Photovoltaic power station cloud layer distribution state prediction method, system, equipment and medium

Abstract

A photovoltaic power station cloud layer distribution state prediction method, system, device and medium relate to the technical field of new energy power generation, and comprise the following steps: obtaining forecast data of current satellite weather, observation data of current irradiator and forecast data of current numerical weather according to the position of a photovoltaic power station group; performing hybrid dynamic analysis based on the three types of acquired data to obtain the distribution states of the current high cloud layer, middle cloud layer and bottom cloud layer; establishing cloud layer change prediction models under different meteorological conditions; in combination with the acquired three types of data and the current distribution state of the cloud layers with different heights, analyzing the change trend of the cloud layers with different heights by a prediction model; respectively predicting the distribution states of the cloud layers with different heights according to the change trends of the cloud layers with different heights; the method has the advantages of high accuracy, high adaptability and high prediction efficiency.

Description

Technical Field

[0001] This invention relates to the field of new energy power generation technology, specifically to a method, system, equipment, and medium for predicting cloud distribution status in photovoltaic power plants. Background Technology

[0002] Because the geographical location, terrain, and meteorological conditions of photovoltaic power stations vary, each photovoltaic power station has its own unique characteristics. Furthermore, the power generation of a photovoltaic power station has a direct linear relationship with the intensity of sunlight. Since the power generation of new energy sources is easily affected by various environmental factors, resulting in large fluctuations in the power generation of new energy sources, it is generally necessary to predict the power generation of new energy power stations in advance, so as to adjust the power grid power in advance based on the predicted power generation, thereby improving the success rate of grid connection.

[0003] For the prediction of photovoltaic power generation, the decisive factor is the solar power. Accurate solar power forecasting relies on accurate mesoscale data as well as the capture and simulation of key information such as electric field topography, grid type, and especially cloud cover. Traditionally, such accurate simulation is achieved based on probabilistic statistical models, artificial neural networks (ANN), and numerical weather prediction (NWP), combined with satellite cloud images. This involves identifying meteorological parameters such as irradiance and cloud trajectories in satellite cloud images to accurately depict cloud formation and dissipation, thereby improving the accuracy of solar power prediction.

[0004] However, while numerical weather prediction (NWP) methods can provide weather forecasts over a longer period of time, their resolution and accuracy may be insufficient in local areas, especially under complex terrain conditions.

[0005] While using cloud images obtained from Earth observation satellites to directly monitor the current cloud distribution can provide a relatively intuitive image of cloud distribution, its ability to predict short-term trends is limited.

[0006] Real-time radiation intensity data is collected by devices such as ground-based irradiators to reflect the impact of clouds on actual power generation. However, this point-based data collection method is difficult to fully cover the entire photovoltaic power station group.

[0007] In summary, traditional methods are mostly limited to the analysis of single-type data sources and lack the ability to comprehensively consider multiple meteorological factors, resulting in limitations in forecast accuracy, adaptability, and efficiency. Summary of the Invention

[0008] The purpose of this invention is to provide a method, system, equipment, and medium for predicting cloud distribution in photovoltaic power plants, which has the advantages of high accuracy, strong adaptability, and high prediction efficiency.

[0009] This invention is achieved through the following technical solution:

[0010] A method for predicting cloud distribution status in photovoltaic power plants includes:

[0011] Based on the location of the photovoltaic power station cluster, obtain current satellite meteorological forecast data, current irradiance observation data, and current numerical weather prediction data;

[0012] Based on the acquired three types of data, a hybrid dynamic analysis is performed to obtain the current distribution status of the high cloud layer, middle cloud layer, and low cloud layer;

[0013] Establish a cloud layer change prediction model under different meteorological conditions;

[0014] By combining the three types of data obtained and the current distribution of clouds at different altitudes, the predictive model analyzes the changing trends of clouds at different altitudes.

[0015] Based on the changing trends of cloud layers at different altitudes, the distribution of cloud layers at different altitudes is predicted.

[0016] Furthermore, the method for obtaining the current satellite meteorological forecast data is as follows: using satellite meteorological measurement data to extract spatiotemporal information about the cloud layer of the photovoltaic power station; the method for obtaining the current irradiance observation data is as follows: based on the irradiance observation data of the station; the method for obtaining the current numerical weather prediction data is as follows: using a numerical weather prediction model.

