Photovoltaic power characteristic difference prediction method and system under different weather conditions and medium

Abstract

The embodiment of the invention discloses a photovoltaic power characteristic difference prediction method and system under different weather conditions and a medium. The method comprises the following steps: determining thermodynamic diagram data carrying dust type distribution information according to particle size analysis data; determining airflow turbulence characteristic parameters based on the three-dimensional velocity field data and the pressure distribution data; constructing a dust removal effect model according to the thermodynamic diagram data and the airflow turbulence characteristic parameters, and establishing a mapping relation between turbulence intensity and a dust removal effect; a power recovery speed prediction model is constructed according to the power output data of the photovoltaic panel under different dust removal effects, and a mapping relation between the dust removal effects and the power recovery speed is established; constructing a quantitative relation model under the severe convection weather condition based on the mapping relation; and according to the quantitative relation model and a preset photovoltaic power prediction model under a non-severe convection weather condition, constructing a photovoltaic power characteristic difference prediction model to perform photovoltaic power prediction under different weather conditions, thereby realizing accurate prediction of power recovery of the photovoltaic power station under severe convection weather.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power prediction, and in particular to a method, system and medium for predicting photovoltaic power characteristic differences under different weather conditions. Background Art

[0002] The power generation efficiency of photovoltaic power stations will be significantly affected under different weather conditions. Especially in strong convective weather, dust deposition on the surface of photovoltaic panels may be more serious, resulting in a significant drop in photovoltaic power. Although strong convective weather will bring strong turbulence, the influence of turbulence on the removal of dust on the surface of photovoltaic panels is still unclear.

[0003] The intensity, scale, frequency and other characteristics of turbulence will affect the dust removal effect, but the quantitative relationship still needs to be studied in depth. In addition, the turbulent characteristics of the airflow around the photovoltaic panels are closely related to the arrangement, spacing, inclination and other factors of the photovoltaic panels. Different photovoltaic power station design schemes will lead to different turbulent characteristics. Traditional research often only focuses on one aspect of the influencing factors and lacks systematic integrated analysis. Therefore, it is urgent to carry out research on the mechanism of dust deposition and removal on the surface of photovoltaic panels under severe convective weather, reveal the quantitative relationship between turbulent characteristics and dust removal effect, and optimize the design of photovoltaic power stations accordingly to improve photovoltaic power generation efficiency. Summary of the invention

[0004] Based on this, it is necessary to propose a method, system and medium for predicting the differences in photovoltaic power characteristics under different weather conditions to address the above problems.

[0005] A method for predicting photovoltaic power characteristics under different weather conditions, the method comprising:

[0006] Obtain particle size analysis data of dust samples on the surface of photovoltaic panels, measure the dust samples through the particle size analysis data and emit an intensity signal reflecting the chemical composition of the dust, and determine thermal map data carrying dust type distribution information based on the particle size analysis data and the intensity signal.

[0007] The three-dimensional velocity field data of the airflow around the photovoltaic panel and the pressure distribution data of the photovoltaic panel surface and the surrounding area are obtained, and the airflow around the photovoltaic panel is numerically simulated based on the three-dimensional velocity field data and the pressure distribution data to determine the airflow turbulence characteristic parameters.

[0008] A dust removal effect model is constructed according to the thermal map data and the airflow turbulence characteristic parameters, the dust removal effect is predicted by the dust removal effect model, and a mapping relationship between turbulence intensity and dust removal effect is established.

[0009] A power recovery speed prediction model is constructed based on power output data of photovoltaic panels under different dust removal effects, the power recovery speed is determined by the power recovery speed prediction model, and a mapping relationship between the dust removal effect and the power recovery speed is established.

[0010] Based on the mapping relationship between the turbulence intensity and the dust removal effect and the mapping relationship between the dust removal effect and the power recovery speed, a quantitative relationship model under severe convective weather conditions is constructed according to the turbulence intensity, dust removal effect and photovoltaic power recovery speed in the historical operation data, and the photovoltaic power recovery speed is predicted by the quantitative relationship model.

[0011] Feature extraction is performed on the quantitative relationship model under severe convective weather conditions and the preset photovoltaic power prediction model under non-severe convective weather conditions to obtain model feature vectors, and a photovoltaic power characteristic difference prediction model is constructed based on historical weather data and the model feature vectors. Short-term, medium-term and long-term predictions of photovoltaic power under different weather conditions are performed using the photovoltaic power characteristic difference prediction model.

[0012] The step of obtaining particle size analysis data of dust samples on the surface of photovoltaic panels, measuring dust samples through the particle size analysis data and emitting intensity signals reflecting the chemical composition of dust, and determining thermal map data carrying dust type distribution information according to the particle size analysis data and the intensity signals specifically includes:

[0013] Obtain particle size analysis data for dust samples from photovoltaic panel surfaces.

[0014] An element analysis instruction is sent according to the particle size analysis data, a dust sample is measured by the element analysis instruction, and an intensity signal reflecting the chemical composition of the dust is sent.

[0015] The particle size analysis data and the intensity signal are converted into feature vectors.

[0016] It is determined whether there is a classification hyperplane matching the feature vector in the support vector machine model library.

[0017] If there is a classification hyperplane matching the feature vector in the support vector machine model library, the feature vector is classified to determine the thermal map data carrying the dust type distribution information.

[0018] If the feature vector does not have a classification hyperplane matching the feature vector in the support vector machine model library, the feature vector will not be classified.

[0019] The method of acquiring three-dimensional velocity field data of the airflow around the photovoltaic panel and pressure distribution data on the surface of the photovoltaic panel and the surrounding area, and numerically simulating the airflow around the photovoltaic panel based on the three-dimensional velocity field data and the pressure distribution data to determine the airflow turbulence characteristic parameters specifically includes:

[0020] The airflow around the photovoltaic panel is sampled at high frequency to obtain the three-dimensional velocity field data of the airflow around the photovoltaic panel.

[0021] The pressure distribution data on the surface of the photovoltaic panel and the surrounding area is measured by a pressure sensor array.

[0022] The three-dimensional velocity field data and pressure distribution data are preprocessed.

[0023] A calculation domain is constructed according to the preprocessed three-dimensional velocity field data and pressure distribution data.

[0024] The airflow around the photovoltaic panel is numerically simulated in the computational domain by using a large eddy simulation method;

[0025] Determine whether the numerical simulation result meets the preset conditions.

[0026] If the numerical simulation results meet the preset conditions, the airflow turbulence characteristic parameters are determined according to the three-dimensional velocity field data.

[0027] If the numerical simulation result does not meet the preset conditions, the numerical simulation is performed again.

[0028] The step of building a dust removal effect model based on the thermal map data and airflow turbulence characteristic parameters, predicting the dust removal effect through the dust removal effect model, and establishing a mapping relationship between turbulence intensity and dust removal effect specifically includes:

[0029] The dust distribution features in the thermal map data are extracted, and the dust distribution features and airflow turbulence characteristic parameters are subjected to standard deviation standardization to obtain a standardized feature set.

[0030] The Pearson correlation coefficient is used to analyze the correlation between the standardized feature set and the dust removal effect, and feature data with an absolute value of the correlation coefficient greater than a preset threshold is selected as key feature data.

[0031] A dust removal effect model is constructed according to the key feature data.

[0032] The characteristic data of turbulence intensity within a preset range is used as input of the dust removal effect model, and the corresponding dust removal effect is predicted by the dust removal effect model.

[0033] A cubic spline interpolation method is used to obtain a continuous mapping function between the turbulence intensity and the corresponding dust removal effect.

[0034] A relationship curve is drawn through the continuous mapping function to determine the mapping relationship between the turbulence intensity and the dust removal effect.

[0035] The power recovery speed prediction model is constructed according to the power output data of the photovoltaic panel under different dust removal effects, the power recovery speed is determined by the power recovery speed prediction model, and the mapping relationship between the dust removal effect and the power recovery speed is established, which specifically includes:

[0036] Obtain power output data of photovoltaic panels under different dust removal effects.

[0037] An abnormal value detection instruction is sent according to the power output data, an abnormal point is identified according to the abnormal value detection instruction, and an interpolation process is performed on the identified abnormal point.

[0038] The power output data after the interpolation processing is pre-processed.

[0039] The pre-processed power output data generates trend terms, seasonal terms and residual terms through a time series decomposition method, and the trend terms, seasonal terms and residual terms are used as time series decomposition results.

[0040] According to the existence of the time series decomposition result in the autocorrelation function database, it is determined that the time series decomposition result meets the modeling condition.

[0041] A power recovery speed prediction model is constructed according to the time series decomposition result.

[0042] The power recovery speed is determined according to the power recovery speed prediction model.

[0043] Support vector regression is used to establish the mapping relationship between dust removal effect and power recovery speed.

[0044] Wherein, determining, according to the existence of the time series decomposition result in the autocorrelation function database, that the time series decomposition result meets the modeling condition specifically includes:

[0045] An autocorrelation function database is queried, wherein the autocorrelation function database contains function data with a range of model orders.

[0046] It is determined whether there is the function data matching the time series decomposition result in the autocorrelation function database.

[0047] If the function data matching the time series decomposition result exists in the autocorrelation function database, it is determined that the time series decomposition result meets the modeling condition.

[0048] If the function data matching the time series decomposition result does not exist in the autocorrelation function database, it is determined that the time series decomposition result does not meet the modeling condition.

