A method, device and system for obtaining high-precision irradiation data

Abstract

The present invention provides a method, apparatus and system for obtaining high-precision irradiance data, relating to the technical field of photovoltaic systems. The method for obtaining high-precision irradiance data according to the present invention includes: obtaining low-precision horizontal irradiance data of a reference region, and calibrating the low-precision horizontal irradiance data through a low-precision irradiance data calibration model to obtain high-precision horizontal irradiance data of the reference region; inputting meteorological data, geographical data and the high-precision horizontal irradiance data of the reference region into an irradiance data conversion model between different regions to predict high-precision horizontal irradiance data of any target region; and converting the high-precision horizontal irradiance data of the any target region into high-precision inclined-plane irradiance data with any inclination angle of the any target region through a horizontal-to-inclined-plane irradiance data conversion model. Compared with the solution of directly using satellite data, the present invention has higher precision, reduces the number of installed irradiance meters, and effectively reduces the cost of obtaining irradiance data.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic systems, and more particularly, to a method, apparatus, and system for obtaining high-precision irradiation data. Background Art

[0002] Currently, the main provider of irradiation data in the industry is the Solargis database. However, it calculates the actual ground irradiation data based on satellite data through an earth model, resulting in large data errors and making it difficult to meet the data requirements for power station construction site selection and operation and maintenance monitoring. Therefore, to obtain high-precision irradiation data, only by installing expensive high-precision irradiance meters. It is costly to obtain high-precision irradiation data for all actual operation and maintenance sites by installing expensive irradiation equipment. Moreover, considering the large number of component tilts in distributed power stations, each tilt requires a corresponding high-precision irradiance meter, resulting in too high equipment costs and low economic feasibility. Summary of the Invention

[0003] The problem solved by the present invention is how to reduce the cost of obtaining irradiation data while ensuring accuracy.

[0004] To solve the above problems, the present invention provides a method for obtaining high-precision irradiation data, including: obtaining low-precision horizontal irradiation data of a reference region, calibrating the low-precision horizontal irradiation data through a low-precision irradiation data calibration model to obtain high-precision horizontal irradiation data of the reference region; inputting meteorological data, geographical data, and the high-precision horizontal irradiation data of the reference region into an irradiation data conversion model between different regions to predict high-precision horizontal irradiation data of any target region; and converting the high-precision horizontal irradiation data of the any target region into high-precision inclined-plane irradiation data of any inclination angle of the any target region through a horizontal-to-inclined-plane irradiation data conversion model.

[0005] The method for obtaining high-precision irradiation data according to the present invention calibrates, optimizes, converts, and then predicts the inclined-plane irradiation data of any inclination angle of any target region by calibrating the low-precision horizontal irradiation data of the reference region. Compared with the solution of directly using satellite data to obtain irradiation data, it has higher accuracy, reduces the number of installed irradiance meters, and effectively reduces the cost of obtaining irradiation data.

[0006] Optionally, the construction process of the low-precision irradiation data calibration model includes: eliminating systematic errors for different irradiance meters; deploying low-precision irradiance meters and high-precision irradiance meters with the same inclination angle at the same location to collect low-precision total irradiation data and high-precision total irradiation data; and fitting the sensitivities of the low-precision irradiance meter and the high-precision irradiance meter based on the random forest algorithm according to the low-precision total irradiation data and the high-precision total irradiation data to construct the low-precision irradiation data calibration model.

[0007] The method for obtaining high-precision irradiance data according to the present invention realizes the fitting of the sensitivities of low-precision irradiance meters and high-precision irradiance meters through the random forest algorithm, so that high-precision irradiance data can be obtained at a relatively low cost, effectively reducing the cost of obtaining irradiance data.

[0008] Optionally, the fitting of the sensitivities of low-precision irradiance meters and high-precision irradiance meters based on the random forest algorithm includes: constructing an original training set; performing random sampling with replacement on the original training set, randomly selecting multiple features in each sampled training set to establish different classification models, and determining the optimal classification by voting to perform the fitting of the sensitivities of low-precision irradiance meters and high-precision irradiance meters.