[0017] Furthermore, the specific steps of the hybrid dynamic analysis include:

[0018] a. Integrate and preprocess the three types of data acquired;

[0019] b. Extract cloud features from the preprocessed data;

[0020] c. Perform feature fusion based on cloud conditions and the three types of data acquired;

[0021] d. Based on the results of feature fusion, and using a time series prediction model, the trend of cloud cover changes over time is obtained;

[0022] e. Based on the trend of cloud layer changes over time, analyze the distribution of cloud layer height and thickness at different altitudes;

[0023] f. Generate a detailed description of the cloud distribution at the current moment.

[0024] Furthermore, the feature fusion employs a Bayesian network or a random forest algorithm.

[0025] Furthermore, the calculation formula for feature fusion using Bayesian networks is as follows:

[0026]

[0027] Among them, X i Parents(X) represents specific variables or characteristics related to cloud distribution. i () refers to those things that directly affect or are related to X i Other variables or characteristics.

[0028] Furthermore, the steps for constructing the prediction model include:

[0029] a. Model cloud layers at different altitudes separately;

[0030] b. Use historical meteorological data and cloud change data to train the trend model, so as to summarize the relationship between cloud changes and meteorological data;

[0031] c. Verify the accuracy of the model using cross-validation, adjust the regression coefficients based on the prediction error, and determine the final trend model.

[0032] Furthermore, the modeling of cloud distribution at different altitudes is performed, and the evolution relationship of the clouds is fitted using multiple regression analysis to construct a cloud change trend model. The calculation formula is as follows:

[0033] C h (t)=α h ·W h (t)+β h R h (t)+γ h T h (t)+ε h

[0034] Among them, C h (t) represents the changing state of the h-th layer of cloud at time t, W h (t), R h (t) and T h (t) represent the wind speed, irradiance, and temperature data in the meteorological conditions, respectively, α h β h and γ h ε is the regression coefficient. h This is the error term.

[0035] A cloud distribution state prediction system for photovoltaic power plants includes a data acquisition module, a dynamic analysis module, a model building module, and a processing module;

[0036] The data acquisition module is used to acquire current satellite meteorological forecast data, current radiometer observation data, and current numerical weather prediction data.

[0037] The dynamic analysis module is used to perform hybrid dynamic analysis based on the data in the data acquisition module to obtain the current distribution status of the high cloud layer, middle cloud layer and low cloud layer;

[0038] The model building module contains cloud change prediction models under different meteorological conditions;

[0039] The processing module is used to perform calculations based on the current distribution data of high, middle and low cloud layers output by the dynamic analysis module, the output data from the data acquisition module, and the model data output by the model building module, to predict the distribution of cloud layers at different altitudes.

[0040] A computer-readable storage medium storing a computer program, characterized in that the program, when executed by a processor, implements the aforementioned method for predicting the cloud distribution state of a photovoltaic power station.

[0041] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable thereon, characterized in that the processor, when executing the program, implements the aforementioned method for predicting the cloud distribution state of a photovoltaic power station.

[0042] The technical solution of the present invention has at least the following advantages and beneficial effects:

[0043] The present invention discloses a method, system, equipment and medium for predicting cloud distribution status in photovoltaic power plants. By performing mixed dynamic analysis on current satellite meteorological forecast data, current irradiance observation data and current numerical weather prediction data, the accuracy of the prediction results can be ensured.

[0044] In addition, a hybrid dynamic analysis dimensionality reduction operation mode was adopted in the process of obtaining the current distribution status of the high cloud layer, middle cloud layer and low cloud layer, which effectively improved the overall calculation efficiency.

[0045] Furthermore, the cloud change prediction model effectively improves the adaptability of this prediction method by using different meteorological conditions as input. Attached Figure Description

[0046] Figure 1 This is a schematic diagram of a method flow of the present invention. Detailed Implementation

[0047] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0048] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0049] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0050] In the description of this invention, it should be noted that if terms such as "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," or "outer" are used to indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product is in use, they are only for the convenience of describing this invention and simplifying the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this invention.

