strategy through the temperature distribution data and solar radiation intensity, that is, the first photovoltaic panel installation strategy obtained after considering various factors such as system power generation efficiency and photovoltaic panel installation environment, providing a data basis for subsequent optimization.

[0100] Specifically, the present embodiment calculates and determines the photovoltaic panel temperature and system power generation efficiency of the distributed photovoltaic system according to the temperature distribution data and the solar radiation intensity, and outputs the first photovoltaic panel installation strategy, specifically:

[0101] Inputting the temperature distribution data and the solar radiation intensity into a preset Sandia array performance model for calculation and processing to obtain the photovoltaic panel temperature and the system power generation efficiency;

[0102] Acquire building density data and green coverage data of the target city, and construct a photovoltaic array multi-objective optimization model based on the building density data, the green coverage data, the photovoltaic panel temperature, and the system power generation efficiency;

[0103] The photovoltaic array multi-objective optimization model is optimized and solved by a genetic algorithm, and the first photovoltaic panel installation strategy is output.

[0104] The temperature distribution data and solar radiation intensity are processed through the Sandia array performance model. While ensuring the spatiotemporal continuity and consistency of the temperature data and providing ambient temperature boundary conditions for subsequent photovoltaic array performance simulations, the solar radiation intensity data is used to comprehensively consider factors such as the material properties, electrical parameters and installation conditions of the photovoltaic panels, thereby obtaining the photovoltaic panel temperature and the power generation efficiency of the photovoltaic system, thereby improving the accuracy and reliability of the obtained data.

[0105] After obtaining the above data, the system will also obtain the city's building density data and green coverage data to ensure that the subsequently generated photovoltaic panel installation strategy fully considers the shading effect of greening on photovoltaic panel radiation. The photovoltaic array multi-objective optimization model constructed based on the above data not only comprehensively considers various factors such as system power generation, green building shading and landscape coordination, but also further optimizes it through genetic algorithms, thereby improving the practicality, reliability and accuracy of the output photovoltaic panel installation strategy.

[0106] As another example of obtaining temperature distribution data in this embodiment, the system can use high spatiotemporal resolution temperature distribution data of the city center area, utilize multi-source data such as thermal infrared remote sensing and ground meteorological stations, and use spatiotemporal interpolation and data assimilation methods such as Kriging interpolation and Kalman filtering to generate a temperature distribution grid dataset in the three-dimensional space of the city, ensure the spatiotemporal continuity and consistency of the temperature data, and provide ambient temperature boundary conditions for subsequent photovoltaic array performance simulation, i.e., the temperature distribution data obtained in this embodiment.

[0107] When obtaining urban temperature distribution data, the system can use thermal infrared remote sensing images such as Landsat8TIRS to extract surface temperature information with a spatial resolution of 100 meters and a temporal resolution of 16 days. At the same time, the temperature observation data of the city's meteorological station is used every hour to obtain the temperature distribution of a 1000-meter grid through Kriging interpolation. The ensemble Kalman filter algorithm is used to fuse the surface temperature inverted by remote sensing and the temperature interpolated by the meteorological station to obtain three-dimensional temperature distribution data that is continuous in time and space.

[0108] The temperature distribution data and solar radiation intensity are processed through the Sandia array performance model. Specifically, the Sandia photovoltaic array performance model is used to comprehensively consider factors such as the material properties, electrical parameters and installation conditions of the photovoltaic panels, and the operating temperature and output power of the photovoltaic panels at different installation angles and positions are calculated. Combined with the inverter efficiency and system loss, the power generation efficiency curve of the photovoltaic array is obtained.

[0109] In this embodiment, the "PVLib" photovoltaic array performance calculation model proposed by Sandia National Laboratories is preferably used, and the photovoltaic components are set to monocrystalline silicon cells with a rated power of 320W and an efficiency of 19.1%. The photovoltaic panel temperature and output power curves at different inclination angles (0-90 degrees) and azimuth angles (0-360 degrees) are calculated.

[0110] After obtaining the building density data and green coverage data of the target city, the system will also use the area ratio method and viewshed analysis method to calculate the available area and shading factor of the roof and wall of the building, extract the geometric information of the building, calculate the available area of ​​the roof and wall, and consider the shading effect of greening on the radiation of photovoltaic panels. In this embodiment, a green coverage rate of 40% is preferably considered to correct the effective radiation intensity of the photovoltaic array.

[0111] A photovoltaic array multi-objective optimization model is constructed based on the above data, and the first photovoltaic panel installation strategy is output, specifically: a multi-objective function for photovoltaic array layout optimization is constructed, factors such as maximizing power generation, minimizing building shading, and landscape coordination are comprehensively considered, and a genetic algorithm is used for multi-objective optimization solution. The key elements such as the encoding method of decision variables, the construction of fitness functions, the design of genetic operators (selection, crossover, mutation) and parameter tuning, and the judgment of convergence conditions are clarified. Combined with the characteristics of photovoltaic layout optimization, a specific algorithm flow and pseudo code are given to obtain an optimized configuration scheme that satisfies the photovoltaic panel installation angle, orientation, and density under different weight combinations, that is, the first photovoltaic panel installation strategy in this embodiment.

[0112] In this embodiment, three objective functions for photovoltaic layout optimization are preferably constructed, namely annual power generation, building shading ratio and landscape coordination. The NSGA-II multi-objective genetic algorithm is used, the population number is set to 100, the crossover probability is 0.8, the mutation probability is 0.1, and it is iterated 500 times to obtain the Pareto optimal solution set. Generate multiple combinations of initial installation angles and positions as the initial population, and calculate the power generation efficiency and photovoltaic panel temperature of each combination, and then evaluate its fitness according to the objective function, and select the combination with better performance according to the fitness for crossover and mutation operations. Generate a new generation of combinations through crossover and mutation operations, and output the optimal installation angle and position when the set number of iterations is reached or the fitness is no longer significantly improved.

[0113] Step 104: Calculate and determine the radiation intensity of the photovoltaic panels based on the first photovoltaic panel installation strategy and the three-dimensional building model, and construct a photovoltaic panel thermal balance model based on the photovoltaic panel radiation intensity.

[0114] After obtaining the first photovoltaic panel installation strategy, the system adjusts the installation position of the photovoltaic panels in the three-dimensional building model according to the strategy, and determines the radiation intensity of the photovoltaic panels based on the adjusted model calculation. That is, the direct solar radiation and the reflected radiation on the building surface at different solar altitudes and azimuths are determined through model simulation, so as to determine the radiation intensity on the photovoltaic panels, and construct a photovoltaic panel thermal balance model based on the determined photovoltaic panel radiation intensity, thereby improving the accuracy and reliability of the model in predicting the temperature changes and power generation performance of the photovoltaic panels.

[0115] Specifically, the present embodiment calculates and determines the radiation intensity of the photovoltaic panel based on the first photovoltaic panel installation strategy and the three-dimensional building model, and constructs a photovoltaic panel thermal balance model based on the photovoltaic panel radiation intensity, including:

[0116] Adjusting parameters of the three-dimensional building model according to the first photovoltaic panel installation strategy, and simulating the reflection and shielding of sunlight by Monte Carlo ray tracing method to obtain the radiation flux per unit area of ​​the photovoltaic panel;

[0117] Constructing a corresponding building surface radiation transmission mathematical model according to the radiation flux, and calculating the photovoltaic panel radiation intensity of the photovoltaic panel through the building surface radiation transmission mathematical model;

[0118] The thermal balance model of the photovoltaic panel is constructed according to the radiation intensity of the photovoltaic panel and the heat exchange process of the photovoltaic panel.