[0009] The method for obtaining high-precision irradiance data according to the present invention realizes the fitting of the sensitivities of low-precision irradiance meters and high-precision irradiance meters through the random forest algorithm, so that high-precision irradiance data can be obtained at a relatively low cost, effectively reducing the cost of obtaining irradiance data.

[0010] Optionally, the construction process of the irradiance data conversion model between different regions includes: eliminating systematic errors for different irradiance meters; establishing multiple typical surface meteorological stations within the same region to obtain the horizontal total irradiance data, numerical weather forecasts, and geographical locations of each place; using deep data mining to obtain the influence degrees of the numerical weather forecasts and the geographical locations, and establishing the irradiance data conversion model between different regions according to the horizontal total irradiance data and the influence degrees.

[0011] The method for obtaining high-precision irradiance data according to the present invention converts the horizontal irradiance data of the reference region within the region to obtain the horizontal irradiance data of the adjacent target region, realizing the prediction of irradiance data without ground meteorological equipment, and effectively reducing the cost of obtaining irradiance data.

[0012] Optionally, the construction process of the horizontal and inclined surface irradiance data conversion model includes: eliminating systematic errors for different irradiance meters; arranging a horizontal irradiance meter and inclined surface irradiance meters with different tilting angles at the same location to collect the horizontal total irradiance data of the typical region and the inclined total irradiance data of the typical region, and simultaneously collecting various meteorological index data; based on the characteristic relationship between the irradiance data and the meteorological index data, using the GRU neural network model to construct a direct and diffuse separation model; constructing the horizontal and inclined surface irradiance data conversion model according to the direct and diffuse separation model and the Perez physical model.

[0013] The method for obtaining high-precision irradiance data according to the present invention constructs a direct and diffuse separation model through the GRU neural network model to construct a horizontal and inclined surface irradiance data conversion model, and can obtain the inclined surface irradiance data at any tilting angle without installing any inclined surface irradiance meter, effectively reducing the cost of obtaining irradiance data.

[0014] Optionally, the conversion of the high-precision horizontal irradiance data of any target area into the high-precision inclined-plane irradiance data of any inclination angle in the any target area by the horizontal and inclined-plane irradiance data conversion model includes: inputting the high-precision horizontal irradiance data of the any target area into the direct-diffuse separation model to obtain the direct and diffuse irradiance data on the horizontal plane, inputting the direct and diffuse irradiance data on the horizontal plane into the Perez physical model to obtain the direct and diffuse irradiance data on the inclined plane of any inclination angle, and obtaining the high-precision inclined-plane irradiance data of any inclination angle in the any target area according to the direct and diffuse irradiance data on the inclined plane of any inclination angle.

[0015] In the method for obtaining high-precision irradiance data according to the present invention, the high-precision horizontal irradiance data of any target area is converted into the high-precision inclined-plane irradiance data of any inclination angle in the any target area through the direct-diffuse separation model and the Perez physical model, and the inclined-plane irradiance data of any inclination angle can be obtained without installing any inclined-plane irradiance meter, effectively reducing the cost of obtaining irradiance data.

[0016] Optionally, the meteorological index data includes solar altitude angle, hour angle, azimuth angle, temperature and humidity, wind speed, wind direction, altitude, air pressure, and clear sky coefficient.

[0017] In the method for obtaining high-precision irradiance data according to the present invention, by setting the meteorological index data to include solar altitude angle, hour angle, azimuth angle, temperature and humidity, wind speed, wind direction, altitude, air pressure, and clear sky coefficient, the training effect of the GRU neural network model is improved, which is beneficial to improving the accuracy of the horizontal and inclined-plane irradiance data conversion model.

[0018] The present invention also provides an apparatus for obtaining high-precision irradiance data, including: a calibration module, configured to obtain the low-precision horizontal irradiance data of a reference area, and calibrate the low-precision horizontal irradiance data through a low-precision irradiance data calibration model to obtain the high-precision horizontal irradiance data of the reference area; a conversion module, configured to input meteorological data, geographical data, and the high-precision horizontal irradiance data of the reference area into an irradiance data conversion model between different areas, and predict to obtain the high-precision horizontal irradiance data of any target area; a transformation module, configured to convert the high-precision horizontal irradiance data of the any target area into the high-precision inclined-plane irradiance data of any inclination angle in the any target area through a horizontal and inclined-plane irradiance data conversion model. The advantages of the apparatus for obtaining high-precision irradiance data compared with the prior art are the same as those of the above method for obtaining high-precision irradiance data, and will not be elaborated here.