[0049] Among them, based on the mapping relationship between the turbulence intensity and the dust removal effect and the mapping relationship between the dust removal effect and the power recovery speed, a quantitative relationship model under severe convective weather conditions is constructed according to the turbulence intensity, dust removal effect and photovoltaic power recovery speed in historical operation data, and the photovoltaic power recovery speed is predicted by the quantitative relationship model, specifically including:

[0050] The historical operation data of the photovoltaic power station is obtained, wherein the historical operation data includes turbulence intensity, dust removal effect and photovoltaic power recovery speed.

[0051] The historical operation data are processed to extract key features of the turbulence intensity, dust removal effect and photovoltaic power recovery speed.

[0052] An input feature vector and an output label are constructed according to the key features, the input feature matrix is ​​the turbulence intensity and dust removal effect, and the output label vector is the photovoltaic power recovery speed.

[0053] Based on the mapping relationship between the turbulence intensity and the dust removal effect and the mapping relationship between the dust removal effect and the power recovery speed, a quantitative relationship model under severe convective weather conditions is constructed according to the input feature matrix and the output label vector, and the photovoltaic power recovery speed is predicted by the quantitative relationship model.

[0054] Among them, based on the mapping relationship between the turbulence intensity and the dust removal effect and the mapping relationship between the dust removal effect and the power recovery speed, according to the input feature matrix and the output label vector, a quantitative relationship model under severe convective weather conditions is constructed, and the photovoltaic power recovery speed is predicted by the quantitative relationship model, specifically including:

[0055] Based on the mapping relationship between the turbulence intensity and the dust removal effect and the mapping relationship between the dust removal effect and the power recovery speed, the quantitative relationship model under severe convective weather conditions is determined according to P=αI+βD, wherein P is the photovoltaic power recovery speed, I is the turbulence intensity, D is the dust removal effect, and α and β are model parameters.

[0056] Wherein, the method of establishing a quantitative relationship model among turbulence intensity, dust removal effect and photovoltaic power recovery speed according to the input feature vector and the output label, and predicting the photovoltaic power recovery speed by the quantitative relationship model, further specifically includes:

[0057] The photovoltaic power output curve is determined according to the predicted photovoltaic power recovery speed and in combination with the initial power and area information of the photovoltaic panel.

[0058] The photovoltaic power output curve is used to determine the power recovery time under different turbulence intensities.

[0059] A cubic spline interpolation method is adopted to establish a corresponding relationship between turbulence intensity and power recovery time based on the power recovery time under different turbulence intensities.

[0060] The layout scheme of different photovoltaic panels and the distribution of turbulence intensity under installation angles are determined based on actual meteorological data and terrain information.

[0061] The power recovery performance under each layout scheme and installation angle is evaluated according to the corresponding relationship between the turbulence intensity distribution and the power recovery time, and the layout schemes and installation angles of the different photovoltaic panels are optimized.

[0062] The method comprises the following steps: extracting features from the quantitative relationship model under severe convective weather conditions and the preset photovoltaic power prediction model under non-severe convective weather conditions to obtain a model feature vector, constructing a photovoltaic power characteristic difference prediction model based on historical weather data and the model feature vector, and performing short-term, medium-term and long-term predictions of photovoltaic power under different weather conditions through the photovoltaic power characteristic difference prediction model, specifically including:

[0063] The K-means clustering algorithm is used to classify the historical weather data to obtain multiple weather categories, and a weather feature vector is constructed according to the characteristic parameters of each weather category.

[0064] Feature extraction is performed on the quantitative relationship model under severe convective weather conditions and the preset photovoltaic power prediction model under non-severe convective weather conditions to obtain model feature vectors.

[0065] A photovoltaic power characteristic difference prediction model is constructed according to the weather characteristic vector and the model characteristic vector.

[0066] The real-time weather data is input into the photovoltaic power characteristic difference prediction model, and the photovoltaic power is predicted in the short term, medium term and long term by the sliding time window method to obtain the final photovoltaic power prediction value.

[0067] The step of inputting the real-time weather data into the photovoltaic power characteristic difference prediction model, performing short-term prediction, medium-term prediction and long-term prediction on the photovoltaic power by the sliding time window method, and obtaining the final photovoltaic power prediction value further specifically includes:

[0068] The photovoltaic power characteristics under different weather conditions are determined according to the photovoltaic power prediction value.

[0069] The difference between the photovoltaic power prediction value and the photovoltaic power average prediction value is used to analyze the influence of each feature in the weather feature vector and the model feature vector on the photovoltaic power prediction result.

[0070] A photovoltaic power characteristic difference prediction system under different weather conditions, the system comprising:

[0071] The thermal map data determination module is used to obtain the particle size analysis data of the dust sample on the surface of the photovoltaic panel, measure the dust sample through the particle size analysis data and send out an intensity signal reflecting the chemical composition of the dust, and determine the thermal map data carrying the dust type distribution information based on the particle size analysis data and the intensity signal.

[0072] The airflow turbulence characteristic parameter determination module is used to obtain the three-dimensional velocity field data of the airflow around the photovoltaic panel and the pressure distribution data on the surface of the photovoltaic panel and the surrounding area, and to perform numerical simulation of the airflow around the photovoltaic panel based on the three-dimensional velocity field data and the pressure distribution data to determine the airflow turbulence characteristic parameters.

[0073] A module for establishing a mapping relationship between turbulence intensity and dust removal effect is used to construct a dust removal effect model based on the thermal map data and the airflow turbulence characteristic parameters, predict the dust removal effect through the dust removal effect model, and establish a mapping relationship between turbulence intensity and the dust removal effect.

[0074] A module for establishing a mapping relationship between dust removal effect and power recovery speed is used to construct a power recovery speed prediction model based on the power output data of the photovoltaic panel under different dust removal effects, determine the power recovery speed through the power recovery speed prediction model, and establish a mapping relationship between the dust removal effect and the power recovery speed.

[0075] A quantitative relationship model building module is used to build a quantitative relationship model under severe convective weather conditions based on the mapping relationship between the turbulence intensity and the dust removal effect and the mapping relationship between the dust removal effect and the power recovery speed, according to the turbulence intensity, dust removal effect and photovoltaic power recovery speed in the historical operation data, and predict the photovoltaic power recovery speed through the quantitative relationship model.

[0076] The photovoltaic power prediction module is used to extract features from the quantitative relationship model under severe convective weather conditions and the preset photovoltaic power prediction model under non-severe convective weather conditions, obtain model feature vectors, construct a photovoltaic power characteristic difference prediction model based on historical weather data and the model feature vectors, and use the photovoltaic power characteristic difference prediction model to perform short-term, medium-term and long-term predictions of photovoltaic power under different weather conditions.

[0077] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor executes the steps of the method described above.

[0078] The embodiments of the present invention have the following beneficial effects:

[0079] The present invention can understand the size distribution of dust particles in detail by performing particle size analysis on dust samples on the surface of photovoltaic panels. The dust samples are measured through particle size analysis data and an intensity signal reflecting the chemical composition of the dust is emitted. Measuring the chemical composition of the dust helps to identify specific components that may affect the performance and life of the photovoltaic panels, and provides a basis for targeted cleaning strategies. The thermal map data carrying the dust type distribution information is determined based on the particle size analysis data and the intensity signal, which can intuitively display the distribution of dust on the surface of the photovoltaic panel, and help to accurately evaluate the degree of pollution in different areas. Furthermore, the airflow around the photovoltaic panel is numerically simulated to determine the turbulence characteristic parameters, and a dust removal effect model is constructed based on the thermal map data and the airflow turbulence characteristic parameters. The model can quantify the dust removal effect under different turbulence intensities, providing a scientific basis for designing an efficient cleaning system or using natural wind for cleaning. The dust removal effect is predicted by the dust removal effect model, and a mapping relationship between turbulence intensity and dust removal effect is established. Furthermore, a power recovery speed prediction model is constructed based on the power output data of photovoltaic panels under different dust removal effects. The model directly links the dust removal effect with the recovery speed of photovoltaic panel power output, so that the power generation capacity recovery time of photovoltaic panels after cleaning can be accurately predicted. The power recovery speed is determined by the power recovery speed prediction model, and a mapping relationship between dust removal effect and power recovery speed is established. Furthermore, by integrating the mapping relationship between turbulence intensity, dust removal effect and power recovery speed, a quantitative relationship model under severe convective weather conditions is constructed based on turbulence intensity, dust removal effect and photovoltaic power recovery speed in historical operation data. The model can accurately predict the recovery speed of photovoltaic power under severe convective weather conditions, which is crucial for coping with extreme weather conditions, ensuring grid stability and improving energy supply reliability. Furthermore, the quantitative relationship model under severe convective weather conditions and the preset photovoltaic power prediction model under non-severe convective weather conditions are feature extracted to obtain model feature vectors, and a photovoltaic power characteristic difference prediction model is constructed based on historical weather data and model feature vectors. The model combines the photovoltaic power characteristics under different weather conditions and can achieve short-term, medium-term and long-term power predictions. Through comprehensive analysis with the preset photovoltaic power prediction model under non-severe convective weather conditions, a photovoltaic power characteristic difference prediction model applicable to various weather conditions is constructed. The present invention realizes the accurate prediction of power recovery of photovoltaic power stations under severe convective weather, and provides technical support for improving power generation efficiency and stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0081] in:

[0082] Figure 1 A schematic diagram of a flow chart of an embodiment of a method for predicting differences in photovoltaic power characteristics under different weather conditions;

[0083] Figure 2 A schematic diagram of a flow chart of another embodiment of a method for predicting differences in photovoltaic power characteristics under different weather conditions;

[0084] Figure 3 A schematic flow chart of another embodiment of a method for predicting differences in photovoltaic power characteristics under different weather conditions;

[0085] Figure 4 A schematic flow chart of another embodiment of a method for predicting differences in photovoltaic power characteristics under different weather conditions;

[0086] Figure 5 A schematic diagram of the structure of an embodiment of a photovoltaic power characteristic difference prediction system under different weather conditions;

[0087] Figure 6 A schematic structural diagram of an embodiment of the medium provided by the present invention. DETAILED DESCRIPTION

[0088] 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.

