Photovoltaic power generation prediction method and device based on digital twinning and medium

Abstract

The invention discloses a photovoltaic power generation prediction method and device based on digital twinning and a medium, and relates to the technical field of new energy, and the method comprises the steps: constructing a three-dimensional geometric model of a photovoltaic power station; acquiring electrical parameters and environmental data of each component in the photovoltaic power station, and inputting the electrical parameters and environmental data into the three-dimensional geometric model according to the three-dimensional coordinates corresponding to the components to form a digital twinborn model; outputting a future environment data sequence in a preset time period based on the environment data through a trained generative model, and inputting the future environment data sequence into a digital twinborn model; performing power generation simulation through a digital twinborn model to obtain a power prediction value in a preset time period; and adjusting model parameters of the digital twinborn model according to the power predicted value and the actual power. And in combination with a high-resolution future environment sequence output by the generative model, power generation simulation of the digital twin model is fully fused with physical position characteristics and meteorological evolution laws.

Description

Technical Field

[0001] The present application relates to the field of new energy technology, and specifically to a photovoltaic power generation prediction method, device and medium based on digital twins. Background Art

[0002] Photovoltaic power generation is a clean energy technology that uses the photovoltaic effect to convert sunlight directly into electricity through solar cells. Its system primarily consists of solar panels, controllers, and inverters, and features no mechanical components, a long lifespan, and easy maintenance. Fluctuations in photovoltaic power generation affect grid stability. The widespread application of photovoltaic power generation technology in areas such as home power supply, transportation, communications, and large-scale power plants has placed higher demands on power generation stability. Therefore, accurate prediction of photovoltaic power generation is crucial for optimizing energy allocation, reducing grid fluctuations, and improving energy efficiency.

[0003] Traditional photovoltaic power generation forecasting methods primarily include physical models, statistical models, and machine learning. Physical models rely on a deep understanding of the physical properties of photovoltaic modules, but struggle to accurately reflect complex and changing environmental factors. Statistical models establish statistical relationships based on historical data, but lack a deep understanding of power generation characteristics under varying conditions. Machine learning can capture complex relationships between data, but its interpretability and generalization capabilities need improvement. This is particularly true given the scarcity of historical monitoring data for extreme weather events, resulting in weak generalization capabilities and large prediction errors in extreme weather conditions. Summary of the Invention

[0004] To solve the above problems, this application proposes a photovoltaic power generation prediction method based on digital twins, including: Construct a 3D geometric model of a photovoltaic power station; Obtaining electrical parameters and environmental data of each component in the photovoltaic power station, and inputting the data into the three-dimensional geometric model according to the three-dimensional coordinates corresponding to the components to form a digital twin model; Outputting a sequence of future environmental data within a preset time period based on the environmental data through the trained generative model and inputting the sequence into the digital twin model; Based on the electrical parameters of the component and the future environmental data sequence, power generation simulation is performed through the digital twin model to obtain a power prediction value within the preset time period; The actual power of the photovoltaic power station within the preset time period is obtained, and the model parameters of the digital twin model are adjusted according to the power prediction value and the actual power.

[0005] On the other hand, this application also proposes a photovoltaic power generation prediction device based on digital twins, including: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute a photovoltaic power generation prediction method based on digital twins as described in the above example.

[0006] On the other hand, the present application also proposes a non-volatile computer storage medium storing computer executable instructions, wherein the computer executable instructions are configured as: a photovoltaic power generation prediction method based on digital twins as described in the above example.

[0007] This application proposes a photovoltaic power generation prediction method based on digital twins, which can bring the following beneficial effects: By constructing a component-level three-dimensional geometric model and binding spatial coordinates, accurate spatial mapping of environmental data and electrical parameters is achieved. Combined with the high-resolution future environmental sequence output by the generative model, the power generation simulation of the digital twin model fully integrates the physical location characteristics and meteorological evolution laws.