[0019] The present invention also provides a system for obtaining high-precision irradiation data, including a computer-readable storage medium storing a computer program and a processor. When the computer program is read and run by the processor, the method for obtaining high-precision irradiation data as described above is implemented. The advantages of the system for obtaining high-precision irradiation data and the above method for obtaining high-precision irradiation data over the prior art are the same and will not be elaborated here.

[0020] The present invention also provides a computer-readable storage medium storing a computer program. When the computer program is read and run by a processor, the method for obtaining high-precision irradiation data as described above is implemented. The advantages of the computer-readable storage medium and the above method for obtaining high-precision irradiation data over the prior art are the same and will not be elaborated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 Schematic diagram of the method for obtaining high-precision irradiation data according to an embodiment of the present invention Figure One ;

[0022] Figure 2 Schematic diagram of the method for obtaining high-precision irradiation data according to an embodiment of the present invention Figure Two ;

[0023] Figure 3 Schematic diagram of the construction process of the calibration model for low-precision irradiation data according to an embodiment of the present invention;

[0024] Figure 4 Schematic diagram of the fitting process of the sensitivity of the low-precision irradiation meter and the high-precision irradiation meter based on the random forest algorithm according to an embodiment of the present invention;

[0025] Figure 5 Schematic diagram of the construction process of the conversion model for irradiation data between different regions according to an embodiment of the present invention;

[0026] Figure 6 Schematic diagram of the construction process of the conversion model for horizontal and inclined plane irradiation data according to an embodiment of the present invention;

[0027] Figure 7 Schematic diagram of the GRU neural network structure according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0028] First, some professional terms are explained.

[0029] (1) Solar radiation: Refers to the energy transmitted by the sun in the form of electromagnetic waves.

[0030] (2) Irradiation meter: A sensor that converts solar radiation energy into an electrical signal, used to detect the irradiation intensity of the sun, that is, irradiance, and is used in conjunction with a specially developed data collector.

[0031] (3) Direct-diffuse separation model: The total solar irradiance on a horizontal plane consists of the direct irradiance and the diffuse irradiance on the horizontal plane. Currently, the total solar radiation meter can only obtain the total solar irradiance. The direct-diffuse separation model can decompose the total irradiance into the direct irradiance and the diffuse irradiance.

[0032] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following detailed description of specific embodiments of the present invention will be given with reference to the accompanying drawings.

[0033] As Figure 1 shown, an embodiment of the present invention provides a method for obtaining high-precision irradiance data, including: obtaining low-precision horizontal irradiance data of a reference region, calibrating the low-precision horizontal irradiance data through a low-precision irradiance data calibration model to obtain high-precision horizontal irradiance data of the reference region; inputting meteorological data, geographical data, and the high-precision horizontal irradiance data of the reference region into an irradiance data conversion model between different regions to predict high-precision horizontal irradiance data of any target region; and converting the high-precision horizontal irradiance data of the any target region into high-precision inclined-plane irradiance data of any inclination angle of the any target region through a horizontal-to-inclined-plane irradiance data conversion model.

[0034] Specifically, in this embodiment, as Figure 2 shown, the method for obtaining high-precision irradiance data includes: (1) By using the low-precision horizontal irradiance data C1 of the reference region and calibrating it through a low-precision irradiance data calibration model, high-precision horizontal irradiance data C1 of the reference region can be obtained ， ; (2) Inputting the high-precision horizontal irradiance data of the reference region and meteorological and geographical data into an irradiance data conversion model between different regions, high-precision horizontal irradiance data of an adjacent target region can be predicted, that is, high-precision horizontal irradiance data C2 of any target region can be obtained ， ; (3) After conversion by a horizontal-to-inclined-plane irradiance data conversion model (i.e., a horizontal-to-inclined-plane irradiance data conversion model with any inclination angle), high-precision inclined-plane irradiance data C2 of any inclination angle of any target region can be obtained ，， .