[0051] In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0052] Example 1

[0053] Combined with appendix Figure 1 The method for predicting cloud distribution status of a photovoltaic power station, as shown, includes:

[0054] Based on the location of the photovoltaic power station cluster, obtain current satellite meteorological forecast data, current irradiance observation data, and current numerical weather prediction data;

[0055] Specifically, the method for obtaining the satellite meteorological forecast data is as follows: using satellite meteorological measurement data, extracting spatiotemporal information about the cloud layer of the photovoltaic power station, providing space-based observation data for subsequent analysis, and the data specifically includes high-resolution images obtained from the satellite, providing spatiotemporal information such as cloud coverage and height;

[0056] The method for obtaining the irradiance observation data is as follows: based on the irradiance observation data of the site, the real-time radiation intensity is obtained, reflecting the actual impact of the cloud layer on photovoltaic power generation; and this data is obtained through the real-time radiation intensity data collected by the ground-mounted irradiance instrument, reflecting the actual impact of the current cloud layer on photovoltaic power generation.

[0057] The method for obtaining the numerical weather prediction data is as follows: it is obtained by using a numerical weather prediction model, and the prediction data includes physical parameter data such as temperature and humidity of cloud layers at different altitudes;

[0058] Furthermore, these different types of data sources each provide a unique perspective on cloud conditions and are dynamic in both time and space. For example, satellite images can show the changing trends of clouds over time, while numerical weather prediction can provide possible weather changes in the future, and irradiance observation data directly reflects the degree to which clouds block sunlight and their impact on photovoltaic power generation.

[0059] Based on the acquired three types of data, a hybrid dynamic analysis is performed to obtain the current distribution status of the high cloud layer, middle cloud layer, and low cloud layer;

[0060] The hybrid dynamic analysis steps include:

[0061] a. Integrate and preprocess the three types of data: First, acquire real-time data from multiple data sources (satellite weather forecasts, irradiance observations, and numerical weather predictions); then remove noise and outliers to ensure data quality; next, synchronize data from different sources in time to ensure that all data reflect the state at the same moment; finally, convert the data format to make it suitable for subsequent analysis and processing.

[0062] b. Extract cloud features from preprocessed data: First, identify cloud boundaries and determine the location and shape of clouds; then expand to adjacent pixels based on the initial seed points to form a complete cloud region; then use a convolutional neural network (CNN) to automatically extract complex features, such as cloud texture and shape;

[0063] c. Based on the cloud state and the three types of data obtained, feature fusion is performed: a probabilistic graphical model is constructed based on a Bayesian network to represent the dependencies between variables, and the most likely cloud distribution state is calculated through Bayesian inference.

[0064] The calculation formula for constructing a probabilistic graphical model is as follows:

[0065]

[0066] Among them, X i This refers to a specific variable or characteristic related to the state of cloud distribution, and it can be any one or a combination of the following:

[0067] Cloud height: indicates the height of a certain layer (such as high cloud layer, middle cloud layer, low cloud layer);

[0068] Cloud density: describes the density of clouds in a certain area;

[0069] Cloud movement speed and direction: reflects the speed at which clouds change over time and their direction of movement;

[0070] Meteorological parameters such as temperature and humidity: These physical quantities have a significant impact on the formation and development of clouds;

[0071] Irradiance: The intensity of solar radiation received by the ground, affected by cloud cover;

[0072] Parents (X) i This refers to those things that directly affect or are related to X. i Other variables or characteristics. For example:

[0073] If X i If it is "cloud height", then its parent node could be "temperature", "humidity" and "wind speed", because these factors together determine the height of the cloud.

[0074] If X i It is "cloud density", and its parent node may include "water vapor content", "atmospheric pressure", etc., which are important factors affecting cloud formation;

[0075] For variables like "irradiance", their parent nodes can be "cloud coverage area", "cloud thickness", etc., because they directly determine the amount of solar radiation reaching the ground.

[0076] Specific examples include:

[0077] X1 = cloud height, corresponding to Parents(X1) = {temperature, humidity, wind speed}

[0078] X2 = cloud density, corresponding to Parents(X2) = {water vapor content, atmospheric pressure}

[0079] X3 = Irradiance, corresponding to Parents(X3) = {Cloud coverage area, Cloud thickness}

[0080] d. Based on the results of feature fusion, and using a time series prediction model, the trend of cloud changes over time is obtained: First, LSTM or other time series prediction models are used to capture the trend of cloud changes over time; then, by comparing consecutive frames, motion parameters such as speed and direction of the clouds are calculated; finally, based on historical data and known patterns, the evolution process of clouds under specific conditions is simulated.