[0008] A dynamic calibration mechanism based on real-time power deviations automatically optimizes the key parameters of the digital twin model, solving the problem of inaccurate predictions in traditional models caused by environmental degradation. At the same time, the generative model significantly enhances its generalization capabilities for extreme weather through pre-training on cross-regional historical data and fine-tuning on target power stations. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings: Figure 1 Schematic diagram of a process of photovoltaic power generation prediction method based on digital twin in an embodiment of the present application; Figure 2 This is a schematic diagram of the system architecture of the photovoltaic power generation system in the embodiment of the present application; Figure 3 This is a schematic diagram of a photovoltaic power generation prediction device based on digital twins in an embodiment of the present application. DETAILED DESCRIPTION

[0010] To make the purpose, technical solutions, and advantages of this application more clear, the technical solutions of this application will be clearly and completely described below in conjunction with the specific embodiments of this application and the corresponding drawings. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0011] The following describes in detail the technical solutions provided by various embodiments of the present application in conjunction with the accompanying drawings.

[0012] like Figure 1 As shown, the embodiment of the present application provides a photovoltaic power generation prediction method based on digital twins, including: S101: Construct a three-dimensional geometric model of a photovoltaic power station.

[0013] Specifically, LiDAR technology was used to perform a 3D point cloud scan of the photovoltaic power station site, covering components, brackets, terrain, and surrounding obstructions. This generated point cloud data containing the spatial distribution of each component. This point cloud data was then preprocessed, with statistical outlier filtering used to remove noise points. An iterative closest point algorithm was used to stitch the multi-site clouds together and unify them into an engineering coordinate system. Based on this preprocessed point cloud data, a moving least squares method was used for surface reconstruction to generate an initial 3D model of the photovoltaic power station, ensuring that features such as bracket corners and photovoltaic panel edges were fully preserved.

[0014] Furthermore, the spatial distribution of the photovoltaic power station is measured by a total station, the physical space coordinate system is defined, the physical space coordinates of each component are determined, a homogeneous transformation matrix from the measurement coordinate system to the model coordinate system is constructed, and the physical coordinates are mapped to the initial three-dimensional model to form a three-dimensional geometric model aligned with the actual spatial position.

[0015] It should be noted that physical space data is acquired through a total station, a physical coordinate system is established, and the spatial coordinates of each component are clarified, forming the digital foundation for the actual spatial position. A homogeneous transformation matrix is ​​constructed, containing transformation parameters such as rotation, translation, and scaling. Matrix operations are used to map the measurement coordinate system to the model coordinate system, eliminating deviations between different coordinate systems. The transformed coordinates are then substituted into the initial three-dimensional model, ensuring that the component positions and relative relationships in the model are essentially consistent with those of the actual power plant. This achieves high-precision spatial matching between the physical scene and the digital model, providing an accurate spatial reference for subsequent power generation simulation, shadow analysis, maintenance planning, and other tasks. This ensures the uniqueness and reproducibility of coordinate transformations, improving the efficiency and reliability of model construction.

[0016] S102: Obtain electrical parameters and environmental data of each component in the photovoltaic power station, and input the data into the three-dimensional geometric model according to the three-dimensional coordinates corresponding to the components to form a digital twin model.

[0017] Specifically, through the sensor devices deployed in the components, electrical parameters and environmental data are collected in real time to form a multi-modal real-time data stream, the sensor data is unified with the timestamp of the three-dimensional model through high-precision clock synchronization technology, the mapping relationship between the sensor position and the vertex of the three-dimensional model is unified based on the spatial coordinate system, the multi-modal real-time data stream is spatio-temporally calibrated, the spatio-temporally calibrated multi-modal real-time data stream is embedded into the three-dimensional geometric model to form a digital twin model.

[0018] It should be noted that high-precision sensors such as thin-film irradiance sensors and platinum resistance temperature sensors are deployed on the surface of the photovoltaic components to collect physical data of each component in real time, including environmental data such as irradiance and temperature and voltage parameters such as component current and voltage. An edge computing gateway is deployed to realize real-time transmission and preprocessing of sensor data through a 5G / optical fiber link.