[0035] Based on a low-cost and low-precision horizontal irradiance meter, this embodiment can provide accurate horizontal irradiance data, replace expensive high-precision irradiance meters, and has higher accuracy compared to irradiance data calculated through satellite data, etc., meeting the high-precision irradiance data requirements for power station site selection and construction, power station operation and maintenance evaluation, etc.; at the same time, it can provide accurate inclined-plane irradiance data of any inclination angle without the need to install irradiance meters with any inclination; in addition, based on the irradiance data of the reference region, it can provide high-precision inclined-plane irradiance data of any inclination angle in an adjacent target region without the need to install irradiance meters in the target region, significantly reducing costs.

[0036] Among them, the sensors used to obtain irradiance data are divided into Class A, Class B, and Class C. In terms of accuracy, A > B > C. The low-precision level irradiance data can be obtained by Class C sensors, and the high-precision level irradiance data can be obtained by Class A sensors.

[0037] In this embodiment, by calibrating, optimizing, converting, and transforming the low-precision level irradiance data of the reference area, the prediction of the irradiance data of the inclined plane with any inclination angle in any target area is realized. Compared with the scheme of directly using satellite data to obtain irradiance data, the accuracy is higher, the number of installed irradiance meters is reduced, and the cost of obtaining irradiance data is effectively reduced.

[0038] Optionally, the construction process of the low-precision irradiance data calibration model includes: eliminating systematic errors for different irradiance meters; deploying low-precision and high-precision irradiance meters with the same inclination angle at the same location to collect low-precision total irradiance data and high-precision total irradiance data; and fitting the sensitivities of the low-precision irradiance meter and the high-precision irradiance meter based on the random forest algorithm according to the low-precision total irradiance data and the high-precision total irradiance data to construct the low-precision irradiance data calibration model.

[0039] Specifically, in this embodiment, combined with Figure 3 As shown, the construction process of the low-precision irradiance data calibration model includes: (1) eliminating systematic errors for different irradiance meters; (2) installing low-precision and high-precision irradiance meters at the same location and the same angle to collect irradiance data; (3) using the random forest algorithm for data training and model construction.

[0040] Among them, the method for eliminating the systematic error of the device is as follows: Place different irradiance meters at the same location under sunny weather, and place a reference meter at the same time. Adjust the inclination angles of all irradiance meters to horizontal. After confirming that there is no occlusion on site, collect the irradiance values under sunny weather for three days. Compare the data of the other irradiance meters with the reference meter, and eliminate the original systematic errors of all irradiance meters by adjusting the irradiance coefficient parameters.

[0041] Among them, data collection is to deploy low-precision and high-precision irradiance meters at the same location with the same inclination angle of the irradiance meter. The irradiance data collection intervals are the same, both 20s. At the same time, data collection needs to cover different types of weather such as sunny, cloudy, overcast, light rain, shower, moderate rain, heavy rain, etc.

[0042] Among them, after collecting low-precision total irradiance data and high-precision total irradiance data, preprocessing is also required. The data preprocessing methods are as follows: a. Abnormalities in the irradiance data acquisition device or communication device are identified through digital judgment functions / empty judgment functions; b. When the irradiance data does not refresh, it is judged whether it is a dead value by identifying whether the numerical values of consecutive points are exactly the same; c. When the numerical values are abnormally large or small, the real-time irradiance data is judged based on irradiance data and power generation data under the same type of weather, etc.

[0043] There are non-linear differences in sensitivity between low / high-precision irradiance meters, and the random forest algorithm can be used for fitting. The random forest has excellent accuracy, can effectively run on large datasets, and can evaluate the importance of each feature in classification problems.

[0044] In this embodiment, the sensitivity of the low-precision irradiance meter and the sensitivity of the high-precision irradiance meter are fitted through the random forest algorithm, so that high-precision irradiance data can be obtained at a lower cost, effectively reducing the cost of obtaining irradiance data.