[0119] After adjusting the parameters of the three-dimensional building model through the output of the first photovoltaic panel installation strategy, the system will once again simulate the reflection and shielding of sunlight through the Monte Carlo ray tracing method to determine the solar radiation intensity per unit area of ​​each photovoltaic panel after the placement of the photovoltaic panels is adjusted, that is, the radiation flux, and construct the corresponding mathematical model of radiation transmission on the building surface through the determined radiation flux to further clarify the propagation and distribution process of radiation energy on the building surface.

[0120] The radiation intensity of the photovoltaic panel can be calculated by constructing a transmission mathematical model. According to the radiation intensity of the photovoltaic panel and the heat exchange process of the photovoltaic panel, a thermal balance model of the photovoltaic panel can be constructed. The heat transfer mechanisms including solar radiation absorption, photoelectric conversion, convection heat transfer and thermal radiation are considered, and the mathematical relationship between the photovoltaic panel temperature and environmental conditions, material properties, structural parameters and other factors is established. In this way, a thermal balance model of the photovoltaic panel is constructed, which enables the model to accurately predict the temperature changes and power generation performance of the photovoltaic panel.

[0121] In this embodiment, when the system adjusts the parameters of the three-dimensional building model according to the first photovoltaic panel installation strategy, it includes using the equatorial coordinate system model of the sun's position according to the geographic coordinates and time parameters of the area where the building is located, determining the position of the sun on the celestial sphere through parameters such as solar right ascension and declination, and then calculating the altitude angle and azimuth of the sun relative to the horizon according to the longitude and latitude and time zone of the building's location. The main calculation formulas include solar hour angle, declination angle, altitude angle, azimuth angle, etc., and the influence of the earth's revolution needs to be considered, and correction items such as the mean time difference and the ecliptic obliquity are introduced to obtain the incident angle and azimuth of the sun relative to the building surface at different times.

[0122] In the case where photovoltaic panels are installed on the facade of a building, this embodiment will first obtain a three-dimensional model of the building facade. When obtaining the three-dimensional model of the building facade, drone photogrammetry or laser scanning can be used to collect point cloud data on the building surface with a spatial resolution of 0.1 meters, and point cloud data processing software such as RealityCapture or CloudCompare is used to remove noise, segment point clouds, and reconstruct triangular meshes to obtain a three-dimensional solid model of the building facade. For a building with a height of 50 meters, a width of 30 meters, and a length of 80 meters, a mesh model consisting of 2 million triangular facets can be generated, and the size of each triangular facet is about 0.5-1 square meters. Using the Ladybug plug-in, the sun position at different times on June 22, 2022 is calculated based on the longitude and latitude coordinates of the building location (such as Beijing, 116.3° east longitude, 39.9° north latitude) and the UTC+8 time zone. According to the WGS84 ellipsoid parameters and the Meeus algorithm, the solar right ascension, declination and hour angle were calculated, and the atmospheric refraction correction of -0.8° and the solar correction of ±8 minutes were taken into account. The solar altitude angles at 9:00, 12:00, 15:00 and 18:00 were 28.3°, 73.2°, 52.1° and 7.6° respectively, and the azimuths were 74.5° east of south, due south, 63.8° west of south and 86.4° west of north.

[0123] The Monte Carlo ray tracing method is used to simulate the reflection and shielding of sunlight and obtain the radiation flux per unit area of ​​the photovoltaic panel. Specifically, the Monte Carlo ray tracing method is used to simulate the reflection and shielding process of sunlight on the building surface. Starting from the sunlight source, a large amount of light is emitted to the building surface according to the direction and intensity of direct solar radiation. At the same time, in order to balance the calculation accuracy and efficiency, methods such as layered sampling or importance sampling are used to project more light in areas with large changes in radiation intensity and less light in areas with small changes. Parallel computing technologies such as GPU acceleration are also used to improve the calculation efficiency of ray tracing. When the light intersects with the building surface, the reflection direction and intensity attenuation of the light are calculated according to the optical parameters such as reflectivity, absorptivity and roughness of the surface material, and the next reflection is performed until the light is absorbed or leaves the building scene. In the ray tracing process, the radiation shielding of the photovoltaic panel at different positions on the building facade is judged by the intersection test between the light and the building surface. Algorithms such as ray intersection with triangular mesh and octree space partition are used to accelerate the intersection test between the light and the building surface.

[0124] In the Monte Carlo ray tracing, this embodiment preferably adopts the ImportanceSampling algorithm, generates 10^7 rays in the solid angle space according to the importance distribution of the BSDF reflectivity function, and uses the Embree ray tracing kernel and GPU parallel calculation to achieve real-time intersection and multiple reflections of the light and the scene. For a photovoltaic panel with a size of 1m×1.5m, 100 candidate installation positions are divided from a height of 10m-40m and a spacing of 2m on the south facade of the building. Using the octree algorithm, 10^4 rays are projected for visibility analysis at each position, and the cumulative values ​​of direct solar radiation and sky diffuse radiation during the period of 10:00-14:00 are counted. The results show that the annual average radiation at position 12 is the highest, which is 1280kWh / m^2, but there are 20% and 35% occlusion rates during the periods of 9:00-10:00 and 16:00-17:00.

[0125] For each candidate installation location, the number and intensity of direct and reflected light falling on the photovoltaic panel surface are counted, and the radiation flux per unit area of ​​the photovoltaic panel is calculated by taking into account the inclination and azimuth of the photovoltaic panel itself. For the location of the photovoltaic panel that is blocked, its cumulative blocking time and blocking ratio in a day are calculated as the constraint conditions for the optimization of the installation location.

[0126] According to the radiation flux, a mathematical model of radiation transmission on the corresponding building surface is constructed, which specifically includes considering multiple radiation sources such as direct solar radiation, scattered radiation from the sky, and reflected radiation from the building surface, and adopts the radiation transmission equation to describe the propagation and distribution process of radiation energy on the building surface. At the same time, this embodiment will also combine the statistical results of ray tracing to calculate the total radiation intensity change curve of each photovoltaic panel installation position at different times of the day. Taking the maximum average radiation per unit area of ​​the photovoltaic panel as the objective function, and the geometric constraints and aesthetic requirements of the building facade as the constraints, a genetic algorithm (GA) or a particle swarm optimization algorithm (PSO) is used to optimize the search for the installation angle and orientation of the photovoltaic panel.

[0127] Preferably, this embodiment adopts the radiation transfer equation and the finite element discretization method to establish a numerical model of radiation transfer on the building surface, and solves it in combination with the Radiance software to obtain the radiation intensity distribution of each photovoltaic panel position at different times.

[0128] Specifically, the optimization model is constructed with the inclination and azimuth of the photovoltaic panel as optimization variables, the annual average radiation as the objective function, and the installation area, shading ratio, aesthetics, etc. as constraints. The GA algorithm continuously updates the population and searches for the optimal solution through operations such as selection, crossover, and mutation; the PSO algorithm searches for the optimal particle in the solution space by updating the position and speed of the particle.