[0019] In the embodiments of the present application, after the digital twin model is constructed, dynamic correction is performed. Specifically, based on multi-dimensional indexes, the index error values of the digital twin model are calculated, wherein the multi-dimensional indexes include geometric accuracy, physical parameters, and dynamic behavior, the correction threshold values corresponding to the multi-dimensional indexes are obtained, it is respectively judged whether the index error values are greater than the corresponding correction threshold values, when the index error values are greater than the corresponding correction threshold values and the number of index error values is greater than 1, the correction process of the index corresponding to the index error value is triggered, and the digital twin model is corrected. If the error still exceeds the threshold value after one correction, optimization is started again until the accuracy requirement is met.

[0020] For geometric accuracy correction, the actual coordinates of a plurality of feature points in the entity structure of the power station are obtained through high-precision measurement equipment, which are compared with the corresponding point coordinates in the digital twin model to calculate the spatial position deviation; when the deviation exceeds the preset accuracy threshold, the model is adjusted in position.

[0021] For physical parameter correction, the temperature, stress and other data during the operation of the components are collected in real time, compared and analyzed with the simulated physical parameters of the model, and the material properties, electrical parameters and the like in the model are optimized according to the data difference, so that the error between the model output and the actual measurement data does not exceed the preset physical parameter threshold.

[0022] For dynamic behavior correction, the historical operation data of the power station are analyzed, the actual operation characteristics (such as vibration, temperature change, etc.) under different working conditions are identified, the dynamic parameters (such as damping, friction coefficient) in the model are adjusted based on the actual operation characteristics, so that the matching degree between the model simulated equipment operation state and the actual situation is not less than the preset dynamic behavior threshold.

[0023] S103: Output a future environmental data sequence within a preset time period based on the environmental data through the trained generative model, and input it into the digital twin model.

[0024] Specifically, the measured environmental data at the current moment or the measured environmental data sequence within a period of time is used as the initial condition of the generative model and input into the generative model. According to the preset time period, the generative model outputs the environmental parameter prediction sequence at each time point (such as the illumination sequence of each time step Δt). , temperature series wait).

[0025] It should be noted that each parameter value in the environmental parameter prediction sequence corresponds to a specific prediction moment, forming structured time series data.

[0026] By setting up a standardized data interface in the digital twin model, the generated environmental data sequence is received, the generated environmental parameters are associated with the spatial position of the photovoltaic components in the digital twin model, and the prediction sequence is input into the digital twin model frame by frame according to the time step to drive the model to perform real-time status updates.

[0027] In an embodiment of the present application, a generative model is pre-constructed and trained. Specifically, multi-source historical environmental data is obtained and pre-processed, wherein the multi-source historical environmental data includes satellite cloud image data, ground meteorological station time series data, and component monitoring environmental data. The spatial dimension characteristics of each component data point in the satellite cloud image data are extracted, and the time series period characteristics of each component data point in the ground meteorological station time series data are analyzed. Based on the spatial dimension characteristics and the time series period characteristics, a mapping relationship between the spatial position and timestamp of each component data point is established to form a spatiotemporal correlation matrix. The generative model is trained based on the spatiotemporal correlation matrix, and the training results of the generative model are optimized by preset constraints.

[0028] Furthermore, the spatiotemporal correlation matrix is ​​input into the generative model, and the future environmental data of the component data points are predicted by the generative model. The output is a sequence of future environmental data at each time point in a preset future time period. According to the preset spatiotemporal constraints, the component environmental prediction parameters are corrected, and the corrected future environmental data sequence is compared with the measured environmental data sequence to calculate the prediction deviation. According to the prediction deviation, the model parameters of the generative model are adjusted.