[0045] Optionally, the fitting of the sensitivity of the low-precision irradiance meter and the sensitivity of the high-precision irradiance meter based on the random forest algorithm includes: constructing an original training set; performing random sampling with replacement on the original training set, randomly selecting multiple features in each sampled training set to establish different classification models, and determining the optimal classification by voting to perform the fitting of the sensitivity of the low-precision irradiance meter and the sensitivity of the high-precision irradiance meter.

[0046] Specifically, in this embodiment, as shown in Figure 4 The fitting of the sensitivity of the low-precision irradiance meter and the sensitivity of the high-precision irradiance meter based on the random forest algorithm includes: (1) constructing an original training set; (2) performing random sampling with replacement on the original training set to form multiple training sets, randomly selecting m features in each sampled training set to establish classification model 1, classification model 2, and classification model 3, etc.; (3) determining the optimal classification by voting to perform the fitting of the sensitivity of the low-precision irradiance meter and the sensitivity of the high-precision irradiance meter.

[0047] Among them, the original training set includes low-precision total irradiance data and high-precision total irradiance data. Through algorithm parameter tuning and error weights during training, different results are output under different parameters and error weights, that is, the optimal classification is determined by voting, obtaining the corresponding input dataset and output dataset of the low-precision irradiance data calibration model, so that the low-precision irradiance data calibration model can be constructed.

[0048] In this embodiment, the sensitivity of the low-precision irradiance meter and the sensitivity of the high-precision irradiance meter are fitted through the random forest algorithm, so that high-precision irradiance data can be obtained at a lower cost, effectively reducing the cost of obtaining irradiance data.

[0049] Optionally, the process of constructing the irradiation data conversion model between different regions includes: eliminating systematic errors for different irradiance meters; establishing multiple typical ground meteorological stations within the same region to obtain the horizontal total irradiance data, numerical weather forecasts, and geographical locations of each place; using deep data mining to obtain the influence degrees of the numerical weather forecasts and the geographical locations, and establishing the irradiation data conversion model between different regions according to the horizontal total irradiance data and the influence degrees.

[0050] Specifically, in this embodiment, as shown in Figure 5 For multiple photovoltaic power stations located in adjacent regions, due to similar geographical location conditions and under the action of similar weather and other meteorological conditions, the solar radiation received by the photovoltaic power stations shows a certain degree of correlation. The process of constructing the irradiation data conversion model between different regions includes: (1) eliminating systematic errors for different irradiance meters; (2) by establishing multiple typical ground meteorological stations within the same region, obtaining the horizontal total irradiance data of each place (referring to each typical ground meteorological station), and simultaneously obtaining the numerical weather forecasts and geographical location models of each place; (3) using the method of deep data mining to obtain the influence degrees of the numerical weather forecasts and the geographical locations, establishing the irradiation data conversion model between different regions, further obtaining the horizontal irradiance data of any target region, and realizing the prediction of irradiation data without ground meteorological equipment.

[0051] Among them, establishing the irradiation data conversion model between different regions according to the horizontal total irradiance data and the influence degrees can be achieved through model training. Taking the horizontal total irradiance data and the influence degrees of each place as inputs and the horizontal irradiance data of any target region as outputs for training, so as to establish the irradiation data conversion model between different regions.

[0052] Among them, after obtaining the horizontal total irradiance data, numerical weather forecasts, and geographical locations of each place, preprocessing is also required. The preprocessing method of the data refers to the above text and will not be elaborated here.

[0053] In this embodiment, based on the horizontal irradiance data of the reference region within the area, the horizontal irradiance data of the adjacent target region is converted to realize the prediction of irradiation data without ground meteorological equipment, effectively reducing the cost of obtaining irradiation data.

[0054] Optionally, the process of constructing the horizontal and inclined surface irradiation data conversion model includes: eliminating systematic errors for different irradiance meters; arranging a horizontal irradiance meter and inclined surface irradiance meters with different inclinations at the same location to collect the horizontal total irradiance data of the typical region and the inclined total irradiance data of the typical region, and simultaneously collecting various meteorological index data; based on the characteristic relationship between the irradiation data and the meteorological index data, using the GRU neural network model to construct a direct and diffuse separation model; constructing the horizontal and inclined surface irradiation data conversion model according to the direct and diffuse separation model and the Perez physical model.