[0129] Preferably, this embodiment uses the NSGA-II multi-objective genetic algorithm to optimize the installation angle and orientation of photovoltaic panels, with annual power generation, radiation uniformity and shading rate as optimization targets, installation angle 0°-90°, orientation angle -90°-90° as constraint variables, setting the population size to 50, crossover probability to 0.8, and mutation probability to 0.1, and iterating 200 generations to obtain the Pareto optimal solution set. After weighing the three objectives, the optimal installation angle is 30°, the optimal orientation angle is 15° south-west, the annual power generation is increased by 5%, and the shading rate is reduced by 10%.

[0130] In addition, this embodiment will also reasonably set algorithm parameters, such as population size, evolutionary generations, crossover probability, mutation probability, etc., and perform parameter sensitivity analysis and algorithm performance evaluation to obtain the optimal installation position and angle of photovoltaic panels on the building facade. The optimized photovoltaic panel installation parameters are applied to the design of building photovoltaic integration, a three-dimensional visualization model combining photovoltaic panels and building facades is generated, and light and shadow rendering and energy balance analysis are performed for the subsequent construction of photovoltaic panel thermal balance models. According to the rated power and efficiency parameters of the photovoltaic panels, the power generation and benefits of the photovoltaic system in different seasons and weather conditions are estimated to provide data support for the decision-making of the building photovoltaic integration plan. At the same time, a sensitivity analysis is performed on the installation position and angle of the photovoltaic panels to evaluate the impact of changes in building height-width ratio, spacing, surface materials and other parameters on photovoltaic power generation performance, optimize the building design parameters, and maximize the comprehensive benefits of building photovoltaic integration.

[0131] Preferably, this embodiment builds a parametric design model in the Rhino platform, embeds the optimization results into the building facade, generates a three-dimensional model of photovoltaic panels and building integration, and introduces the DIVA plug-in in Grasshopper to analyze the hourly power generation and daily power generation of typical days in different seasons. After optimization, the peak power of the photovoltaic system is 80kW, and the daily power generation on the summer solstice, vernal equinox and winter solstice is 540kWh, 380kWh and 210kWh respectively, which is 8%-12% higher than before optimization. The analysis shows that for every 0.1 increase in the building aspect ratio, the annual average radiation of the facade decreases by 1.5%; for every 1 meter reduction in the building spacing, the photovoltaic panel shielding rate increases by 2%; the facade material changes from light color to dark color, and the reflected radiation is reduced by 30%. Therefore, the introduction of photovoltaic optimization analysis in the architectural design stage can provide architects with a scientific and quantitative design basis and promote the promotion and application of building photovoltaic integration.

[0132] A thermal balance model of photovoltaic panels is constructed based on the radiation intensity of photovoltaic panels and the heat exchange process of photovoltaic panels. Specifically, it includes: considering the heat exchange process between photovoltaic panels and the surrounding environment, including solar radiation absorption, photoelectric conversion, convection heat transfer, thermal radiation and other heat transfer mechanisms, and establishing the mathematical relationship between photovoltaic panel temperature and environmental conditions, material properties, structural parameters and other factors.

[0133] At the same time, the system will also collect typical annual meteorological data for the location of the photovoltaic panels, including hourly ambient temperature, wind speed, solar radiation intensity and other meteorological parameters as input conditions for energy simulation; obtain the technical parameters of the photovoltaic panels, such as photoelectric conversion efficiency, surface roughness, heat capacity, thermal conductivity, etc., as boundary conditions and material property parameters of the thermal balance model; and use energy simulation software to calibrate and verify the parameters of the photovoltaic panel thermal balance model. By comparing with the measured data, the empirical coefficients and key parameters in the model, such as convection heat transfer coefficient, radiation absorption rate, etc., are adjusted to enable the model to accurately predict the temperature changes and power generation performance of the photovoltaic panels.

[0134] Preferably, when constructing the thermal balance model of photovoltaic panels, this embodiment can select the Sandia photovoltaic array performance model and related components in the EnergyPlus software, set the geometric size, orientation, inclination and other parameters of the photovoltaic panels, define the photoelectric conversion efficiency curve, temperature coefficient and other material properties of the silicon cells, establish convection, radiation and other heat exchange processes between the photovoltaic panels and the air, ground and other environments, input the hourly meteorological parameters of a typical meteorological year in the target city, such as dry-bulb temperature, wind speed, total horizontal radiation intensity, etc., and simulate the temperature changes and power generation of photovoltaic panels under different environmental conditions.

[0135] In addition, the system of this embodiment also uses indicators such as root mean square error RMSE and mean relative error MAPE to evaluate the accuracy of the model. The smaller the RMSE and MAPE, the higher the accuracy of the model.

[0136] Preferably, this embodiment can utilize the parameter estimation tool GenOpt in EnergyPlus, adopt the Hooke-Jeeves pattern search algorithm, use the measured photovoltaic panel temperature and output power as training data, optimize the key parameters such as the convective heat transfer coefficient and the backplane infrared emissivity in the model, so that the root mean square error RMSE between the model output and the measured data is less than 1.2°C, and the average relative error MAPE is less than 5%.

[0137] Step 105: Determine key influencing factors according to the photovoltaic panel thermal balance model, optimize and adjust the first photovoltaic panel installation strategy according to the key influencing factors, output and optimize and adjust the distributed photovoltaic system according to the second photovoltaic panel installation strategy.

[0138] The constructed photovoltaic panel thermal balance model can further determine the key factors affecting the photovoltaic panel temperature and photovoltaic power generation efficiency, namely the key influencing factors, and readjust the weight parameters of the first photovoltaic panel installation strategy based on the determined key influencing factors. The strategy is readjusted with the photovoltaic panel temperature change and power generation efficiency as the optimization goals to obtain the second photovoltaic panel installation strategy.

[0139] Specifically, in this embodiment, determining key influencing factors according to the photovoltaic panel thermal balance model, and optimizing and adjusting the first photovoltaic panel installation strategy according to the key influencing factors include:

[0140] The photovoltaic panel temperature and the power generation efficiency of the system are simulated and analyzed by the photovoltaic panel thermal balance model to obtain a corresponding change curve;

[0141] Screening the influencing factors of the change curve by using a sensitivity analysis algorithm to obtain the key influencing factors;

[0142] Then, the first photovoltaic panel installation strategy is optimized and adjusted in parameters based on the key influencing factors through a genetic algorithm.

[0143] By simulating the temperature and power generation efficiency of photovoltaic panels through the constructed thermal balance model, the curves of the change of photovoltaic panel temperature and power generation efficiency over time can be obtained. By analyzing the above curves, various factors that may affect the temperature and power generation efficiency of photovoltaic panels can be screened out. The most significant factors affecting the temperature and efficiency of photovoltaic panels, such as ambient temperature, wind speed or conversion efficiency, can be screened out through the sensitivity analysis algorithm. After determining the key factors, the system will make secondary optimization adjustments to the installation strategy of photovoltaic panels based on the key factors, further improving the power generation efficiency of photovoltaic panels while reducing their temperature impact.

[0144] In this embodiment, based on the verified thermal balance model, the system will also carry out simulation analysis of photovoltaic panel temperature and power generation performance, set different ambient temperature and radiation intensity conditions, simulate the dynamic temperature response and output power changes of photovoltaic panels, and draw curves of temperature and efficiency changes over time.