[0081] e. Based on the changing trend of cloud layers over time, analyze the distribution of cloud height and thickness at different altitudes: First, divide the atmospheric space into several layers (such as high cloud layer, middle cloud layer, and low cloud layer), and analyze the cloud distribution in each layer; then, combine multi-angle observation data to construct a three-dimensional structure of the cloud layer, and more accurately describe its height, thickness, and distribution.

[0082] f. Generate a detailed description of the cloud distribution at the current moment, including information such as the height, density, and movement speed of each layer. At the same time, by comparing the predicted results with the actual observations, the accuracy of the analysis results is evaluated using indicators such as mean squared error (MSE) and mean absolute error (MAE).

[0083] Establish a cloud layer change prediction model under different meteorological conditions;

[0084] The steps involved in building the prediction model include:

[0085] a. Model cloud layers at different altitudes, such as high-level clouds, middle-level clouds, and low-level clouds;

[0086] Specifically, multiple regression analysis is used to fit the evolution relationship of clouds, that is, to construct a model of cloud change trends. The calculation formula is as follows:

[0087] C h (t)=α h ·W h (t)+β h R h (t)+γ h T h (t)+ε h

[0088] Among them, C h (t) represents the changing state of the h-th layer of cloud at time t, W h (t), R h (t) and T h (t) represent the wind speed, irradiance, and temperature data in the meteorological conditions, respectively, α h β h and γ h ε is the regression coefficient. h This is the error term;

[0089] b. Use historical meteorological data and cloud change data to train the trend model, so as to summarize the relationship between cloud changes and meteorological data;

[0090] c. Verify the accuracy of the model using cross-validation, adjust it based on the prediction error, optimize the regression coefficients, and determine the final trend model;

[0091] It is important to note that the cross-validation method uses current meteorological data to validate the trained trend model, thereby determining the final trend model.

[0092] By combining the three types of data obtained and the current distribution of clouds at different altitudes, the predictive model analyzes the changing trends of clouds at different altitudes.

[0093] Specifically, time series forecasting or Markov chains are used for trend analysis to estimate future cloud distribution. The Markov chain method describes state transition probabilities and is based on the Markov assumption that the future state of a system depends only on the current state and not on past historical states. The formula for calculating a Markov chain is:

[0094]

[0095] P(C h (t+1)|C h (t) is the cloud state C at the current time t. h Given (t), predict the cloud state C at the next time step t+1. h The probability of (t+1) indicates the corresponding trend of cloud layer change; P(C h (t) = i) is the probability that the current time t is in state i;

[0096] Since Markov chains assume that the future state of a cloud depends only on the current state and is not affected by past states, the modeling process is simplified. In the prediction of long-term trends of cloud changes, this means that we can directly predict the future state through the current cloud state and its transition probability without dealing with complex historical data.

[0097] Furthermore, since Markov chains only focus on the current state and transition probabilities, they avoid the complexity of needing to model historical states in detail, thereby reducing the difficulty of computation and data processing.

[0098] In addition, Markov chains can handle time series data very well, and are particularly suitable for modeling systems such as cloud distribution that change over time. Therefore, when predicting the trend of cloud changes, Markov chains can be used to dynamically capture the changing patterns of cloud state over time.

[0099] By establishing a state transition matrix, Markov chains can predict the cloud distribution state after multiple time steps, thereby providing long-term trends. This is particularly important for the operation and management of photovoltaic power plants, as they rely on long-term irradiance prediction.

[0100] Furthermore, cloud changes are a stochastic process that can be influenced by various factors (such as weather, temperature, humidity, etc.). Markov chains describe the state transitions of the system through probabilistic modeling, which can effectively handle this uncertainty. For example, cloud changes are not absolute, but rather transition from one state to another with a certain probability, which makes predictions more flexible and realistic.

[0101] Furthermore, by predicting the probability distribution of different cloud states, Markov chain models can help quantify the likelihood of different cloud changes and provide more accurate risk assessments, especially in weather-sensitive scenarios such as photovoltaic power plants.