[0055] Specifically, in this embodiment, in combination with Figure 6 As shown, the construction process of the horizontal and inclined plane irradiance data conversion model includes: (1) eliminating systematic errors for different irradiance meters; (2) arranging a horizontal irradiance meter and inclined plane irradiance meters with different inclinations at the same location, and simultaneously collecting data such as solar altitude angle, hour angle, azimuth angle, temperature and humidity, wind speed, wind direction, altitude, air pressure, clear sky coefficient, etc. at this location; (3) based on the characteristic relationship between irradiance data and influencing factors, using the GRU neural network model to construct a direct and diffuse separation model, in combination with Figure 7 As shown, X t-1 , X t and X t+1 represent input variables, G represents a unit, C t-1 , C t , C t+1 , H t-1 , H t and H t+1 represent state variables. The GRU neural network can use the current input and the previous output to generate the output at the current moment in the same unit, and has good time series data processing ability. The solar irradiance is a variable with strictly periodic changes, and the magnitude correlation between the solar irradiance at the current moment and the solar irradiance at the previous moment is relatively high, making it easier to train and having a better convergence time. (4) Through the direct and diffuse separation model, the direct and diffuse irradiance data on the horizontal plane can be obtained, and then based on the Perez physical model, the direct and diffuse irradiance data on the inclined plane at any inclination can be obtained, and further the total irradiance data on the inclined plane at this inclination can be obtained.

[0056] Among them, typical regions need to cover all solar energy resource level regions, and generally can be divided according to the resource level. For example, the most abundant regions (such as Lhasa and other regions with sufficient sunlight), very abundant regions (such as Kunming), abundant regions (such as Guangzhou), and general regions (such as Chongqing).

[0057] Among them, the mean square error is selected as the training error judgment index during the training process, and the formula is as follows:

[0058]

[0059] In the formula: MSE is the mean square error, is the value predicted by the model, is the measured value, and n is the data volume.

[0060] Among them, after collecting the horizontal total irradiance data of typical regions, the inclined total irradiance data of typical regions, and various meteorological index data, preprocessing is also required. For the data preprocessing method, refer to the above text and will not be elaborated here.

[0061] In this embodiment, a direct-diffuse separation model is constructed through a GRU neural network model to construct a conversion model for horizontal and inclined-plane irradiance data, so that the inclined-plane irradiance data at any inclination angle can be obtained without installing any inclined-plane irradiance meter, effectively reducing the cost of obtaining irradiance data.

[0062] Optionally, the conversion of the high-precision horizontal irradiance data of any target area into the high-precision inclined-plane irradiance data at any inclination angle of any target area by the horizontal and inclined-plane irradiance data conversion model includes: inputting the high-precision horizontal irradiance data of any target area into the direct-diffuse separation model to obtain the direct and diffuse irradiance data on the horizontal plane, inputting the direct and diffuse irradiance data on the horizontal plane into the Perez physical model to obtain the direct and diffuse irradiance data on the inclined plane at any inclination angle, and obtaining the high-precision inclined-plane irradiance data at any inclination angle of any target area according to the direct and diffuse irradiance data on the inclined plane at any inclination angle.

[0063] Specifically, in this embodiment, the direct and diffuse irradiance data on the horizontal plane can be obtained through the direct-diffuse separation model, and then the direct and diffuse irradiance data on the inclined plane at any inclination angle can be obtained based on the Perez physical model, and further the total irradiance data on the inclined plane at this inclination angle can be obtained.

[0064] In this embodiment, by using the direct-diffuse separation model and the Perez physical model to convert the high-precision horizontal irradiance data of any target area into the high-precision inclined-plane irradiance data at any inclination angle of any target area, the inclined-plane irradiance data at any inclination angle can be obtained without installing any inclined-plane irradiance meter, effectively reducing the cost of obtaining irradiance data.

[0065] Optionally, the meteorological index data includes solar altitude angle, hour angle, azimuth angle, temperature and humidity, wind speed, wind direction, altitude, air pressure, and clear sky coefficient.

[0066] Specifically, in this embodiment, the selection of typical areas mainly considers the following points:

[0067] (1) Geographic longitude and latitude

[0068] The geographical location directly affects the solar radiation energy. For example, the solar irradiance decreases gradually from low latitudes to high latitudes.