[0145] Preferably, based on the verified thermal balance model, this embodiment sets the ambient temperature to -10°C, 0°C, 10°C, 25°C, and 40°C, and the radiation intensity to 200W / m 2 , 400W / m 2 、600W / m 2 、800W / m 2 , 1000W / m 2 , a total of 25 working condition combinations were used to simulate the hourly variation curves of photovoltaic panel temperature and power generation efficiency under each working condition. It was found that when the radiation intensity was 1000W / m 2 When the ambient temperature is 40℃, the maximum temperature of the photovoltaic panel can reach 78℃, and the power generation efficiency drops to 8.5%.

[0146] The influencing factors screening by the sensitivity analysis algorithm described in this embodiment specifically includes: using the Morris global sensitivity analysis method to evaluate the significance of the influence of environmental factors, material properties and structural parameters on the temperature and power generation efficiency of photovoltaic panels. Among them, the Morris method calculates the elementary effect (EE) of each factor through factor perturbation and trajectory sampling. EE is defined as the approximate value of the partial derivative of the model output with respect to a certain factor, the mean μ of EE is defined as the overall influence of the factor, and the standard deviation σ of EE is defined as the nonlinear influence of the factor or the interaction with other factors. In the Morris diagram, the larger the μ, the stronger the main effect of the factor, and the larger the σ, the stronger the nonlinear or interaction effect of the factor.

[0147] Preferably, the present embodiment adopts Morris sensitivity analysis method, and the initial temperature of photovoltaic panel is 25±5℃, wind speed is 2.5±0.5m / s, and radiation intensity is 800±160W / m 2 50 Morris experimental samples were generated within the variation range of 10 parameters, including 0.18±0.036 for silicon cell conversion efficiency and 0.9±0.18 for backplane roughness, and the EnergyPlus model was run to calculate the maximum temperature and average power generation efficiency of the photovoltaic panel, and the Morris index μ and σ of each factor were obtained. The above data indicate that the μ values ​​of radiation intensity, silicon cell conversion efficiency and wind speed are the largest, which are 15.8, 12.6 and 9.4 respectively, and are the key factors affecting the temperature and efficiency of photovoltaic panels; the σ values ​​of backplane roughness, photovoltaic panel length and silicon cell temperature coefficient are relatively large, which are 4.2, 3.5 and 2.8 respectively, indicating that these factors have a strong interactive effect with other factors.

[0148] Therefore, the factors to be analyzed and their range of variation can be determined by performing sensitivity analysis on the influencing factors, randomly generating perturbation vectors in the factor space, constructing the Morris trajectory matrix, taking the trajectory matrix as the input of the model, running the model to calculate the output response, calculating the EE value, μ value and σ value of each factor, drawing the Morris diagram, and screening the key influencing factors. Within the range of variation, multiple key parameters such as the initial temperature, wind speed, solar radiation intensity, conversion efficiency, and surface roughness of the photovoltaic panel are perturbed by factors, and the factors that have the most significant impact on the temperature and efficiency of the photovoltaic panel, such as ambient temperature, wind speed, conversion efficiency, etc., are screened out.

[0149] According to the sensitivity analysis results, the system of this embodiment can determine the key influencing factors of photovoltaic panel temperature and efficiency, and use genetic algorithms to optimize the design of the heat dissipation structure and temperature control strategy of the photovoltaic panel, that is, the first photovoltaic panel installation strategy is optimized and adjusted based on the key influencing factors described in this embodiment. A multi-objective optimization model is established with the size, spacing, and material thermal conductivity of the photovoltaic panel heat sink as optimization variables, and the minimum maximum temperature of the photovoltaic panel, the minimum temperature fluctuation, and the maximum power generation efficiency as optimization goals.

[0150] Preferably, according to the sensitivity analysis results, this embodiment adopts the NSGA-II multi-objective genetic algorithm, takes the four variables of heat sink width, thickness, spacing, and material thermal conductivity as optimization parameters, takes the three indicators of photovoltaic panel maximum temperature, temperature fluctuation, and power generation efficiency as optimization targets, generates an initial population of 50 individuals, crossover probability 0.8, mutation probability 0.1, iterates 500 generations, and obtains 8 non-dominated solutions of the Pareto optimal frontier. After weighing the heat dissipation performance and cost, the solution of heat sink width 0.08m, thickness 0.002m, spacing 0.04m, and thermal conductivity 205W / (m·K) is selected and applied to a 1m×1.5m, 250W rated power polycrystalline silicon photovoltaic panel.

[0151] Combined with hourly weather forecast data and load forecast data, the temperature control strategy is designed. The upper limit of the photovoltaic panel surface temperature is 50℃ as the target. According to the predicted ambient temperature and radiation intensity, the cooling fan is started one hour in advance. The fuzzy PID control algorithm is used to dynamically adjust the speed of the cooling fan according to the deviation and deviation change rate between the photovoltaic panel surface temperature and the target temperature, so that the photovoltaic panel temperature is stabilized at 50±2℃, and the photovoltaic panel power generation performance is maximized. After optimization, under sunny conditions in summer, the maximum temperature of the photovoltaic panel is reduced from 78℃ to 52℃, the temperature fluctuation is reduced from 30℃ to 5℃, the power generation efficiency is increased from 8.5% to 14.3%, the annual power generation is increased by 10.2%, the annual power consumption of the cooling fan is about 1.5% of the power generation of the photovoltaic panel, and the operation and maintenance cost is equivalent to the annual power generation income of the photovoltaic panel.

[0152] Step 106: Acquire actual operation data of the distributed photovoltaic system after optimization and adjustment, construct a multivariate regression model based on the actual operation data, optimize and adjust the multivariate regression model according to the heat island intensity index and the reflected light distribution, and output the optimized distributed photovoltaic system deployment strategy.

[0153] After obtaining the second photovoltaic panel installation strategy, the distributed photovoltaic system can be adjusted through the strategy, and the strategy can be further optimized according to the actual operation data of the photovoltaic system, including the heat island intensity index and the distribution of reflected light, to obtain the optimized photovoltaic system optimization strategy, which improves the power generation efficiency of the optimized photovoltaic system deployment strategy, and also reduces the temperature of the photovoltaic panels caused by the urban heat island effect and building reflected radiation, thereby improving the overall efficiency of the photovoltaic power generation system. At the same time, by identifying the power generation mode and efficiency characteristics of the photovoltaic system under different environmental conditions, and determining the correlation between the actual power generation efficiency and environmental factors, the deployment strategy of the photovoltaic system is optimized, and the intelligent management and efficient operation of the photovoltaic power generation system are also realized.

[0154] Specifically, the method of acquiring the actual operation data of the distributed photovoltaic system after optimization and adjustment and constructing a multivariate regression model according to the actual operation data in this embodiment includes:

[0155] Acquiring the actual operation data of the distributed photovoltaic system, and performing dimensionality reduction and feature extraction processing on the actual operation data to obtain a photovoltaic feature data set;

[0156] A cluster analysis is performed on the photovoltaic characteristic data set, the photovoltaic characteristic data set is divided into different clusters according to preset power generation indicators, and photovoltaic power generation influencing factors are determined according to the clusters, and then the multivariate regression model is constructed according to the photovoltaic power generation influencing factors and power generation efficiency.

[0157] After the secondary adjustment of the photovoltaic panel installation strategy is determined, the system can adjust the distributed photovoltaic system according to it, so as to obtain the actual operation data of the adjusted photovoltaic system, and obtain the photovoltaic feature data set by performing data dimension reduction and feature extraction on the actual operation data of the photovoltaic system. Then, the photovoltaic feature data set is clustered to explore the power generation law and efficiency characteristics of photovoltaic panels under different environmental conditions, and to construct a corresponding multivariate regression model to represent the quantitative relationship between power generation efficiency and environmental factors.