[0089] like Figure 1 As shown, Figure 1 The flowchart of an embodiment of a method for predicting the difference of photovoltaic power characteristics under different weather conditions is shown in FIG. A method for predicting the difference of photovoltaic power characteristics under different weather conditions includes:

[0090] S101: Obtain particle size analysis data of dust samples on the surface of photovoltaic panels, measure the dust samples through the particle size analysis data and send out intensity signals reflecting the chemical composition of the dust, and determine thermal map data carrying dust type distribution information based on the particle size analysis data and the intensity signal.

[0091] Exemplarily, the particle size analysis data of dust samples on the surface of photovoltaic panels are obtained; an element analysis instruction is sent according to the particle size analysis data, the dust sample is measured by the element analysis instruction, and an intensity signal reflecting the chemical composition of the dust is sent; the particle size analysis data and the intensity signal are converted into feature vectors; it is determined whether there is a classification hyperplane matching the feature vector in the support vector machine model library; if there is a classification hyperplane feature vector matching the feature vector in the support vector machine model library, the feature vector is classified to determine the thermal map data carrying the dust type distribution information; if there is no classification hyperplane feature vector matching the feature vector in the support vector machine model library, the feature vector is not classified.

[0092] S102: Acquire three-dimensional velocity field data of the airflow around the photovoltaic panel and pressure distribution data on the surface of the photovoltaic panel and the surrounding area, perform numerical simulation on the airflow around the photovoltaic panel based on the three-dimensional velocity field data and the pressure distribution data, and determine the airflow turbulence characteristic parameters.

[0093] Exemplarily, high-frequency sampling is performed on the airflow around the photovoltaic panel to obtain three-dimensional velocity field data of the airflow around the photovoltaic panel; pressure distribution data of the photovoltaic panel surface and surrounding areas are measured by a pressure sensor array; the three-dimensional velocity field data and pressure distribution data are preprocessed; a calculation domain is constructed based on the preprocessed three-dimensional velocity field data and pressure distribution data; the airflow around the photovoltaic panel is numerically simulated in the calculation domain by a large eddy simulation method; it is determined whether the numerical simulation results meet the preset conditions; if the numerical simulation results meet the preset conditions, the airflow turbulence characteristic parameters are determined based on the three-dimensional velocity field data; if the numerical simulation results do not meet the preset conditions, the numerical simulation is performed again.

[0094] S103: constructing a dust removal effect model according to the thermal map data and the airflow turbulence characteristic parameters, predicting the dust removal effect through the dust removal effect model, and establishing a mapping relationship between the turbulence intensity and the dust removal effect.

[0095] Exemplarily, dust distribution features in thermal map data are extracted, and standard deviation normalization is performed on dust distribution features and airflow turbulence characteristic parameters to obtain a standardized feature set; the Pearson correlation coefficient is used to analyze the correlation between the standardized feature set and the dust removal effect, and feature data with an absolute value of the correlation coefficient greater than a preset threshold is selected as key feature data; a dust removal effect model is constructed based on the key feature data; the feature data of turbulence intensity within a preset range is used as input to the dust removal effect model, and the corresponding dust removal effect is predicted through the dust removal effect model; the cubic spline interpolation method is used to obtain a continuous mapping function between turbulence intensity and the corresponding dust removal effect; a relationship curve is drawn through the continuous mapping function to determine the mapping relationship between turbulence intensity and dust removal effect.

[0096] S104: constructing a power recovery speed prediction model according to power output data of the photovoltaic panel under different dust removal effects, determining the power recovery speed through the power recovery speed prediction model, and establishing a mapping relationship between the dust removal effect and the power recovery speed.

[0097] Exemplarily, power output data of photovoltaic panels under different dust removal effects are obtained; an outlier detection instruction is sent according to the power output data, anomalies are identified according to the outlier detection instruction, and the identified anomalies are interpolated; the power output data after interpolation is preprocessed; the preprocessed power output data generates trend terms, seasonal terms and residual terms through a time series decomposition method, and the trend terms, seasonal terms and residual terms are used as time series decomposition results; according to the existence of the time series decomposition results in an autocorrelation function database, it is determined that the time series decomposition results meet the modeling conditions; a power recovery speed prediction model is constructed according to the time series decomposition results; the power recovery speed is determined according to the power recovery speed prediction model; and support vector regression is used to establish a mapping relationship between dust removal effect and power recovery speed.

[0098] S105: Based on the mapping relationship between turbulence intensity and dust removal effect and the mapping relationship between dust removal effect and power recovery speed, a quantitative relationship model under severe convective weather conditions is constructed according to the turbulence intensity, dust removal effect and photovoltaic power recovery speed in historical operation data, and the photovoltaic power recovery speed is predicted by the quantitative relationship model.

[0099] Exemplarily, historical operation data of a photovoltaic power station is obtained, the historical operation data including turbulence intensity, dust removal effect and photovoltaic power recovery speed; the historical operation data is processed to extract key features of turbulence intensity, dust removal effect and photovoltaic power recovery speed; input feature vectors and output labels are constructed according to the key features, the input feature matrix is ​​turbulence intensity and dust removal effect, and the output label vector is the photovoltaic power recovery speed; based on the mapping relationship between turbulence intensity and dust removal effect and the mapping relationship between the dust removal effect and power recovery speed, a quantitative relationship model under severe convective weather conditions is constructed according to the input feature matrix and the output label vector, and the photovoltaic power recovery speed is predicted by the quantitative relationship model.

[0100] Specifically, the quantitative relationship model under severe convective weather conditions is shown in the following formula:

[0101] P = αI + βD;

[0102] Among them, P is the photovoltaic power recovery rate, I is the turbulence intensity, D is the dust removal effect, and α and β are model parameters.

[0103] Furthermore, the photovoltaic power output curve is determined based on the predicted photovoltaic power recovery speed and combined with the initial power and area information of the photovoltaic panels. The power recovery time under different turbulence intensities is determined using the photovoltaic power output curve. The corresponding relationship between turbulence intensity and power recovery time is established based on the power recovery time under different turbulence intensities using the cubic spline interpolation method. The turbulence intensity distribution under different layout schemes and installation angles of photovoltaic panels is determined based on actual meteorological data and terrain information. The power recovery performance under each layout scheme and installation angle is evaluated based on the corresponding relationship between turbulence intensity distribution and power recovery time.

[0104] According to the predicted photovoltaic power recovery speed, the impact of turbulence on the power recovery of photovoltaic power stations under severe convective weather is evaluated, and corresponding suggestions and countermeasures are provided, including optimizing the layout and angle of photovoltaic panels, improving cleaning technology and equipment, so as to improve the power generation efficiency and stability of photovoltaic power stations under severe convective weather. Specifically, the layout and angle of photovoltaic panels are optimized: the spacing, arrangement and inclination of photovoltaic panels are set as optimization variables; the genetic algorithm is used to minimize the average power recovery time as the objective function for optimization; if the number of iterations reaches the preset value, the optimal layout plan and installation angle are output.

[0105] Design the motion trajectory and cleaning frequency of the automatic cleaning system, including: based on the optimal layout plan, using the particle swarm algorithm to maximize cleaning efficiency and minimize energy consumption as the optimization goal; if the convergence condition is met, output the optimal cleaning strategy.

[0106] Develop emergency plans for extreme weather, including: determining whether extreme weather such as strong winds or sandstorms will occur; if extreme weather occurs, automatically adjusting the tilt angle of the photovoltaic panels; and activating the dust cover system to protect the photovoltaic panels.

[0107] S106: Extract features from the quantitative relationship model under severe convective weather conditions and the preset photovoltaic power prediction model under non-severe convective weather conditions to obtain model feature vectors, construct a photovoltaic power characteristic difference prediction model based on historical weather data and the model feature vectors, and use the photovoltaic power characteristic difference prediction model to perform short-term, medium-term and long-term predictions of photovoltaic power under different weather conditions.

[0108] Exemplarily, the K-means clustering algorithm is used to classify historical weather data to obtain multiple weather categories, and a weather feature vector is constructed based on the characteristic parameters of each weather category; feature extraction is performed on the quantitative relationship model under severe convective weather conditions and the preset photovoltaic power prediction model under non-severe convective weather conditions to obtain the model feature vector; a photovoltaic power characteristic difference prediction model is constructed based on the weather feature vector and the model feature vector; real-time weather data is input into the photovoltaic power characteristic difference prediction model, and the photovoltaic power is predicted in the short term, medium term and long term through the sliding time window method to obtain the final photovoltaic power prediction value.