[0029] It should be noted that a pre-built generative model based on a Transformer or GAN (Generative Adversarial Network) network processes satellite cloud imagery, weather radar, and historical environmental data to generate high-resolution sequences of light intensity, temperature, and cloud motion for future time periods. This model employs a spatiotemporal Transformer architecture, using a 3D-CNN to extract spatial features from satellite cloud imagery and an LSTM to extract periodicity and mutations in meteorological time series data. An attention mechanism is used to generate hourly sequences of light intensity for future time periods and output a probability distribution. This, combined with temporal continuity constraints and a spatial consistency loss, improves the robustness of predictions for sudden changes in cloudy conditions.

[0030] S104: Based on the electrical parameters of the component and the future environmental data sequence, power generation simulation is performed through the digital twin model to obtain a power prediction value within the preset time period.

[0031] Specifically, the future environmental data sequence is matched to the component model corresponding to the digital twin model, and the instantaneous basic power generation power is calculated based on the formula, where the formula is: , where G is the light intensity, A is the component area, η is the rated efficiency, γ is the temperature coefficient, T is the ambient temperature, and the future environmental parameters corresponding to the hourly moments in the future environmental data sequence are used to make the first correction to the standard power generation power.

[0032] Furthermore, historical operating data is analyzed to identify nonlinear factors that are difficult to describe by physical models, including dynamic shading loss, component surface contamination attenuation, and the influence of non-uniform temperature fields. Based on historical power deviation data, a mapping relationship between nonlinear factors and power loss is established. For example, a regression model is used to fit the curve of contamination coefficient and power attenuation. According to the current environmental parameters and component status, the data model is called to generate a nonlinear correction coefficient, and the corrected standard power generation power is corrected for the second time to obtain the power generation power forecast value corresponding to each hour in the preset time period.

[0033] Among them, after the second correction, it also includes: the basic power calculated according to the comprehensive physical laws and the correction results of nonlinear factors, automatically adjusting the contribution ratio of the two according to the complexity of the real-time environment, and generating the power generation prediction value at each time point.

[0034] It's important to note that while future environmental data represents predicted real-time operating parameters, basic power calculations are typically based on standard conditions. This requires eliminating the deterministic differences between standard and predicted operating conditions, necessitating a first correction. Furthermore, since future environmental data is inherently probabilistic, environmental complexity (such as the frequency of cloud cover changes and the complexity of terrain obstruction) can amplify prediction errors. This requires converting these uncertainties into power corrections through statistical models or empirical coefficients, necessitating a second correction.

[0035] S105: Obtain the actual power of the photovoltaic power station within the preset time period, and adjust the model parameters of the digital twin model according to the power prediction value and the actual power.

[0036] Specifically, through the power monitoring device of the photovoltaic power station, the actual power generation data is collected in real time within a preset time period, the power prediction value output by the digital twin model is aligned with the preprocessed actual power at each time point, the root mean square error is calculated, and the formula is: Wherein, P_pred(t) is the power prediction value, and P_real(t) is the actual power. When the error index exceeds the threshold value, the model parameter adjustment process is started; if the error is within the allowable range, the current parameters are maintained.

[0037] It should be noted that the model parameters of the digital twin model include electrical characteristic parameters such as the short-circuit current temperature coefficient of the photovoltaic module, the open-circuit voltage temperature coefficient, the fill factor, etc.; loss model parameters such as shadow shading loss rate, dust stain attenuation coefficient, cable resistance loss coefficient, etc.; physical model parameters such as the series resistance Rs, parallel resistance Rp and thermal conductivity coefficient in the equivalent circuit of the photovoltaic cell; and environmental response parameters such as the temperature correction coefficient and the illumination nonlinear response factor.

[0038] For a single dominant error scenario, the adjustment direction is determined by calculating the gradient of the error with respect to the parameters; when the error is caused by multiple parameters, a global optimization algorithm (such as genetic algorithm, particle swarm optimization algorithm) is used to search for the optimal parameter combination, with the error minimization as the objective function, and the parameter value range is set to generate the optimal parameter set through iterative calculation. The adjusted parameters need to be verified for effectiveness through historical data backtesting.