[0158] In this embodiment, the actual operation data of the photovoltaic panel is obtained through the photovoltaic power station monitoring system, including the electrical parameters such as voltage, current, power on the DC side and AC side of the inverter, and the temperature data on the surface of the photovoltaic panel. The photovoltaic operation data is preprocessed to remove abnormal values ​​and invalid values, interpolate and complete the missing data, and perform data normalization and standardization according to the statistical characteristics of the data to eliminate the influence of different dimensions and orders of magnitude, so as to prepare for subsequent data mining and analysis.

[0159] Preferably, in order to obtain the actual operation data of the photovoltaic panel, the present embodiment uses data acquisition equipment to collect data of multiple dimensions such as the voltage, current, power and temperature of the photovoltaic panel in real time, and the sampling frequency can be set to once per minute. The collected data is cleaned and preprocessed to remove abnormal data and noise data that exceed the normal range, such as voltage exceeding the rated value by ±5%, current exceeding the rated value by ±10%, etc. Then the data is normalized to the maximum and minimum values, and the data is mapped to the [0,1] interval.

[0160] This embodiment preferably uses feature extraction algorithms such as principal component analysis PCA and independent component analysis ICA to perform dimensionality reduction processing on photovoltaic operation data and extract the main eigenvectors of the data. PCA converts the original data set into a set of linearly independent principal components through orthogonal transformation to extract the main features of the data. The original data is centralized, the covariance matrix of the data is calculated, the covariance matrix is ​​subjected to eigenvalue decomposition, the eigenvectors corresponding to the largest k eigenvalues ​​are selected to form a principal component matrix, and the original data is multiplied by the principal component matrix to obtain the principal component score.

[0161] Specifically, the system of this embodiment uses the principal component analysis (PCA) algorithm to extract key characteristic parameters such as photovoltaic panel power generation efficiency, power generation, temperature change, etc. based on the preprocessed data, and selects the top k principal components with a cumulative variance contribution rate of 95% as feature vectors.

[0162] ICA, on the other hand, separates independent source signals from mixed signals by maximizing statistical independence. Commonly used ICA algorithms include FastICA and Infomax, which center and whiten the data, randomly initialize the unmixing matrix, iteratively update the unmixing matrix, maximize the non-Gaussianity of the output signal, and multiply the output signal by the transpose of the unmixing matrix to obtain independent components. The extracted features include statistical indicators such as the mean, variance, and peak-to-peak value of voltage, current, power, and temperature, as well as external factors such as ambient temperature, radiation intensity, and wind speed, to form a photovoltaic power generation feature data set.

[0163] The cluster analysis of the photovoltaic characteristic data set described in this embodiment is specifically as follows: using the K-means clustering algorithm and the hierarchical clustering algorithm to perform cluster analysis on the photovoltaic power generation characteristic data set, and dividing the data into several clusters according to the similarity of indicators such as voltage, current, power and efficiency, each cluster represents a typical power generation mode. Randomly select K samples as the initial cluster center, calculate the Euclidean distance of each sample to each cluster center, divide the sample into the cluster with the closest distance, recalculate the mean of each cluster as the new cluster center, and repeat the calculation of the Euclidean distance and the cluster division until the cluster center no longer changes. Use the aggregation method to initialize each sample as a cluster, calculate the distance matrix between each cluster (such as the shortest distance, the longest distance or the average distance), merge the two clusters with the closest distance, and update the distance matrix, and repeatedly merge the clusters until all samples are classified into one category or the preset number of clusters is reached. The clustering process can be intuitively represented by a tree diagram.

[0164] Preferably, this embodiment uses K-means clustering algorithm to perform cluster analysis on the extracted feature parameters, and determines the optimal number of clusters to be 3 through the elbow rule, and obtains the photovoltaic panel power generation mode and efficiency feature set under three typical environmental conditions of sunny, cloudy and rainy days. At the same time, the environmental factor data of the area where the photovoltaic panel is located is obtained, including solar radiation intensity, temperature, humidity, wind speed, etc., and data preprocessing and feature extraction are performed.

[0165] This embodiment constructs a multivariate regression model, specifically including: exploring the power generation rules and efficiency characteristics of photovoltaic panels under different environmental conditions, such as high temperature and low efficiency mode, low temperature and high efficiency mode, etc. For each power generation mode, the distribution of factors such as ambient temperature, radiation intensity, wind speed and photovoltaic panel temperature is statistically analyzed, and scatter plots and box plots of environmental factors and power generation efficiency are drawn to analyze the correlation and influence trend between each factor and power generation efficiency, and preliminarily determine the key environmental factors affecting photovoltaic power generation. Select power generation efficiency as the dependent variable, ambient temperature, radiation intensity, wind speed, photovoltaic panel temperature, etc. as independent variables, construct a multivariate linear regression model, use the least squares method to estimate the model coefficients, and obtain the quantitative relationship between power generation efficiency and environmental factors.

[0166] When constructing a regression model, the system provided in this embodiment also needs to check whether there is a significant linear relationship between the independent variable and the dependent variable. The F test can be used to compare the ratio of the regression mean square to the residual mean square at a given confidence level. If it is greater than the critical value, it indicates that the regression model is significant. A t test is performed on each independent variable. If the absolute value of the t statistic is greater than the critical value, it indicates that the independent variable has a significant effect on the dependent variable. According to the sign and size of the regression coefficient, the direction and intensity of the influence of each independent variable on the dependent variable can be explained. At the same time, it is necessary to check whether there is multicollinearity between the independent variables, further introduce factors such as the urban heat island intensity index and the proportion of building reflected radiation, analyze the impact of the urban microclimate environment on the efficiency of photovoltaic power generation, and quantitatively evaluate the contribution rate of the urban heat island effect and building reflected light.

[0167] The environmental factor feature set and the power generation efficiency feature set were analyzed for correlation, and the quantitative relationship model between the actual power generation efficiency and environmental factors was established using the multivariate linear regression algorithm. The regression equation was: power generation efficiency = 78 × radiation intensity + 15 × temperature - 03 × humidity - 02 × wind speed. Based on this model, the impact of the urban heat island effect on the power generation efficiency of photovoltaic panels was evaluated. The urban heat island effect causes the temperature around the photovoltaic panels to rise by 2 to 3 ° C, resulting in an average decrease of about 5% in the power generation efficiency of photovoltaic panels.

[0168] In addition, the system of this embodiment also applies the regression model to different urban areas and building types, predicts the spatial distribution of photovoltaic power generation efficiency, draws contour maps and color maps of photovoltaic power generation efficiency, and identifies high-efficiency power generation areas and low-efficiency power generation areas within the city. Combined with the spatial distribution data of urban heat island intensity and building density, the influence of urban thermal environment and building reflected light on photovoltaic power generation efficiency is quantitatively evaluated.