[0109] Furthermore, the photovoltaic power characteristics under different weather conditions are determined according to the photovoltaic power prediction value; the difference between the photovoltaic power prediction value and the average photovoltaic power prediction value is used to analyze the influence of each feature in the weather feature vector and the model feature vector on the photovoltaic power prediction result.

[0110] From the above description, it can be known that the present invention can understand the size distribution of dust particles in detail by performing particle size analysis on dust samples on the surface of photovoltaic panels. The dust samples are measured through particle size analysis data and an intensity signal reflecting the chemical composition of the dust is emitted. Measuring the chemical composition of the dust helps to identify specific components that may affect the performance and life of photovoltaic panels, and provides a basis for targeted cleaning strategies. Determining thermal map data carrying dust type distribution information based on particle size analysis data and intensity signals can intuitively display the distribution of dust on the surface of photovoltaic panels, which helps to accurately assess the degree of pollution in different areas.

[0111] Furthermore, the airflow around the photovoltaic panels is numerically simulated to determine the turbulence characteristic parameters, and a dust removal effect model is constructed based on the thermal map data and the airflow turbulence characteristic parameters. This model can quantify the dust removal effect under different turbulence intensities, providing a scientific basis for designing efficient cleaning systems or using natural wind for cleaning. The dust removal effect is predicted by the dust removal effect model, and a mapping relationship between turbulence intensity and dust removal effect is established.

[0112] Furthermore, a power recovery speed prediction model is constructed based on the power output data of the photovoltaic panels under different dust removal effects. The model directly links the dust removal effect with the recovery speed of the photovoltaic panel power output, so that the power generation capacity recovery time of the photovoltaic panels after cleaning can be accurately predicted. The power recovery speed is determined through the power recovery speed prediction model, and a mapping relationship between the dust removal effect and the power recovery speed is established.

[0113] Furthermore, by integrating the mapping relationship among turbulence intensity, dust removal effect and power recovery speed, a quantitative relationship model under severe convective weather conditions is constructed according to the turbulence intensity, dust removal effect and photovoltaic power recovery speed in historical operation data. This model can accurately predict the recovery speed of photovoltaic power under severe convective weather conditions, which is crucial for coping with extreme weather conditions, ensuring grid stability and improving energy supply reliability.

[0114] Furthermore, the quantitative relationship model under severe convective weather conditions and the preset photovoltaic power prediction model under non-severe convective weather conditions are feature extracted to obtain model feature vectors, and a photovoltaic power characteristic difference prediction model is constructed based on historical weather data and model feature vectors. The model combines the photovoltaic power characteristics under different weather conditions and can achieve short-term, medium-term and long-term power predictions. Through comprehensive analysis with the preset photovoltaic power prediction model under non-severe convective weather conditions, a photovoltaic power characteristic difference prediction model applicable to various weather conditions is constructed. The present invention realizes the accurate prediction of power recovery of photovoltaic power stations under severe convective weather, and provides technical support for improving power generation efficiency and stability.

[0115] like Figure 2 As shown, Figure 2 A method for predicting the difference in photovoltaic power characteristics under different weather conditions, the method comprising:

[0116] S201: Obtaining particle size analysis data of dust samples on the surface of photovoltaic panels.

[0117] For example, a high-precision laser particle size analyzer is used to scan and measure dust on the surface of photovoltaic panels under severe convective weather to obtain particle size distribution data of particles ranging from 0.1 μm to 1000 μm.

[0118] S202: Sending an element analysis instruction according to the particle size analysis data, measuring the dust sample through the element analysis instruction, and sending an intensity signal reflecting the chemical composition of the dust.

[0119] For example, an X-ray fluorescence spectrometer is used to perform elemental analysis on dust samples to obtain the original intensity data of each chemical element in the dust, and the intensity data is converted into mass percentage by a standard curve method. An elemental analysis instruction is sent to the X-ray fluorescence spectrometer based on the particle size analysis data, and the X-ray fluorescence spectrometer is used to perform elemental analysis on the dust samples to obtain the intensity signal of the chemical composition of the dust.

[0120] S203: Convert the particle size analysis data and the intensity signal into a feature vector.

[0121] For example, the particle size analysis data is discretized into 10 intervals using the logarithmic equal interval method, and the percentage of the number of particles in each interval is used as a feature. The intensity signal of the dust chemical composition is converted into mass percentage by the standard curve method, and the mass percentage of the chemical elements in the dust chemical composition is directly used as a feature to form a feature vector.

[0122] S204: Determine whether there is a classification hyperplane matching the feature vector in the support vector machine model library.

[0123] S2041: If there is a classification hyperplane feature vector matching the feature vector in the support vector machine model library, classify the feature vector to determine the thermal map data carrying the dust type distribution information.

[0124] For example, if there is a classification hyperplane feature vector matching the feature vector in the support vector machine model library, the support vector machine algorithm is used to train and classify the constructed dust feature vector, and the radial basis function is selected as the kernel function to map the feature vector to a high-dimensional space. A one-to-many strategy is used for multi-category classification, and the optimal classification hyperplane is found in the high-dimensional space to classify the dust samples into multiple categories. The accuracy of the classification results is evaluated using five-fold cross validation. Based on the classification results, the number distribution and spatial distribution of different types of dust on the surface of the photovoltaic panel are statistically analyzed, and the heat map drawing tool is used to generate heat map data carrying dust type distribution information. The coverage area ratio and accumulation thickness of each type of dust are quantified through image processing algorithms.

[0125] S2042: If there is no classification hyperplane feature vector matching the feature vector in the support vector machine model library, the feature vector is not classified.

[0126] S205: Perform high-frequency sampling on the airflow around the photovoltaic panel to obtain three-dimensional velocity field data of the airflow around the photovoltaic panel.

[0127] Exemplarily, stereo particle image velocimetry technology is used to perform high-frequency sampling of the airflow around the photovoltaic panel to obtain velocity field data in three-dimensional space. The sampling frequency is set to 1000 Hz, and the measurement range covers 5 times the characteristic length upstream of the photovoltaic panel to 10 times the characteristic length downstream.

[0128] S206: Measure pressure distribution data on the surface of the photovoltaic panel and the surrounding area through a pressure sensor array.

[0129] Exemplarily, the pressure distribution on the surface of the photovoltaic panel and the surrounding area is measured using a pressure sensor array to obtain pressure distribution data on the surface of the photovoltaic panel and the surrounding area, and the sampling frequency of the pressure sensor is synchronized with the sampling frequency of the velocity field.

[0130] S207: Preprocessing the three-dimensional velocity field data and pressure distribution data.

[0131] Exemplarily, the acquired three-dimensional velocity field data and pressure distribution data are preprocessed, including denoising, interpolation and coordinate transformation, to convert the original data into a format suitable for numerical simulation.

[0132] S208: Construct a calculation domain according to the preprocessed three-dimensional velocity field data and pressure distribution data.

[0133] For example, based on the preprocessed velocity field and pressure field data, the computational domain is constructed, and the computational domain is meshed using a structured grid. The grid density is denser in the area near the photovoltaic panel, the minimum grid size is 1 mm, and the far field area is appropriately sparse, with a total number of about 5 million grids. The photovoltaic panel surface is set to a no-slip wall boundary condition, the inlet boundary adopts a velocity inlet condition, and the outlet boundary adopts a pressure outlet condition.

[0134] S209: Numerical simulation of the airflow around the photovoltaic panel in the computational domain is performed using the large eddy simulation method.

[0135] Exemplarily, the large eddy simulation method is used to numerically simulate the airflow around the photovoltaic panel. The Smagorinsky subgrid scale model is selected, the time step is set to 0.0001 second, the total simulation time is 10 seconds, the finite volume method is used to discretize the control equations, and the fractional step method is used to solve the velocity field and pressure field.

[0136] S210: Determine whether the numerical simulation result meets the preset conditions.

[0137] S2101: If the numerical simulation results meet the preset conditions, the airflow turbulence characteristic parameters are determined based on the three-dimensional velocity field data.

[0138] Exemplarily, if the numerical simulation results meet the preset conditions, the airflow turbulence characteristic parameters are determined based on the three-dimensional velocity field data. Among them, the airflow turbulence characteristic parameters include turbulence intensity and turbulence scale. The turbulence intensity is obtained by the ratio of the velocity pulsation root mean square to the average velocity, and the turbulence scale is calculated by the integral scale of the two-point velocity correlation function. Furthermore, a spatial distribution cloud map of the turbulence characteristic parameters is drawn, and compared with the original data (three-dimensional velocity field data and pressure distribution data) for verification, and the relative error is calculated. If the relative error exceeds 10%, the grid density and time step are adjusted, and the numerical simulation is repeated.

[0139] S2102: If the numerical simulation result does not meet the preset conditions, perform the numerical simulation again.

[0140] S211: extracting dust distribution features from the thermal map data, performing standard deviation normalization processing on the dust distribution features and airflow turbulence characteristic parameters, and obtaining a standardized feature set.

[0141] For example, features are extracted from the heat map data of dust type distribution information, including dust coverage, dust particle size distribution, and dust type ratio, and features such as turbulence intensity, turbulence scale, and Reynolds stress are extracted from the airflow turbulence characteristic parameters. The dust removal effect is calculated, which is defined as removal rate CR = (M_initial-M_residual) / M_initial×100%, where M is the dust mass. All features are normalized by standard deviation to obtain a standardized feature set.