[0102] It is important to note that Markov chains do not require a large number of external variables, but rely on historical cloud state transition data to estimate the transition probability. Therefore, with sufficient historical data, we can build an accurate state transition model and then predict the future distribution of clouds.

[0103] As new meteorological data is continuously collected, the Markov chain model can gradually adjust its prediction results by updating the transition matrix. This allows the model to adapt to new data and has strong flexibility and self-updating ability.

[0104] In particular, Markov chains can be extended to multiple levels to adapt to changes in cloud layers at different altitudes. For example, high clouds, mid clouds, and low clouds can be viewed as different state spaces and modeled using their respective transition matrices; this allows the model to capture the changing trends of cloud layers at different altitudes with finer granularity, helping to more accurately predict changes in irradiance at photovoltaic power plants.

[0105] Based on the changing trends of cloud layers at different altitudes, predict the distribution of cloud layers at different altitudes.

[0106] Based on the changing trends of clouds at different altitudes and combined with the timeline, the distribution of clouds at different altitudes on the future timeline can be predicted.

[0107] The predicted distribution can then be used to predict the power generation of subsequent photovoltaic power plants.

[0108] Example 2

[0109] A cloud distribution state prediction system for photovoltaic power plants includes a data acquisition module, a dynamic analysis module, a model building module, and a processing module;

[0110] The data acquisition module is used to acquire current satellite meteorological forecast data, current radiometer observation data, and current numerical weather prediction data.

[0111] The dynamic analysis module is used to perform hybrid dynamic analysis based on the data in the data acquisition module to obtain the current distribution status of the high cloud layer, middle cloud layer and low cloud layer;

[0112] The model building module contains cloud change prediction models under different meteorological conditions;

[0113] The processing module is used to perform calculations based on the current distribution data of high, middle and low cloud layers output by the dynamic analysis module, the output data in the data acquisition module, and the model data output by the model building module, to predict the distribution of cloud layers at different altitudes.

[0114] In addition, during the analysis in the dynamic analysis module, this system can invert the distribution of different altitudes (high, medium, and low) and different types of clouds based on the current satellite meteorological forecast data. Then, based on the current numerical weather prediction data, it can obtain the differences in physical quantities of clouds at different altitudes. Combined with cloud identification technology, it can invert and analyze parameters such as cloud morphology, altitude, and movement speed. Finally, it can evaluate the analysis results in conjunction with numerical weather forecast results.

[0115] Furthermore, during dynamic analysis, feature engineering can be used to automatically identify cloud distribution in meteorological data. When performing feature engineering, the physical laws of cloud movement and cloud formation and dissipation are combined. Then, a machine learning model of cloud-related physical quantities in numerical weather prediction based on satellite cloud images is established. This can further improve the accuracy of cloud cover and other data in numerical weather prediction, and improve the prediction accuracy of cloud distribution. In addition, considering the impact of clouds on irradiance, irradiance predictions can be revised by combining irradiance observation data from the field stations.

[0116] In its predictive applications, this system forecasts cloud distribution patterns in a spatiotemporal sequence, thereby calibrating cloud cover predictions for photovoltaic (PV) power plant clusters in advance. This data is then applied to actual PV power plant operations, further enhancing the ultra-short-term power prediction capabilities of on-site PV power plants. By monitoring and analyzing cloud distribution patterns in real time, PV power plants can adjust their power generation strategies and proactively prevent potential problems such as low power generation efficiency and equipment damage. Furthermore, the prediction results can provide a reference for energy dispatch and grid stability. Accurate prediction of cloud distribution patterns is of significant value for grid dispatch and management in large-scale PV power plant clusters.

[0117] Example 3

[0118] A computer-readable storage medium storing a computer program, characterized in that the program, when executed by a processor, implements the aforementioned method for predicting the cloud distribution state of a photovoltaic power station.

[0119] Example 4

[0120] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable thereon, characterized in that the processor, when executing the program, implements the aforementioned method for predicting the cloud distribution state of a photovoltaic power station.