[0169] The reflected light intensity and reflection angle data of the building where the photovoltaic panels are located are obtained, and the data are collected every 10 minutes using an optical sensor, and data preprocessing and feature extraction are performed. The building reflected light feature set is correlated with the actual power generation efficiency, and the partial least squares regression (PLSR) algorithm is used to evaluate the impact of building reflected light on the power generation efficiency of photovoltaic panels. The regression equation is: power generation efficiency = 82 × reflected light intensity + 12 × reflection angle. The analysis results show that for every 100W / m increase in building reflected light intensity, the building reflected light intensity increases by 100W / m 2 , the photovoltaic panel power generation efficiency can be increased by about 2%; for every 10° increase in reflection angle, the power generation efficiency can be increased by about 5%. Therefore, when laying out photovoltaic panels, we should try to choose a location with a high intensity of building reflected light and a suitable reflection angle to improve power generation efficiency.

[0170] Furthermore, in this embodiment, the multivariate regression model is optimized and adjusted according to the heat island intensity index and the reflected light distribution, and an optimized distributed photovoltaic system optimization strategy is output, including:

[0171] Adjusting the parameters of the multivariate regression model according to the heat island intensity index and the reflected light distribution, predicting photovoltaic power generation efficiency through the adjusted model, and outputting photovoltaic prediction data;

[0172] The photovoltaic prediction data is compared with the actual photovoltaic data, and the model prediction error is evaluated on the comparison result by a deviation analysis method, and the model parameters are adjusted by feedback based on the evaluation result;

[0173] The second photovoltaic panel installation strategy is optimized through the feedback-adjusted model to obtain the distributed photovoltaic system deployment strategy.

[0174] After obtaining the multivariate regression model, the system will adjust the model parameters through the heat island intensity index and the distribution of reflected light, so that the output prediction data takes into account factors such as the city's heat island intensity and reflected light distribution, and predict the power generation efficiency of the photovoltaic system through the adjusted model. The prediction data and the actual power generation efficiency data of the photovoltaic system, that is, the actual photovoltaic data, are compared to determine the error between the model and the actual photovoltaic system, and the error is analyzed and evaluated to determine the source of the error. Then, the model is feedback adjusted according to the source of the error to further improve the practicality, reliability and accuracy of the model.

[0175] In this embodiment, it also includes: collecting historical data of the actual operation of the photovoltaic power station, including key parameters such as inverter output power, conversion efficiency, photovoltaic module backplane temperature, and meteorological elements such as ambient temperature, radiation intensity, and wind speed, and constructing a multi-source heterogeneous operation data set. Perform exploratory analysis and data mining on the operation data, calculate the correlation coefficient between photovoltaic module temperature and power generation efficiency, draw scatter plots and regression curves, and quantitatively evaluate the impact of temperature on efficiency.

[0176] Wavelet transform and spectrum analysis are used to extract the periodicity, trend and other characteristics of temperature and efficiency time series. Wavelet transform uses mother wavelet function to decompose time series at multiple scales and extract the characteristics of different frequency components. Common wavelet basis functions include Haar, Daubechies, Symlets, etc. Appropriate mother wavelet function and decomposition layer are selected to perform wavelet transform on time series to obtain wavelet coefficients of each layer. Threshold processing is performed on wavelet coefficients to remove noise, reconstruct wavelet coefficients, obtain denoised time series, and extract statistical characteristics of wavelet coefficients, such as mean, variance, energy, etc. Spectral analysis converts time series to frequency domain through Fourier transform to identify periodic components. Common spectrum estimation methods include periodogram method, BT method, MUSIC method, etc. Detrending time series is performed, autocorrelation function or cross-correlation function is calculated, Fourier transform is performed on correlation function to obtain power spectrum density function, identify the peak frequency of power spectrum density function, and determine periodic components.

[0177] This embodiment preferably optimizes the passive insulation scheme according to the structural characteristics and material properties of the photovoltaic module. Fill low thermal conductivity materials such as aerogel and vacuum insulation panels between the module backplane and the frame to reduce the heat transfer from the backplane to the battery cell. Coat the surface of the module glass cover with a low emissivity selective coating to reduce the absorption ratio of solar radiation.

[0178] In this embodiment, differentiated active cooling strategies are formulated for different installation methods and climatic conditions. For distributed photovoltaic systems installed on roofs, automatically adjustable shutters can be set on the back of the components to adjust the opening and closing angles of the vents according to changes in temperature and radiation intensity to enhance natural ventilation and heat exchange. For ground power stations, evaporative cooling materials can be laid under the components to reduce the local ambient temperature by using the principle of heat absorption by water evaporation.

[0179] Comprehensively analyze the temperature control effects of active cooling and passive insulation measures, and compare the maximum operating temperature, temperature fluctuation range, temperature uniformity and other indicators of different schemes. According to the climatic characteristics of the location of the photovoltaic power station, give the definition of high temperature and low temperature conditions. In the evaluation of the effect of the temperature control scheme, it is necessary to set the temperature range and time scale of the evaluation according to the high temperature and low temperature characteristics of different climate zones to improve the applicability and comparability of the evaluation results. Select typical high temperature and low temperature weather conditions, simulate the power generation performance of the photovoltaic array, and calculate the annual average power generation efficiency improvement and annual power generation increase. In addition, extreme weather conditions, such as continuous high temperature and typhoons, need to be considered to evaluate the reliability and safety of the temperature control scheme under extreme conditions.

[0180] Furthermore, the present embodiment uses a deviation analysis method to evaluate the model prediction error of the comparison results, and uses the evaluation results to feedback and adjust the model parameters, including:

[0181] Attributing and decomposing the comparison results by the deviation analysis method to obtain error source analysis, and performing parameter correction on the model according to the error source analysis;

[0182] An uncertainty analysis is performed on the comparison results to obtain corresponding sources of uncertainty, and then parameters of the model are adjusted according to the sources of uncertainty.

[0183] The error sources are determined by the deviation analysis method, and the model parameters are corrected according to the error sources, that is, the factors that have a greater impact on the prediction error are determined, so as to achieve dynamic parameter adjustment of the model, so that the model can adapt to the actual operating status and environmental conditions of the photovoltaic system. Through the uncertainty analysis of the error, the uncertainty sources in the model parameters, input data and model structure are clarified, the reliability and risk of the model prediction are evaluated, and the model and algorithm are further optimized, which improves the robustness and adaptability of the model prediction, realizes the continuous iteration and upgrading of the model, and also improves the scientificity and practicality of the simulation prediction of photovoltaic power generation efficiency.

[0184] In this embodiment, the actual power generation efficiency data is compared with the power generation efficiency data predicted by simulation, and the error statistics of the two sets of data at different time scales (such as hourly, daily, and monthly) are calculated, including the mean absolute error MAE, the root mean square error RMSE, the mean relative error MAPE, etc., to quantitatively evaluate the accuracy and deviation of the model prediction. The deviation analysis method is used to attribute and decompose the model prediction error, distinguish between systematic deviation and random deviation, analyze the time trend, spatial distribution and influencing factors of the error, such as decomposing the MSE of the hourly error into three parts: mean deviation, variance deviation and covariance deviation, and determine the main source of the error.

[0185] First, this embodiment will diagnose and correct the structure, parameters and input data of the model based on the deviation analysis results, and determine the factors that have a greater impact on the prediction error through sensitivity analysis methods such as Morrisscreening or Sobol index.

[0186] If the Morris method is used, the basic influence values ​​μ and σ are calculated by perturbing each input parameter, where μ represents the overall influence of the parameter and σ represents the nonlinear influence of the parameter or the interaction with other parameters. The value range of each parameter is divided into several levels, a reference point is randomly selected in the parameter space, and each parameter is perturbed in turn to obtain a set of Morris trajectories. The basic influence value of the output response is calculated for each trajectory. The Morris trajectory and the basic influence value are repeatedly obtained to obtain the distribution of μ and σ for each parameter, and the importance of the parameters is sorted according to the size of μ and σ.