[0069] (2) Component tilt angle

[0070] The component tilt angle directly affects the received irradiance. When deploying inclined irradiance meters, more angles should be covered as much as possible to ensure data difference.

[0071] (3) Geomorphic features

[0072] Different topographies and landforms can affect the local microclimate, resulting in differences in local temperature, humidity, meteorological data, etc. It is recommended to deploy for different types of power stations, such as: mountain power stations, water surface power stations, industrial and commercial rooftops, etc.

[0073] (4) Altitude

[0074] The higher the altitude, the shorter the atmospheric optical path, the better the atmospheric transparency, the smaller the weakening degree by meteorological factors, and the higher the proportion of direct solar radiation.

[0075] Considering the correlation and predictability with irradiance, it is determined that meteorological indicators such as solar altitude angle, hour angle, azimuth angle, temperature and humidity, wind speed, wind direction, altitude, air pressure, and clear sky coefficient are used as model inputs. Among them, the clear sky coefficient k refers to the transparency of the atmosphere, which mainly reflects the influence of the atmosphere on the process of solar radiation from outside the atmosphere to the ground surface, and is the ratio of the total solar radiation incident on the horizontal plane of the ground surface to the astronomical radiation.

[0076] In this embodiment, by setting the meteorological indicator data to include solar altitude angle, hour angle, azimuth angle, temperature and humidity, wind speed, wind direction, altitude, air pressure, and clear sky coefficient, the training effect of the GRU neural network model is improved, which is beneficial to improving the accuracy of the horizontal and inclined surface irradiance data conversion model.

[0077] Another embodiment of the present invention provides a device for obtaining high-precision irradiance data, including: a calibration module, configured to obtain low-precision horizontal irradiance data of a reference region, and calibrate the low-precision horizontal irradiance data through a low-precision irradiance data calibration model to obtain high-precision horizontal irradiance data of the reference region; a conversion module, configured to input the high-precision horizontal irradiance data, meteorological data, and geographical data of the reference region into an irradiance data conversion model between different regions, and predict high-precision horizontal irradiance data of any target region; a transformation module, configured to convert the high-precision horizontal irradiance data of any target region into high-precision inclined surface irradiance data of any inclination angle of any target region through a horizontal and inclined surface irradiance data conversion model.

[0078] Another embodiment of the present invention provides a system for obtaining high-precision irradiance data, including a computer-readable storage medium storing a computer program and a processor. When the computer program is read and run by the processor, the above method for obtaining high-precision irradiance data is implemented.

[0079] Another embodiment of the present invention provides a computer-readable storage medium, which stores a computer program. When the computer program is read and run by a processor, the above method for obtaining high-precision irradiance data is implemented.

[0080] Although the present invention is disclosed as above, the scope of protection of the present invention is not limited thereto. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention, and these changes and modifications will all fall within the scope of protection of the present invention.

Claims

1. A method for obtaining high-precision irradiation data, characterized in that, Including: Obtain the low-precision horizontal irradiance data of the reference region, and calibrate the low-precision horizontal irradiance data through a low-precision irradiance data calibration model to obtain the high-precision horizontal irradiance data of the reference region; Input the meteorological data, geographical data, and the high-precision horizontal irradiance data of the reference region into an irradiance data conversion model between different regions, and predict the high-precision horizontal irradiance data of any target region; Convert the high-precision horizontal irradiance data of the any target region into high-precision inclined-plane irradiance data with any inclination angle of the any target region through a horizontal-to-inclined-plane irradiance data conversion model; Among them, the construction process of the low-precision irradiance data calibration model includes: Eliminate systematic errors for different irradiance meters; Deploy low-precision irradiance meters and high-precision irradiance meters with the same inclination angle at the same location to collect low-precision total irradiance data and high-precision total irradiance data; Based on the random forest algorithm, fit the sensitivities of the low-precision irradiance meter and the high-precision irradiance meter according to the low-precision total irradiance data and the high-precision total irradiance data to construct the low-precision irradiance data calibration model.