[0142] S212: using the Pearson correlation coefficient to analyze the correlation between the standardized feature set and the dust removal effect, and selecting feature data whose absolute value of the correlation coefficient is greater than a preset threshold as key feature data.

[0143] Exemplarily, the Pearson correlation coefficient is used to analyze the correlation between the standardized feature set and the dust removal effect, and feature data with an absolute value of the correlation coefficient greater than 0.5 is selected as key feature data.

[0144] S213: Constructing a dust removal effect model according to the key feature data.

[0145] Exemplarily, the input feature matrix and output label vector of the random forest algorithm are constructed based on the key feature data. The input features include the filtered dust distribution and airflow turbulence characteristics, and the output label is the dust removal effect. The grid search method is used to determine the hyperparameters of the random forest, the number of decision trees ranges from [50, 100, 200], the maximum depth of the tree ranges from [5, 10, 15], and the minimum number of leaf node samples ranges from [1, 2, 4]. The key feature data are divided into a training set and a validation set using the 10-fold cross validation method to train the dust removal effect model (the dust removal effect model is a random forest model). The root mean square error, mean absolute error, and R 2 The score evaluates the model performance. If R 2 If the score is lower than 0.8, return to tuning hyperparameters.

[0146] S214: Using the characteristic data of the turbulence intensity within a preset range as input of a dust removal effect model, and predicting the corresponding dust removal effect through the dust removal effect model.

[0147] Exemplarily, the trained random forest model is used to input feature data of turbulence intensity ranging from 0 to 0.5 with a step size of 0.01 to predict the corresponding dust removal effect.

[0148] S215: using a cubic spline interpolation method to obtain a continuous mapping function between the turbulence intensity and the corresponding dust removal effect.

[0149] S216: Draw a relationship curve through a continuous mapping function to determine a mapping relationship between the turbulence intensity and the dust removal effect.

[0150] For example, a continuous mapping function of turbulence intensity and dust removal effect is obtained by cubic spline interpolation, and a relationship curve between the two is plotted. The 95% confidence interval of the predicted corresponding dust removal effect is calculated, and the uncertainty range is marked on the relationship curve. The test set is divided by the holdout method, and the reliability of the model prediction results is verified using the test set data. The R 2 Score and RMSE.

[0151] like Figure 3 As shown, Figure 3 A method for predicting the difference in photovoltaic power characteristics under different weather conditions, the method comprising:

[0152] S301: Obtaining power output data of photovoltaic panels under different dust removal effects.

[0153] Exemplarily, a data collector is used to obtain power output data of photovoltaic panels under different dust removal effects, the sampling interval is set to 5 minutes, and the data are collected continuously for 7 days to obtain 2016 data points.

[0154] S302: Sending an abnormal value detection instruction according to the power output data, identifying abnormal points according to the abnormal value detection instruction, and performing interpolation processing on the identified abnormal points.

[0155] Exemplarily, outlier detection is performed on the collected raw data, the 3σ rule is used to identify abnormal points, and the linear interpolation method is used to interpolate missing values ​​and abnormal values.

[0156] S303: Preprocessing the power output data after the interpolation processing.

[0157] S304: The pre-processed power output data generates a trend term, a season term and a residual term through a time series decomposition method, and the trend term, the season term and the residual term are used as the time series decomposition result.

[0158] Exemplarily, the data is smoothed using the 5-point moving average method to obtain a power data sequence after preliminary processing. To quantify the dust removal effect, the removal rate CR is defined as (M initial - M residual) / M initial × 100%, where M is the dust mass per unit area. The pre-processed power output data is decomposed into a time series term, a seasonal term, and a residual term, and the trend term, seasonal term, and residual term are used as the time series decomposition result. Considering the periodic effects of sunshine changes, temperature changes, and humidity changes on power output, the effects of these factors are eliminated through multivariate seasonal adjustment to obtain a time series decomposition result without seasonality.

[0159] S305: Determine whether the time series decomposition result meets the modeling condition according to the existence of the time series decomposition result in the autocorrelation function database.

[0160] Exemplarily, an autocorrelation function database is queried, the autocorrelation function database contains function data with a range of model orders; it is determined whether there is function data in the autocorrelation function database that matches the time series decomposition result; if there is function data in the autocorrelation function database that matches the time series decomposition result, it is determined that the time series decomposition result meets the modeling conditions; if there is no function data in the autocorrelation function database that matches the time series decomposition result, it is determined that the time series decomposition result does not meet the modeling conditions.

[0161] S306: Construct a power recovery speed prediction model according to the time series decomposition result.

[0162] S307: Determine the power recovery speed according to the power recovery speed prediction model.

[0163] For example, if the time series decomposition result meets the modeling conditions, the preprocessed power output data is allowed to be modeled through the autoregressive integrated moving average model to obtain the power recovery speed prediction model. The power recovery speed is calculated according to the power recovery speed prediction model, and the power recovery speed is defined as the relative rate of change of power within 24 hours.

[0164] S308: Using support vector regression to establish a mapping relationship between dust removal effect and power recovery speed.

[0165] For example, support vector regression is implemented using the SVR class of sklearn, and the RBF kernel function is used. GridSearchCV is used to optimize the parameters, and the optimal parameters C = 100, gamma = 0.1 are obtained. The nonlinear mapping relationship between the cleaning effect and the recovery speed is established. The test set R 2 Reaching 0.92.

[0166] S309: Acquire historical operation data of the photovoltaic power station, where the historical operation data includes turbulence intensity, dust removal effect, and photovoltaic power recovery speed.

[0167] S310: Processing historical operation data to extract key features of the turbulence intensity, dust removal effect, and photovoltaic power recovery speed.

[0168] For example, historical operation data, including turbulence intensity, dust removal effect, and photovoltaic power recovery speed, are extracted from the photovoltaic power station database, and the data are cleaned and standardized to remove outliers and missing values. A variety of data completion methods are used, including linear interpolation, mean filling, and nearest neighbor filling. The most suitable method is selected according to the data type to obtain a complete sample data set. Feature engineering is performed on the processed sample data to extract key features of turbulence intensity, dust removal effect, and photovoltaic power recovery speed.

[0169] S311: construct input feature vectors and output labels based on key features, the input feature matrix is ​​turbulence intensity and dust removal effect, and the output label vector is photovoltaic power recovery speed.

[0170] Exemplarily, the correlation coefficient matrix between features is calculated, feature combinations with strong correlation are selected, and input feature vectors and output labels are constructed.

[0171] S312: Based on the mapping relationship between turbulence intensity and dust removal effect and the mapping relationship between the dust removal effect and the power recovery speed, a quantitative relationship model under severe convective weather conditions is constructed according to the input feature matrix and the output label vector, and the photovoltaic power recovery speed is predicted by the quantitative relationship model.

[0172] Exemplarily, a support vector regression algorithm is used to construct a quantitative relationship model under severe convective weather conditions between turbulence intensity, dust removal effect and photovoltaic power recovery speed based on the input feature matrix and the output label vector. L1 regularization and early stopping method are used to prevent overfitting, and the optimal hyperparameter combination, including kernel function type, regularization parameter and kernel parameter, is selected through grid search and 5-fold cross validation to obtain a quantitative relationship model with good generalization ability. Using the trained quantitative relationship model, different combinations of turbulence intensity and dust removal effect are input to predict the photovoltaic power recovery speed in the next 24 hours.

[0173] S313: Using the K-means clustering algorithm to classify the historical weather data, a plurality of weather categories are obtained, and a weather feature vector is constructed according to the feature parameters of each weather category.

[0174] Exemplarily, the K-means clustering algorithm is used to classify historical weather data, and weather conditions are divided into categories such as severe convection, sunny, cloudy, overcast, light precipitation, heavy rain, haze, and sandstorms. Feature parameters are extracted for each type of weather, including temperature, humidity, wind speed, light intensity, air pressure, visibility, etc., to construct a weather feature vector.

[0175] S314: Extract features of the quantitative relationship model under severe convective weather conditions and the preset photovoltaic power prediction model under non-severe convective weather conditions to obtain model feature vectors.

[0176] For example, the photovoltaic power prediction model under various weather conditions is feature extracted, the key parameters and prediction accuracy indicators of the model are extracted, the model feature vector is constructed, the principal component analysis method is used to reduce the feature dimension, and the principal component with an explained variance ratio greater than 95% is retained. The specific implementation process includes feature standardization, calculation of covariance matrix, eigenvalue decomposition and selection of principal components.

[0177] S315: Constructing a photovoltaic power characteristic difference prediction model according to the weather characteristic vector and the model characteristic vector.

[0178] Exemplarily, a model fusion framework is constructed using the random forest algorithm, with weather feature vectors and model feature vectors as inputs and actual photovoltaic power as output. The model parameters are trained by setting the number of trees to 500, the maximum depth to 10, and the minimum number of leaf node samples to 5, to obtain a comprehensive prediction model suitable for different weather conditions.

[0179] S316: Input the real-time weather data into the photovoltaic power characteristic difference prediction model, perform short-term prediction, medium-term prediction and long-term prediction on the photovoltaic power through the sliding time window method, and obtain the final photovoltaic power prediction value.

[0180] Exemplarily, based on the trained comprehensive prediction model, real-time weather data and the output results of each single prediction model are input to obtain the final photovoltaic power prediction value, and short-term, medium-term and long-term power predictions are achieved through the sliding time window method. The short-term prediction window size is 24 hours, the sliding step is 1 hour, the medium-term prediction window size is 7 days, the sliding step is 12 hours, and the long-term prediction window size is 30 days, and the sliding step is 1 day.