[0187] If the Sobol index is used, the total variance of the model output is decomposed into the variance contribution of each input parameter through variance decomposition, including the first-order index and global index of the parameter. The value range of the parameter is converted to a uniform distribution between 0 and 1, and two sets of parameter sample matrices A and B are generated by Latin hypercube sampling. The model output vectors YA and YB of A and B are calculated respectively. The i-th parameter in A is replaced with the corresponding value in B to obtain a new set of sample matrices Ci, and its output vector YCi is calculated. The variance of YA, YB and YCi is calculated to obtain the first-order index and global index of the i-th parameter. The output vector YCi and the first-order index and global index of the i-th parameter are calculated repeatedly until all parameters are calculated.

[0188] In addition, this embodiment will also use automatic machine learning algorithms such as RandomForest or XGBoost to optimize the hyperparameters of the model to improve the stability and robustness of the model. Taking random forest as an example, the hyperparameters that need to be optimized include the number of decision trees, the maximum depth of each tree, the minimum number of samples for each node, the maximum number of features for feature selection, etc. A grid search or random search method is used to randomly sample within the range of hyperparameter values ​​to generate a set of hyperparameter combinations. For each set of hyperparameters, a k-fold cross-validation method is used to randomly divide the data set into k subsets. Each time, k-1 subsets are selected as training sets, and the remaining 1 subset is used as a validation set. The model is trained and the performance of the model is evaluated on the validation set, such as the mean square error MSE and the coefficient of determination R. 2 Etc. Repeat k times to obtain k groups of performance indicators, take their average as the performance score of the group of hyperparameters, and select the group of hyperparameters with the best performance score as the final model parameters.

[0189] The uncertainty analysis of the comparison results described in this embodiment is specifically as follows: considering the sources of uncertainty in model parameters, input data, and model structure, using methods such as Monte Carlo simulation or Bayesian inference, estimating the confidence interval or probability distribution of the prediction results, evaluating the reliability and risk of model predictions, and providing a reference for decision making. Based on the results of uncertainty analysis, the model and algorithm are further optimized, and the robustness and adaptability of the prediction are improved through methods such as model integration or Bayesian model averaging, so as to achieve continuous iteration and upgrading of the model, establish a closed-loop optimization mechanism of model prediction, error analysis, parameter correction, and uncertainty assessment, and improve the scientificity and practicality of photovoltaic power generation efficiency simulation prediction.

[0190] Preferably, this embodiment carries out uncertainty analysis of model prediction, considers the measurement errors of input parameters such as ambient temperature and radiation, as well as the uncertainty of model structure and parameters, uses the Monte Carlo method to extract 1000 groups of input samples, and calculates the mean and confidence interval of the model output. The data shows that the 95% confidence interval of the model prediction efficiency is [11.2%, 13.5%], and the uncertainty level is controllable. Through a series of measures such as model error analysis, parameter correction, and uncertainty assessment, the accuracy and reliability of the photovoltaic power generation efficiency prediction model have been significantly improved, providing important data support and decision-making basis for the optimized operation of photovoltaic power stations.

[0191] In order to better illustrate the working principle and step flow of the optimization method and device of a distributed photovoltaic system of the present invention, it is possible but not limited to refer to the relevant records above.

[0192] Accordingly, see Figure 2 , Figure 2 This is a structural diagram of an embodiment of the distributed photovoltaic system optimization device provided by the present invention. Figure 2 As shown, the optimization device includes a heat island intensity determination module 201, a building model calculation module 202, a first strategy output module 203, a radiation intensity calculation module 204, a second strategy adjustment module 205 and a photovoltaic system optimization module 206; wherein:

[0193] The heat island intensity determination module 201 is used to obtain temperature distribution data of the target city, and calculate the temperature distribution data through an interpolation analysis method to obtain a heat island intensity index;

[0194] The building model calculation module 202 is used to obtain the three-dimensional building model of the target city, and calculate the solar radiation intensity and the distribution of reflected light according to the three-dimensional building model;

[0195] The first strategy output module 203 is used to calculate and determine the photovoltaic panel temperature and system power generation efficiency of the distributed photovoltaic system according to the temperature distribution data and the solar radiation intensity, and output a first photovoltaic panel installation strategy;

[0196] The radiation intensity calculation module 204 is used to calculate and determine the radiation intensity of the photovoltaic panel based on the first photovoltaic panel installation strategy and the three-dimensional building model, and to construct a photovoltaic panel thermal balance model based on the photovoltaic panel radiation intensity;

[0197] The second strategy adjustment module 205 is used to determine the key influencing factors according to the photovoltaic panel thermal balance model, optimize and adjust the first photovoltaic panel installation strategy according to the key influencing factors, and output and optimize and adjust the distributed photovoltaic system according to the second photovoltaic panel installation strategy;

[0198] The photovoltaic system optimization module 206 is used to obtain the actual operation data of the distributed photovoltaic system after optimization and adjustment, construct a multivariate regression model according to the actual operation data, optimize and adjust the multivariate regression model according to the heat island intensity index and the reflected light distribution, and output the optimized distributed photovoltaic system optimization strategy.

[0199] In summary, the embodiment of the present invention provides a distributed photovoltaic system optimization method and device, which determines the heat island intensity index through the temperature data of the target city, calculates the solar radiation intensity and the reflected light distribution according to the three-dimensional building model, calculates the photovoltaic panel temperature and the system power generation efficiency of the distributed photovoltaic system through the temperature data and the solar radiation intensity, determines the photovoltaic panel installation strategy, builds a photovoltaic panel thermal balance model according to the installation strategy, optimizes the installation strategy by determining the key influencing factors based on the thermal balance model, adjusts the photovoltaic system according to the optimized strategy, re-optimizes the installation strategy according to the system heat island intensity and the reflected light distribution, and obtains the optimized photovoltaic system deployment strategy. The system optimization strategy is generated by the above optimization method, which improves the power generation efficiency of the photovoltaic system deployment strategy, reduces the photovoltaic panel temperature caused by the heat island effect and the building reflected radiation, and realizes the intelligent management and efficient operation of the photovoltaic power generation system.

[0200] The specific embodiments described above further illustrate the purpose, technical solutions and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A distributed photovoltaic system optimization method, characterized in that: The following steps are involved: Acquire temperature distribution data of the target city, and calculate the temperature distribution data by an interpolation analysis method to obtain a heat island intensity index; Acquire a three-dimensional building model of the target city, and calculate the solar radiation intensity and reflected light distribution according to the three-dimensional building model; Determine the photovoltaic panel temperature and system power generation efficiency of the distributed photovoltaic system according to the temperature distribution data and the solar radiation intensity, and output a first photovoltaic panel installation strategy; Calculating and determining the radiation intensity of the photovoltaic panels based on the first photovoltaic panel installation strategy and the three-dimensional building model, and constructing a photovoltaic panel thermal balance model based on the radiation intensity of the photovoltaic panels; Determine key influencing factors according to the photovoltaic panel thermal balance model, optimize and adjust the first photovoltaic panel installation strategy according to the key influencing factors, output and optimize and adjust the distributed photovoltaic system according to the second photovoltaic panel installation strategy; The actual operation data of the distributed photovoltaic system after optimization and adjustment is obtained, a multivariate regression model is constructed according to the actual operation data, the multivariate regression model is optimized and adjusted according to the heat island intensity index and the reflected light distribution, and the optimized distributed photovoltaic system deployment strategy is output.