[0039] In the embodiments of the present application, as shown in Figure 2 The physical layer is used for hardware foundation and data acquisition, and includes hardware devices of the photovoltaic power station and local processing units; the data layer is used for data storage, processing and management, and includes real-time data high-frequency storage, historical data long-term deposition and real-time processing of streaming data; the model layer includes a digital twin model and a generative model, and realizes digital mapping of physical scenes and intelligent prediction; and the application layer is used to realize business scenario landing.

[0040] The present application realizes accurate spatial mapping of environmental data and electrical parameters by constructing component-level three-dimensional geometric models and binding spatial coordinates, and combines the high-resolution future environmental sequence output by the generative model, so that the power generation simulation of the digital twin model fully integrates physical location characteristics and meteorological evolution rules.

[0041] A dynamic calibration mechanism based on real-time power deviations automatically optimizes the key parameters of the digital twin model, solving the problem of inaccurate predictions in traditional models caused by environmental degradation. At the same time, the generative model significantly enhances its generalization capabilities for extreme weather through pre-training on cross-regional historical data and fine-tuning on target power stations.

[0042] like Figure 3 As shown, the embodiment of the present application also proposes a photovoltaic power generation prediction device based on digital twins, including: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute a photovoltaic power generation prediction method based on digital twins as described in any of the above embodiments.

[0043] An embodiment of the present application also provides a non-volatile computer storage medium storing computer-executable instructions, wherein the computer-executable instructions are configured as: a photovoltaic power generation prediction method based on digital twins as described in any of the above embodiments.

[0044] The various embodiments in this application are described in a progressive manner. Similar portions between the various embodiments can be referred to in conjunction with each other. Each embodiment focuses on the differences between the other embodiments. In particular, the device and medium embodiments are generally similar to the method embodiments, so their descriptions are relatively simple. For relevant portions, refer to the descriptions of the method embodiments.

[0045] The devices and media provided in the embodiments of the present application correspond one-to-one to the methods. Therefore, the devices and media also have similar beneficial technical effects to their corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices and media will not be repeated here.

[0046] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0047] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0048] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0049] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0050] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0051] Memory may include non-permanent storage in a computer-readable medium, in the form of random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0052] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can be implemented using any method or technology to store information. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change RAM (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape, disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media such as modulated data signals and carrier waves.

[0053] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0054] The foregoing is merely an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A photovoltaic power generation prediction method based on digital twins, characterized in that: include: Construct a 3D geometric model of a photovoltaic power station; Obtaining electrical parameters and environmental data of each component in the photovoltaic power station, and inputting the data into the three-dimensional geometric model according to the three-dimensional coordinates corresponding to the components to form a digital twin model; Outputting a sequence of future environmental data within a preset time period based on the environmental data through the trained generative model and inputting the sequence into the digital twin model; Based on the electrical parameters of the component and the future environmental data sequence, power generation simulation is performed through the digital twin model to obtain a power prediction value within the preset time period; The actual power of the photovoltaic power station within the preset time period is obtained, and the model parameters of the digital twin model are adjusted according to the power prediction value and the actual power.

2. A photovoltaic power generation prediction method based on digital twin according to claim 1, characterized in that: The construction of the three-dimensional geometric model of the photovoltaic power station specifically includes: Performing a three-dimensional point cloud scan of the photovoltaic power station to obtain point cloud data including the spatial distribution of each component, and preprocessing the point cloud data; Based on the pre-processed point cloud data, an initial three-dimensional model is generated through a spatial gridding algorithm; Based on the spatial distribution of the photovoltaic power station, a physical space coordinate system is defined to determine the physical space coordinates of the component; The entity space coordinates are converted into model coordinates and mapped into the initial three-dimensional model to form a three-dimensional geometric model.