[0181] S317: Determine photovoltaic power characteristics under different weather conditions according to the photovoltaic power prediction value.

[0182] Exemplarily, a 95% confidence interval of the prediction result is calculated. The differences in photovoltaic power characteristics are quantified by calculating the mean, variance, skewness and kurtosis of photovoltaic power under different weather conditions.

[0183] S318: Analyze the influence of each feature in the weather feature vector and the model feature vector on the photovoltaic power prediction result by using the difference between the photovoltaic power prediction value and the photovoltaic power average prediction value.

[0184] For example, the SHAP (SHapleyAdditiveexPlanations) value is used to analyze the importance of features and understand the impact of different weather factors on photovoltaic power. Specifically, the difference between the photovoltaic power prediction value and the average photovoltaic power prediction value is used to analyze the impact of each feature in the weather feature vector and the model feature vector on the photovoltaic power prediction result.

[0185] like Figure 4 As shown, Figure 4 A method for predicting the difference in photovoltaic power characteristics under different weather conditions, the method comprising:

[0186] S401: Determine a photovoltaic power output curve according to the predicted photovoltaic power recovery speed and in combination with initial power and area information of the photovoltaic panel.

[0187] S402: Determine the power recovery time under different turbulence intensities using the photovoltaic power output curve.

[0188] S403: using a cubic spline interpolation method, based on the power recovery time under different turbulence intensities, a corresponding relationship between turbulence intensity and power recovery time is established.

[0189] Exemplarily, the photovoltaic power recovery speed is predicted according to the output of the photovoltaic power characteristic difference prediction model, and the photovoltaic power output curve is determined according to the predicted photovoltaic power recovery speed and the initial power and area information of the photovoltaic panel. The photovoltaic power output curve is used to determine the power recovery time under different turbulence intensities, and the power recovery threshold is set to 80%, 90% and 95%. The corresponding relationship between turbulence intensity and recovery time is obtained by cubic spline interpolation method, and the power recovery time with an interval of 0.05 within the range of turbulence intensity of 0.1 to 0.5 is determined, and the recovery time curves under the three thresholds of 80%, 90% and 95% are obtained.

[0190] S404: Determine the layout schemes of different photovoltaic panels and the turbulence intensity distribution under the installation angles according to the actual meteorological data and terrain information.

[0191] For example, ANSYS Fluent software is used to simulate the airflow field around the photovoltaic power station under severe convective weather, and actual meteorological data and terrain information are input to determine the turbulence intensity distribution under different photovoltaic panel layout schemes and installation angles. Specifically, ANSYS Fluent software is used to simulate the airflow field under a wind speed of 10m / s, and actual terrain data within a range of 1km×1km is input, the number of grids is set to 1 million, and the turbulence intensity distribution is determined.

[0192] S405: Evaluate the power recovery performance under each layout scheme and installation angle according to the corresponding relationship between the turbulence intensity distribution and the power recovery time, and optimize the layout schemes and installation angles of different photovoltaic panels.

[0193] Exemplarily, the power recovery performance under each layout scheme and installation angle is evaluated in combination with the turbulence intensity-recovery time curve. A genetic algorithm is used to optimize the layout scheme and installation angle of different photovoltaic panels, with minimizing the average power recovery time as the objective function, and setting the photovoltaic panel spacing, arrangement and inclination as optimization variables, where the photovoltaic panel spacing is 5-15m, the arrangement is rectangular or hexagonal, and the inclination is 0-60 degrees. The population size is set to 100, the crossover rate is 0.8, the mutation rate is 0.1, the number of iterations is 1000, and the optimal layout scheme and installation angle are obtained through iterative calculation.

[0194] Furthermore, based on the optimized photovoltaic panel layout, the particle swarm algorithm is used to design the motion trajectory and cleaning frequency of the automatic cleaning system, with the optimization goal of maximizing cleaning efficiency and minimizing energy consumption, taking into account parameters such as the movement speed of the cleaning equipment, the water spray pressure, and the amount of detergent used, to obtain the optimal cleaning strategy. Specifically, the particle swarm algorithm optimizes the cleaning system, with the number of particles set to 50 and the maximum number of iterations set to 500, to obtain the optimal cleaning strategy.

[0195] At the same time, the cleaning equipment was improved, carbon fiber composite materials were selected to reduce weight, modular design was used to improve maintenance efficiency, and the nozzle structure was optimized to improve cleaning effect. The economic benefits of the optimized solution were evaluated, the payback period and net present value were calculated, and an environmental impact assessment was conducted, including water resource consumption and carbon emission analysis. For extreme weather such as strong winds and sandstorms, emergency plans were formulated, including: judging whether extreme weather such as strong winds or sandstorms occurred; if extreme weather occurred, the inclination of the photovoltaic panels was automatically adjusted; and the dust cover system was activated to protect the photovoltaic panels. It should be noted that the weight of the improved cleaning equipment was reduced from the original 50kg to 35kg, and the modular design shortened the maintenance time from 2 hours to 30 minutes. The economic benefit evaluation showed that the investment payback period of the optimized solution was 3.5 years, and the net present value of 20 years was 5 million yuan. The environmental impact assessment showed that the average annual water saving was 30% and the carbon emissions were reduced by 15%. In the extreme weather emergency plan, when the wind speed exceeds 20m / s, the photovoltaic panel is automatically adjusted to a 0-degree inclination, and when the dust concentration exceeds 50μg / m 3 Activate the dust cover.

[0196] Finally, a step-by-step implementation plan is developed to apply the optimization results to the transformation process of the actual PV power station, including equipment updates, control system upgrades, and personnel training.

[0197] From the above description, it can be seen that the present invention combines time series analysis to calculate the power recovery speed and establishes a quantitative relationship model between turbulence intensity, dust removal effect and power recovery speed. The model is used to predict photovoltaic power recovery under different turbulence intensities, evaluate the impact of turbulence and provide optimization suggestions.

[0198] like Figure 5 As shown, Figure 5 The structure diagram of an embodiment of a photovoltaic power characteristic difference prediction system under different weather conditions is shown in FIG. A photovoltaic power characteristic difference prediction system 10 under different weather conditions includes:

[0199] The thermogram data determination module 11 is used to obtain the particle size analysis data of the dust sample on the surface of the photovoltaic panel, measure the dust sample through the particle size analysis data and send out an intensity signal reflecting the chemical composition of the dust, and determine the thermogram data carrying the dust type distribution information based on the particle size analysis data and the intensity signal.

[0200] The airflow turbulence characteristic parameter determination module 12 is used to obtain the three-dimensional velocity field data of the airflow around the photovoltaic panel and the pressure distribution data on the surface of the photovoltaic panel and the surrounding area, and to perform numerical simulation of the airflow around the photovoltaic panel based on the three-dimensional velocity field data and the pressure distribution data to determine the airflow turbulence characteristic parameters.

[0201] The module 13 for establishing the mapping relationship between turbulence intensity and dust removal effect is used to construct a dust removal effect model according to the thermal map data and the airflow turbulence characteristic parameters, predict the dust removal effect through the dust removal effect model, and establish a mapping relationship between turbulence intensity and dust removal effect.

[0202] The dust removal effect and power recovery speed mapping relationship establishment module 14 is used to build a power recovery speed prediction model based on the power output data of the photovoltaic panel under different dust removal effects, determine the power recovery speed through the power recovery speed prediction model, and establish a mapping relationship between the dust removal effect and the power recovery speed.

[0203] The quantitative relationship model building module 15 is used to build a quantitative relationship model under severe convective weather conditions based on the mapping relationship between turbulence intensity and dust removal effect and the mapping relationship between dust removal effect and power recovery speed according to the turbulence intensity, dust removal effect and photovoltaic power recovery speed in historical operation data, and predict the photovoltaic power recovery speed through the quantitative relationship model.

[0204] The photovoltaic power prediction module 16 is used to extract features from the quantitative relationship model under severe convective weather conditions and the preset photovoltaic power prediction model under non-severe convective weather conditions, obtain model feature vectors, build a photovoltaic power characteristic difference prediction model based on historical weather data and model feature vectors, and use the photovoltaic power characteristic difference prediction model to perform short-term, medium-term and long-term predictions of photovoltaic power under different weather conditions.