2. A distributed photovoltaic system optimization method according to claim 1, characterized in that: The step of obtaining temperature distribution data of the target city and calculating the temperature distribution data by an interpolation analysis method to obtain a heat island intensity index includes: Acquire initial temperature distribution data of the target city, perform outlier elimination processing and missing value supplement processing on the initial temperature distribution data, and obtain the temperature distribution data; Performing spatial interpolation analysis on the temperature distribution data by using the Kriging interpolation method to obtain a temperature distribution surface of the target city; The heat island intensity index is calculated based on the temperature distribution surface, and the target city is classified into a regional heat island effect level based on the heat island intensity index to obtain a regional heat island level.

3. The optimization method of a distributed photovoltaic system according to claim 1, characterized in that: The step of obtaining the three-dimensional building model of the target city and calculating the solar radiation intensity and the distribution of reflected light according to the three-dimensional building model includes: Acquire building point cloud data of the target city, and sequentially perform denoising, point cloud classification and three-dimensional reconstruction on the building point cloud data to obtain the three-dimensional building model; The solar radiation data of the target city and the reflectivity data of the building surface materials are obtained, the solar radiation data and the reflectivity data are input into the three-dimensional building model, and the solar radiation intensity and the reflected light distribution of the building surface are calculated by a Monte Carlo simulation algorithm.

4. The optimization method of a distributed photovoltaic system according to claim 1, characterized in that: The method of calculating and determining the photovoltaic panel temperature and the system power generation efficiency of the distributed photovoltaic system according to the temperature distribution data and the solar radiation intensity, and outputting a first photovoltaic panel installation strategy, is specifically: Inputting the temperature distribution data and the solar radiation intensity into a preset Sandia array performance model for calculation and processing to obtain the photovoltaic panel temperature and the system power generation efficiency; Acquire building density data and green coverage data of the target city, and construct a photovoltaic array multi-objective optimization model based on the building density data, the green coverage data, the photovoltaic panel temperature, and the system power generation efficiency; The photovoltaic array multi-objective optimization model is optimized and solved by a genetic algorithm, and the first photovoltaic panel installation strategy is output.

5. The distributed photovoltaic system optimization method according to claim 1, characterized in that: The step of calculating and determining the radiation intensity of the photovoltaic panel based on the first photovoltaic panel installation strategy and the three-dimensional building model, and constructing a photovoltaic panel thermal balance model based on the radiation intensity of the photovoltaic panel, comprises: Adjusting parameters of the three-dimensional building model according to the first photovoltaic panel installation strategy, and simulating the reflection and shielding of sunlight by Monte Carlo ray tracing method to obtain the radiation flux per unit area of ​​the photovoltaic panel; Constructing a corresponding building surface radiation transmission mathematical model according to the radiation flux, and calculating the photovoltaic panel radiation intensity of the photovoltaic panel through the building surface radiation transmission mathematical model; The thermal balance model of the photovoltaic panel is constructed according to the radiation intensity of the photovoltaic panel and the heat exchange process of the photovoltaic panel.

6. A distributed photovoltaic system optimization method according to claim 1, characterized in that: Determining key influencing factors according to the photovoltaic panel thermal balance model, and optimizing and adjusting the first photovoltaic panel installation strategy according to the key influencing factors, includes: The photovoltaic panel temperature and the power generation efficiency of the system are simulated and analyzed by the photovoltaic panel thermal balance model to obtain a corresponding change curve; Screening the influencing factors of the change curve by using a sensitivity analysis algorithm to obtain the key influencing factors; Then, the first photovoltaic panel installation strategy is optimized and adjusted in parameters based on the key influencing factors through a genetic algorithm.

7. The distributed photovoltaic system optimization method according to claim 1, characterized in that: The obtaining of actual operation data of the distributed photovoltaic system after optimization and adjustment, and constructing a multivariate regression model according to the actual operation data, comprises: Acquiring the actual operation data of the distributed photovoltaic system, and performing dimensionality reduction and feature extraction processing on the actual operation data to obtain a photovoltaic feature data set; A cluster analysis is performed on the photovoltaic characteristic data set, the photovoltaic characteristic data set is divided into different clusters according to preset power generation indicators, and photovoltaic power generation influencing factors are determined according to the clusters, and then the multivariate regression model is constructed according to the photovoltaic power generation influencing factors and power generation efficiency.

8. The distributed photovoltaic system optimization method according to claim 1, characterized in that: The optimizing and adjusting the multivariate regression model according to the heat island intensity index and the reflected light distribution, and outputting an optimized distributed photovoltaic system optimization strategy, comprises: Adjusting the parameters of the multivariate regression model according to the heat island intensity index and the reflected light distribution, predicting photovoltaic power generation efficiency through the adjusted model, and outputting photovoltaic prediction data; The photovoltaic prediction data is compared with the actual photovoltaic data, and the model prediction error is evaluated on the comparison result by a deviation analysis method, and the model parameters are adjusted by feedback based on the evaluation result; The second photovoltaic panel installation strategy is optimized through the feedback-adjusted model to obtain the distributed photovoltaic system deployment strategy.

9. A distributed photovoltaic system optimization method as claimed in claim 8, characterized in that: The method of evaluating the model prediction error of the comparison results by using the deviation analysis method and performing feedback adjustment on the model parameters according to the evaluation results includes: Attributing and decomposing the comparison results by the deviation analysis method to obtain error source analysis, and performing parameter correction on the model according to the error source analysis; An uncertainty analysis is performed on the comparison results to obtain corresponding sources of uncertainty, and then parameters of the model are adjusted according to the sources of uncertainty.

10. An optimization device for a distributed photovoltaic system, characterized in that: The optimization device includes a heat island intensity determination module, a building model calculation module, a first strategy output module, a radiation intensity calculation module, a second strategy adjustment module and a photovoltaic system optimization module; wherein: The heat island intensity determination module is used to obtain temperature distribution data of the target city, and calculate the temperature distribution data through an interpolation analysis method to obtain a heat island intensity index; The building model calculation module is used to obtain the three-dimensional building model of the target city, and calculate the solar radiation intensity and the reflected light distribution according to the three-dimensional building model; The first strategy output module is used to calculate and determine the photovoltaic panel temperature and system power generation efficiency of the distributed photovoltaic system according to the temperature distribution data and the solar radiation intensity, and output a first photovoltaic panel installation strategy; The radiation intensity calculation module is used to calculate and determine the radiation intensity of the photovoltaic panel based on the first photovoltaic panel installation strategy and the three-dimensional building model, and to construct a photovoltaic panel thermal balance model based on the photovoltaic panel radiation intensity; The second strategy adjustment module is used to determine the key influencing factors according to the photovoltaic panel thermal balance model, optimize and adjust the first photovoltaic panel installation strategy according to the key influencing factors, and output and optimize and adjust the distributed photovoltaic system according to the second photovoltaic panel installation strategy; The photovoltaic system optimization module is used to obtain the actual operation data of the distributed photovoltaic system after optimization and adjustment, construct a multivariate regression model based on the actual operation data, optimize and adjust the multivariate regression model according to the heat island intensity index and the reflected light distribution, and output the optimized distributed photovoltaic system optimization strategy.

Images