3. The photovoltaic power generation prediction method based on digital twin according to claim 2 is characterized in that: The step of obtaining electrical parameters and environmental data of each component in the photovoltaic power station and inputting the data into the three-dimensional geometric model according to the three-dimensional coordinates corresponding to the components to form a digital twin model specifically includes: By means of sensor devices deployed in the components, electrical parameters and environmental data are collected in real time to form a multimodal real-time data stream; Determining the three-dimensional coordinates of the component in the three-dimensional geometric model according to the physical space coordinates, and performing spatiotemporal calibration on the multimodal real-time data stream based on the three-dimensional coordinates; The time-space calibrated multimodal real-time data stream is embedded into the three-dimensional geometric model to form a digital twin model.

4. The photovoltaic power generation prediction method based on digital twin according to claim 1, characterized in that: Before outputting a future environmental data sequence within a preset time period based on the environmental data using the trained generative model, the method further includes: Acquire multi-source historical environmental data and pre-process the multi-source historical environmental data; the multi-source historical environmental data includes satellite cloud image data, ground weather station time series data, and component monitoring environmental data; Extracting spatial dimension features of each component data point in the satellite cloud image data; Analyzing the time series period characteristics of each component data point in the ground meteorological station time series data; Based on the spatial dimension features and the temporal period features, a mapping relationship between the spatial position and the timestamp of each component data point is established to form a spatiotemporal correlation matrix; The generative model is trained based on the spatiotemporal correlation matrix, and the training results of the generative model are optimized through preset constraints.

5. The photovoltaic power generation prediction method based on digital twin according to claim 4 is characterized in that: The step of training the generative model based on the spatiotemporal correlation matrix and optimizing the training result of the generative model by using preset constraints specifically includes: Inputting the spatiotemporal correlation matrix into a generative model, predicting the future environmental data of the component data points through the generative model, and outputting a sequence of future environmental data at each time point within a preset future time period; Modifying the component environment prediction parameters according to preset time and space constraints; The corrected future environmental data sequence is compared with the measured environmental data sequence to calculate the prediction deviation, and the model parameters of the generative model are adjusted according to the prediction deviation.

6. The photovoltaic power generation prediction method based on digital twin according to claim 1, characterized in that: The power generation simulation is performed by the digital twin model based on the electrical parameters of the component and the future environmental data sequence to obtain the power prediction value within the preset time period, specifically including: Matching the future environment data sequence to the component model corresponding to the digital twin model; Calculating the standard power generation within the preset time period based on the electrical parameters and the future environmental data sequence; performing a first correction on the standard generated power according to future environmental parameters corresponding to the hourly moments in the future environmental data sequence; The corrected standard power generation power is corrected for the second time using the historical nonlinear influence parameter to obtain the power generation power prediction value corresponding to each hour in the preset time period.

7. The photovoltaic power generation prediction method based on digital twin according to claim 6, characterized in that: The adjusting the model parameters of the digital twin model according to the power prediction value and the actual power specifically includes: Comparing the power prediction value with the actual power to obtain a power deviation; When the power deviation is greater than a preset deviation threshold, parameter optimization is triggered; The weights of the historical nonlinear influencing parameters are iteratively optimized, and an optimized power prediction value is output.

8. The photovoltaic power generation prediction method based on digital twin according to claim 1, characterized in that: After inputting the three-dimensional coordinates corresponding to the components into the three-dimensional geometric model to form a digital twin model, the method further includes: Calculating the index error value of the digital twin model based on multidimensional indicators; the multidimensional indicators include geometric accuracy, physical parameters, and dynamic behavior; Obtaining correction thresholds corresponding to the multidimensional indicators, and determining whether the indicator error values ​​are greater than the corresponding correction thresholds; When the indicator error value is greater than the corresponding correction threshold and the number of the indicator error values ​​is greater than 1, the correction process of the indicator corresponding to the indicator error value is triggered to correct the digital twin model.

9. A photovoltaic power generation prediction device based on digital twins, characterized in that: include: at least one processor; as well as, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute a photovoltaic power generation prediction method based on digital twins as described in any one of claims 1 to 8.

10. A non-volatile computer storage medium storing computer executable instructions, characterized in that: The computer executable instructions are configured to execute a photovoltaic power generation prediction method based on digital twins as described in any one of claims 1 to 8.

Images