[0205] Exemplarily, in the thermal map data determination module 11, the particle size analysis data of the dust sample on the surface of the photovoltaic panel is obtained; an element analysis instruction is sent according to the particle size analysis data, the dust sample is measured by the element analysis instruction, and an intensity signal reflecting the chemical composition of the dust is issued; the particle size analysis data and the intensity signal are converted into a feature vector; it is determined whether there is a classification hyperplane matching the feature vector in the support vector machine model library; if there is a classification hyperplane feature vector matching the feature vector in the support vector machine model library, the feature vector is classified to determine the thermal map data carrying the dust type distribution information; if there is no classification hyperplane feature vector matching the feature vector in the support vector machine model library, the feature vector is not classified. In the airflow turbulence characteristic parameter determination module 12, the airflow around the photovoltaic panel is sampled at high frequency to obtain the three-dimensional velocity field data of the airflow around the photovoltaic panel; the pressure distribution data of the photovoltaic panel surface and the surrounding area are measured by a pressure sensor array; the three-dimensional velocity field data and the pressure distribution data are preprocessed; a calculation domain is constructed based on the preprocessed three-dimensional velocity field data and pressure distribution data; the airflow around the photovoltaic panel is numerically simulated in the calculation domain by a large eddy simulation method; it is judged whether the numerical simulation results meet the preset conditions; if the numerical simulation results meet the preset conditions, the airflow turbulence characteristic parameters are determined based on the three-dimensional velocity field data; if the numerical simulation results do not meet the preset conditions, the numerical simulation is performed again. In the module 13 for establishing the mapping relationship between turbulence intensity and dust removal effect, the dust distribution characteristics in the thermal map data are extracted, and the dust distribution characteristics and airflow turbulence characteristic parameters are subjected to standard deviation normalization to obtain a standardized feature set; the Pearson correlation coefficient is used to analyze the correlation between the standardized feature set and the dust removal effect, and the feature data whose absolute value of the correlation coefficient is greater than a preset threshold is selected as the key feature data; a dust removal effect model is constructed based on the key feature data; the feature data of turbulence intensity within a preset range is used as the input of the dust removal effect model, and the corresponding dust removal effect is predicted through the dust removal effect model; the cubic spline interpolation method is used to obtain a continuous mapping function between turbulence intensity and the corresponding dust removal effect; a relationship curve is drawn through the continuous mapping function to determine the mapping relationship between turbulence intensity and dust removal effect.In the module 14 for establishing the mapping relationship between dust removal effect and power recovery speed, the power output data of the photovoltaic panel under different dust removal effects is obtained; an abnormal value detection instruction is sent according to the power output data, an abnormal point is identified according to the abnormal value detection instruction, and the identified abnormal point is interpolated; the power output data after the interpolation processing is preprocessed; the preprocessed power output data generates trend terms, seasonal terms and residual terms through a time series decomposition method, and the trend terms, seasonal terms and residual terms are used as the time series decomposition results; according to the existence of the time series decomposition results in the autocorrelation function database, it is determined that the time series decomposition results meet the modeling conditions; a power recovery speed prediction model is constructed according to the preprocessed power output data; the power recovery speed is determined according to the power recovery speed prediction model; and support vector regression is used to establish a mapping relationship between dust removal effect and power recovery speed. In the quantitative relationship model construction module 15, the historical operation data of the photovoltaic power station is obtained, and the historical operation data includes turbulence intensity, dust removal effect and photovoltaic power recovery speed; the historical operation data is processed to extract the key features of turbulence intensity, dust removal effect and photovoltaic power recovery speed; the input feature vector and output label are constructed according to the key features, the input feature matrix is ​​turbulence intensity and dust removal effect, and the output label vector is photovoltaic power recovery speed; based on the mapping relationship between turbulence intensity and dust removal effect and the mapping relationship between dust removal effect and power recovery speed, according to the input feature matrix and the output label vector, a quantitative relationship model under severe convective weather conditions is constructed, and the photovoltaic power recovery speed is predicted by the quantitative relationship model. In the photovoltaic power prediction module 16, the K-means clustering algorithm is used to classify the historical weather data to obtain multiple weather categories, and a weather feature vector is constructed according to the characteristic parameters of each weather category; the quantitative relationship model under severe convective weather conditions and the preset photovoltaic power prediction model under non-severe convective weather conditions are feature extracted to obtain the model feature vector; a photovoltaic power characteristic difference prediction model is constructed according to the weather feature vector and the model feature vector; the real-time weather data is input into the photovoltaic power characteristic difference prediction model, and the photovoltaic power is predicted in the short term, medium term and long term through the sliding time window method to obtain the final photovoltaic power prediction value.

[0206] like Figure 6 As shown, Figure 6 The structure diagram of an embodiment of the medium provided by the present invention is shown in FIG. The medium 20 stores at least one computer program 21, which is executed by a processor to implement the following Figure 1 - Figure 4 In one embodiment, the medium 20 may be a storage chip, a hard disk, a mobile hard disk, a USB flash drive, an optical disk, or other readable and writable storage tools, or a server, etc.

[0207] The above describes specific embodiments of the present specification, and other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily have to be performed in the specific order or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0208] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and non-volatile computer-readable storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0209] The apparatus, device, non-volatile computer-readable storage medium and method provided in the embodiments of this specification correspond to each other, and therefore, the apparatus, device, and non-volatile computer storage medium also have similar beneficial technical effects as the corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the corresponding apparatus, device, and non-volatile computer storage medium will not be repeated here.

[0210] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0211] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification may be in the form of complete hardware embodiments, complete software embodiments, or embodiments in combination with software and hardware. Moreover, the embodiments of this specification may be in the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0212] This specification is described with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of this specification. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, 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 flowchart and / or block diagram. 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.

[0213] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate 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 A function specified in one or more boxes.

[0214] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0215] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0216] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0217] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by 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 memory (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 technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic 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 temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0218] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0219] This specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. This specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0220] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0221] The above disclosure is only the preferred embodiment of the present invention, which certainly cannot be used to limit the scope of the present invention. Therefore, equivalent changes made according to the claims of the present invention are still within the scope of the present invention.

Claims

1. A method for predicting differences in photovoltaic power characteristics under different weather conditions, characterized in that: The method comprises: Acquire particle size analysis data of dust samples on the surface of photovoltaic panels, measure the dust samples through the particle size analysis data and emit an intensity signal reflecting the chemical composition of the dust, and determine thermal map data carrying dust type distribution information based on the particle size analysis data and the intensity signal; Acquire three-dimensional velocity field data of the airflow around the photovoltaic panel and pressure distribution data on the surface of the photovoltaic panel and the surrounding area, perform numerical simulation on the airflow around the photovoltaic panel based on the three-dimensional velocity field data and the pressure distribution data, and determine the airflow turbulence characteristic parameters; Constructing a dust removal effect model according to the thermal map data and the airflow turbulence characteristic parameters, predicting the dust removal effect through the dust removal effect model, and establishing a mapping relationship between turbulence intensity and dust removal effect; A power recovery speed prediction model is constructed based on power output data of photovoltaic panels under different dust removal effects, a power recovery speed is determined by the power recovery speed prediction model, and a mapping relationship between the dust removal effect and the power recovery speed is established; Based on the mapping relationship between the turbulence intensity and the dust removal effect and the mapping relationship between the dust removal effect and the power recovery speed, a quantitative relationship model under severe convective weather conditions is constructed according to the turbulence intensity, dust removal effect and photovoltaic power recovery speed in the historical operation data, and the photovoltaic power recovery speed is predicted by the quantitative relationship model; Feature extraction is performed on the quantitative relationship model under severe convective weather conditions and the preset photovoltaic power prediction model under non-severe convective weather conditions to obtain model feature vectors, and a photovoltaic power characteristic difference prediction model is constructed based on historical weather data and the model feature vectors. Short-term, medium-term and long-term predictions of photovoltaic power under different weather conditions are performed using the photovoltaic power characteristic difference prediction model.

2. The method for predicting photovoltaic power characteristics under different weather conditions according to claim 2, characterized in that: The step of obtaining particle size analysis data of dust samples on the surface of the photovoltaic panel, measuring the dust samples through the particle size analysis data and emitting an intensity signal reflecting the chemical composition of the dust, and determining thermal map data carrying dust type distribution information according to the particle size analysis data and the intensity signal specifically includes: Obtain particle size analysis data of dust samples on the surface of photovoltaic panels; sending an elemental analysis instruction according to the particle size analysis data, measuring the dust sample through the elemental analysis instruction, and sending an intensity signal reflecting the chemical composition of the dust; converting the particle size analysis data and the intensity signal into a feature vector; Determine whether there is a classification hyperplane matching the feature vector in the support vector machine model library; If there is a classification hyperplane matching the feature vector in the support vector machine model library, classify the feature vector to determine the thermal map data carrying the dust type distribution information; If the feature vector does not have a classification hyperplane matching the feature vector in the support vector machine model library, the feature vector will not be classified.

3. The method for predicting the difference in photovoltaic power characteristics under different weather conditions according to claim 2, characterized in that: The step of acquiring three-dimensional velocity field data of the airflow around the photovoltaic panel and pressure distribution data on the surface of the photovoltaic panel and the surrounding area, and numerically simulating the airflow around the photovoltaic panel based on the three-dimensional velocity field data and the pressure distribution data to determine the airflow turbulence characteristic parameters specifically includes: Perform high-frequency sampling on the airflow around the photovoltaic panel to obtain the three-dimensional velocity field data of the airflow around the photovoltaic panel; The pressure distribution data on the surface of the photovoltaic panel and the surrounding area is measured by a pressure sensor array; Preprocessing the three-dimensional velocity field data and pressure distribution data; Constructing a calculation domain according to the preprocessed three-dimensional velocity field data and pressure distribution data; The airflow around the photovoltaic panel is numerically simulated in the computational domain by using a large eddy simulation method; Determining whether the numerical simulation result meets a preset condition; If the numerical simulation results meet the preset conditions, then determining the airflow turbulence characteristic parameters according to the three-dimensional velocity field data; If the numerical simulation result does not meet the preset conditions, the numerical simulation is performed again.