2. The method for obtaining high-precision irradiation data according to claim 1, wherein The fitting of the sensitivities of the low-precision irradiance meter and the high-precision irradiance meter based on the random forest algorithm includes: Construct an original training set; Perform random sampling with replacement on the original training set. Randomly select multiple features in each sampled training set to establish different classification models, and determine the optimal classification by voting to fit the sensitivities of the low-precision irradiance meter and the high-precision irradiance meter.

3. The method for obtaining high-precision irradiation data according to claim 1, characterized in that, The construction process of the irradiance data conversion model between different regions includes: Eliminate systematic errors for different irradiance meters; Establish multiple typical ground meteorological stations within the same region to obtain the horizontal total irradiance data, numerical weather forecasts, and geographical locations of each place; Use deep data mining to obtain the influence degrees of the numerical weather forecasts and the geographical locations, and establish the irradiance data conversion model between different regions according to the horizontal total irradiance data and the influence degrees.

4. The method for obtaining high-precision irradiation data according to claim 1, characterized in that The construction process of the horizontal-to-inclined-plane irradiance data conversion model includes: Eliminate systematic errors for different irradiance meters; Arrange a horizontal irradiance meter and inclined-plane irradiance meters with different inclination angles at the same location to collect the horizontal total irradiance data of the typical region and the inclined total irradiance data of the typical region, and simultaneously collect various meteorological index data; Based on the characteristic relationship between the irradiance data and the meteorological index data, use a GRU neural network model to construct a direct-diffuse separation model; Construct the horizontal-to-inclined-plane irradiance data conversion model according to the direct-diffuse separation model and the Perez physical model.

5. The method for obtaining high-precision irradiation data according to claim 4, wherein The conversion of the high-precision horizontal irradiance data of the any target region into high-precision inclined-plane irradiance data with any inclination angle of the any target region through the horizontal-to-inclined-plane irradiance data conversion model includes: Input the high-precision horizontal irradiance data of the arbitrary target area into the direct-diffuse separation model to obtain the direct and diffuse irradiance data on the horizontal plane. Input the direct and diffuse irradiance data on the horizontal plane into the Perez physical model to obtain the direct and diffuse irradiance data on the inclined plane with an arbitrary inclination angle, and obtain the high-precision inclined-plane irradiance data with an arbitrary inclination angle in the arbitrary target area according to the direct and diffuse irradiance data on the inclined plane with an arbitrary inclination angle.

6. The method for obtaining high-precision irradiation data according to claim 4, wherein The meteorological index data includes solar altitude angle, hour angle, azimuth angle, temperature and humidity, wind speed, wind direction, altitude, air pressure, and clear sky coefficient.

7. An apparatus for obtaining high-precision irradiation data, characterized in that, It includes: A calibration module, configured to obtain the low-precision horizontal irradiance data of the reference area, and calibrate the low-precision horizontal irradiance data through a low-precision irradiance data calibration model to obtain the high-precision horizontal irradiance data of the reference area; A conversion module, configured to input the meteorological data, geographical data, and the high-precision horizontal irradiance data of the reference area into an irradiance data conversion model between different areas, and predict to obtain the high-precision horizontal irradiance data of an arbitrary target area; A transformation module, configured to convert the high-precision horizontal irradiance data of the arbitrary target area into the high-precision inclined-plane irradiance data with an arbitrary inclination angle in the arbitrary target area through a horizontal-inclined plane irradiance data conversion model; Among them, the construction process of the low-precision irradiance data calibration model includes: Eliminating systematic errors for different irradiance meters; Deploying low-precision irradiance meters and high-precision irradiance meters with the same inclination angle at the same location to collect low-precision total irradiance data and high-precision total irradiance data; Based on the random forest algorithm, fitting the sensitivities of the low-precision irradiance meter and the high-precision irradiance meter according to the low-precision total irradiance data and the high-precision total irradiance data to construct the low-precision irradiance data calibration model.

8. A system for obtaining high-precision irradiation data, characterized in that, It includes a computer-readable storage medium storing a computer program and a processor. When the computer program is read and run by the processor, the method for obtaining high-precision irradiance data according to any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program. When the computer program is read and run by the processor, the method for obtaining high-precision irradiance data according to any one of claims 1 to 6 is implemented.

Images