[0121] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for predicting cloud distribution status in a photovoltaic power station, characterized by: include: According to the location of the photovoltaic power station group, obtain the current satellite weather forecast data, the current irradiator observation data and the current numerical weather forecast data; Based on the three types of data obtained, a hybrid dynamic analysis is performed to obtain the current distribution status of high cloud layer, middle cloud layer and bottom cloud layer; Establish a cloud change prediction model under different meteorological conditions; Combining the three types of data obtained and the current distribution status of clouds at different altitudes, the prediction model analyzes the changing trends of clouds at different altitudes; According to the changing trends of clouds at different altitudes, the distribution status of clouds at different altitudes is predicted respectively.

2. The method for predicting cloud distribution state of a photovoltaic power station according to claim 1, characterized in that: The method for obtaining the current satellite meteorological forecast data is: using satellite meteorological measured data to extract the temporal and spatial information about the cloud layer of the photovoltaic power station; the method for obtaining the current irradiator observation data is: obtaining it based on the site irradiator observation data; the method for obtaining the current numerical weather prediction data is: obtaining it using a numerical weather forecast model.

3. The method for predicting cloud distribution state of a photovoltaic power station according to claim 1, characterized in that: The specific steps of the hybrid dynamic analysis include: a. Integrate and preprocess the three types of data obtained; b. Extract cloud features from preprocessed data; c. Perform feature fusion based on the cloud status and the three types of data obtained; d. Based on the results of feature fusion, the time series prediction model is used to obtain the change trend of cloud layer over time; e. Analyze the height and thickness distribution of clouds at different altitudes based on the changing trend of clouds over time; f. Generate a detailed description of the cloud distribution at the current moment.

4. The method for predicting cloud distribution state of a photovoltaic power station according to claim 3, characterized in that: The feature fusion adopts Bayesian network or random forest algorithm.

5. The method for predicting cloud distribution state of a photovoltaic power station according to claim 4, characterized in that: The calculation formula for feature fusion using Bayesian network is: Among them, X i Represents specific variables or features related to the cloud distribution state, Parents(X i ) refers to those that directly affect or are related to X i other variables or features.

6. The method for predicting cloud distribution state of a photovoltaic power station according to claim 1, characterized in that: The steps of constructing the prediction model include: a. Model clouds at different heights separately; b. Use historical meteorological data and cloud change data to train the change trend model, so as to summarize the relationship between cloud changes and meteorological data; c. Verify the accuracy of the model through cross-validation method, make adjustments based on the prediction error, optimize the regression coefficient, and determine the final change trend model.

7. The method for predicting cloud distribution state of a photovoltaic power station according to claim 6, characterized in that: The cloud distribution at different heights is modeled, and the evolution relationship of the cloud layer is fitted using a multivariate regression analysis method to construct a cloud layer change trend model. The calculation formula is: C h (t)=a h ·W h (t)+β h R h (t)+γ h T h (t)+e h Among them, C h (t) is the changing state of the h-th layer of clouds at time t, W h (t), R h (t) and T h (t) represent the wind speed, radiation intensity and temperature data in meteorological conditions, α h , β h and γ h is the regression coefficient, ε h is the error term.

8. A photovoltaic power station cloud distribution state prediction system, including a data acquisition module, a dynamic analysis module, a model building module and a processing module; The data acquisition module is used to acquire the current satellite meteorological forecast data, the current irradiator observation data and the current numerical weather forecast data; The dynamic analysis module is used to perform hybrid dynamic analysis based on the data in the data acquisition module to obtain the current distribution status of the high cloud layer, the middle cloud layer and the bottom cloud layer; The model building module is written with a cloud layer change prediction model under different meteorological conditions; The processing module is used to perform calculations and processing based on the distribution status data of the current high cloud layer, middle cloud layer and bottom cloud layer output by the dynamic analysis module and the output data in the data acquisition module, combined with the model data output by the model construction module, to predict the distribution status of clouds at different heights.

9. A computer-readable storage medium storing a computer program, characterized in that: When the program is executed by a processor, the method for predicting cloud distribution status in a photovoltaic power station as described in any one of claims 1 to 7 is implemented.

10. An electronic device comprising a memory, a processor and a computer program stored in the memory and executable, characterized in that: When the processor executes the program, the method for predicting cloud distribution status in a photovoltaic power station as described in any one of claims 1-7 is implemented.

Images