4. The method for predicting the difference in photovoltaic power characteristics under different weather conditions according to claim 3, characterized in that: The step of building a dust removal effect model based on the thermal map data and the airflow turbulence characteristic parameters, predicting the dust removal effect through the dust removal effect model, and establishing a mapping relationship between turbulence intensity and dust removal effect specifically includes: Extracting dust distribution features from the thermal map data, performing standard deviation normalization processing on the dust distribution features and airflow turbulence characteristic parameters, and obtaining a standardized feature set; The Pearson correlation coefficient is used to analyze the correlation between the standardized feature set and the dust removal effect, and feature data whose absolute value of the correlation coefficient is greater than a preset threshold is selected as key feature data; Constructing a dust removal effect model according to the key feature data; Using characteristic data of turbulence intensity within a preset range as input of the dust removal effect model, and predicting the corresponding dust removal effect through the dust removal effect model; A continuous mapping function between the turbulence intensity and the corresponding dust removal effect is obtained by using a cubic spline interpolation method; A relationship curve is drawn through the continuous mapping function to determine the mapping relationship between the turbulence intensity and the dust removal effect.

5. The method for predicting photovoltaic power characteristics under different weather conditions according to claim 4, characterized in that: The power recovery speed prediction model is constructed according to the power output data of the photovoltaic panel under different dust removal effects, the power recovery speed is determined by the power recovery speed prediction model, and the mapping relationship between the dust removal effect and the power recovery speed is established, specifically including: Obtain power output data of photovoltaic panels under different dust removal effects; sending an abnormal value detection instruction according to the power output data, identifying abnormal points according to the abnormal value detection instruction, and performing interpolation processing on the identified abnormal points; Preprocessing the power output data after interpolation processing; The pre-processed power output data generates trend terms, seasonal terms and residual terms through a time series decomposition method, and the trend terms, seasonal terms and residual terms are used as time series decomposition results; Determining, based on the existence of the time series decomposition result in the autocorrelation function database, that the time series decomposition result meets the modeling condition; Constructing a power recovery speed prediction model according to the time series decomposition result; Determining the power recovery speed according to the power recovery speed prediction model; Support vector regression is used to establish the mapping relationship between dust removal effect and power recovery speed.

6. The method for predicting the difference in photovoltaic power characteristics under different weather conditions according to claim 5, characterized in that: The determining, based on the existence of the time series decomposition result in the autocorrelation function database, that the time series decomposition result meets the modeling condition specifically includes: querying an autocorrelation function database, wherein the autocorrelation function database contains function data with a range of model orders; Determine whether the function data matching the time series decomposition result exists in the autocorrelation function database; If the function data matching the time series decomposition result exists in the autocorrelation function database, it is determined that the time series decomposition result meets the modeling condition; If the function data matching the time series decomposition result does not exist in the autocorrelation function database, it is determined that the time series decomposition result does not meet the modeling condition.

7. The method for predicting photovoltaic power characteristics under different weather conditions according to claim 6, characterized in that: Based on the mapping relationship between the turbulence intensity and the dust removal effect and the mapping relationship between the dust removal effect and the power recovery speed, a quantitative relationship model under severe convective weather conditions is constructed according to the turbulence intensity, dust removal effect and photovoltaic power recovery speed in historical operation data, and photovoltaic power recovery speed is predicted by the quantitative relationship model, specifically including: Acquiring historical operation data of the photovoltaic power station, wherein the historical operation data includes turbulence intensity, dust removal effect, and photovoltaic power recovery speed; Processing the historical operation data to extract key features of the turbulence intensity, dust removal effect and photovoltaic power recovery speed; Constructing an input feature vector and an output label according to the key features, the input feature matrix is ​​the turbulence intensity and dust removal effect, and the output label vector is the photovoltaic power recovery speed; Based on the mapping relationship between the turbulence intensity and the dust removal effect and the mapping relationship between the dust removal effect and the power recovery speed, a quantitative relationship model under severe convective weather conditions is constructed according to the input feature matrix and the output label vector, and the photovoltaic power recovery speed is predicted by the quantitative relationship model.

8. The method for predicting photovoltaic power characteristics under different weather conditions according to claim 7, characterized in that: The mapping relationship between the turbulence intensity and the dust removal effect and the mapping relationship between the dust removal effect and the power recovery speed are constructed based on the input feature matrix and the output label vector under severe convective weather conditions, and the photovoltaic power recovery speed is predicted by the quantitative relationship model, specifically including: Based on the mapping relationship between the turbulence intensity and the dust removal effect and the mapping relationship between the dust removal effect and the power recovery speed, the quantitative relationship model under severe convective weather conditions is determined according to P=αI+βD, wherein P is the photovoltaic power recovery speed, I is the turbulence intensity, D is the dust removal effect, and α and β are model parameters.

9. The method for predicting photovoltaic power characteristics under different weather conditions according to claim 7, characterized in that: The method further includes establishing a quantitative relationship model between turbulence intensity, dust removal effect and photovoltaic power recovery speed according to the input feature vector and the output label, and predicting the photovoltaic power recovery speed through the quantitative relationship model: Determining a photovoltaic power output curve based on the predicted photovoltaic power recovery speed and combining initial power and area information of the photovoltaic panel; Determining the power recovery time under different turbulence intensities using the photovoltaic power output curve; Using a cubic spline interpolation method, based on the power recovery time under the different turbulence intensities, a corresponding relationship between turbulence intensity and power recovery time is established; Determine the layout of different photovoltaic panels and the distribution of turbulence intensity at different installation angles based on actual meteorological data and terrain information; The power recovery performance under each layout scheme and installation angle is evaluated according to the corresponding relationship between the turbulence intensity distribution and the power recovery time, and the layout schemes and installation angles of the different photovoltaic panels are optimized.

10. The method for predicting photovoltaic power characteristics under different weather conditions according to claim 8, characterized in that: The feature extraction of the quantitative relationship model under severe convective weather conditions and the preset photovoltaic power prediction model under non-severe convective weather conditions is performed to obtain a model feature vector, a photovoltaic power characteristic difference prediction model is constructed according to historical weather data and the model feature vector, and short-term, medium-term and long-term predictions of photovoltaic power under different weather conditions are performed by using the photovoltaic power characteristic difference prediction model, specifically including: The K-means clustering algorithm is used to classify the historical weather data to obtain multiple weather categories, and a weather feature vector is constructed according to the characteristic parameters of each weather category; Extract features from the quantitative relationship model under severe convective weather conditions and the preset photovoltaic power prediction model under non-severe convective weather conditions to obtain model feature vectors; Constructing a photovoltaic power characteristic difference prediction model according to the weather characteristic vector and the model characteristic vector; The real-time weather data is input into the photovoltaic power characteristic difference prediction model, and the photovoltaic power is predicted in the short term, medium term and long term by the sliding time window method to obtain the final photovoltaic power prediction value.

11. The method for predicting photovoltaic power characteristics under different weather conditions according to claim 10, characterized in that: The method of inputting the real-time weather data into the photovoltaic power characteristic difference prediction model, performing short-term prediction, medium-term prediction and long-term prediction on the photovoltaic power by a sliding time window method, and obtaining the final photovoltaic power prediction value further specifically includes: determining photovoltaic power characteristics under different weather conditions according to the photovoltaic power prediction value; The difference between the photovoltaic power prediction value and the photovoltaic power average prediction value is used to analyze the influence of each feature in the weather feature vector and the model feature vector on the photovoltaic power prediction result.

12. A photovoltaic power characteristics difference prediction system under different weather conditions, characterized in that: The system comprises: A thermal map data determination module, used to obtain particle size analysis data of dust samples on the surface of photovoltaic panels, measure the dust samples through the particle size analysis data and send out an intensity signal reflecting the chemical composition of the dust, and determine thermal map data carrying dust type distribution information based on the particle size analysis data and the intensity signal; An airflow turbulence characteristic parameter determination module is used to obtain three-dimensional velocity field data of the airflow around the photovoltaic panel and pressure distribution data on the surface of the photovoltaic panel and the surrounding area, and to perform numerical simulation on the airflow around the photovoltaic panel based on the three-dimensional velocity field data and the pressure distribution data to determine the airflow turbulence characteristic parameters; A module for establishing a mapping relationship between turbulence intensity and dust removal effect, which is used to construct a dust removal effect model according to the thermal map data and the airflow turbulence characteristic parameters, predict the dust removal effect through the dust removal effect model, and establish a mapping relationship between turbulence intensity and the dust removal effect; A module for establishing a mapping relationship between dust removal effect and power recovery speed, which is used to construct a power recovery speed prediction model based on power output data of photovoltaic panels under different dust removal effects, determine the power recovery speed through the power recovery speed prediction model, and establish a mapping relationship between the dust removal effect and the power recovery speed; A quantitative relationship model building module, for building a quantitative relationship model under severe convective weather conditions based on the mapping relationship between the turbulence intensity and the dust removal effect and the mapping relationship between the dust removal effect and the power recovery speed, according to the turbulence intensity, dust removal effect and photovoltaic power recovery speed in the historical operation data, and predicting the photovoltaic power recovery speed through the quantitative relationship model; The photovoltaic power prediction module is used to extract features from the quantitative relationship model under severe convective weather conditions and the preset photovoltaic power prediction model under non-severe convective weather conditions, obtain model feature vectors, construct a photovoltaic power characteristic difference prediction model based on historical weather data and the model feature vectors, and use the photovoltaic power characteristic difference prediction model to perform short-term, medium-term and long-term predictions of photovoltaic power under different weather conditions.

13. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 11.

Images