Training method of minute-level inclined plane irradiation prediction model, medium and electronic equipment
By training the minute-level inclined surface radiation prediction model, using data of different time scales and weather types, the accuracy problem of prediction of minute-level irradiation data in the inclined surface of the photovoltaic power station is solved, and the stability of the power generation of the photovoltaic power station and the grid scheduling are achieved. support.
Patent Information
- Application Number
- CN202310625230.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-05-30
- Publication Date
- 2024-12-03
AI Technical Summary
The existing technology is difficult to achieve high-precision prediction of minute-level irradiation data of the inclined surface of the photovoltaic power station, resulting in unstable power generation of the photovoltaic power station, affecting the grid scheduling and planning, and traditional methods are difficult to effectively predict the ultra-short-term solar tracking photovoltaic systems and photovoltaic power stations. Irradiation data.
By obtaining minute-level training data, including input parameters of different time scales and weather types, a neural network model is used to train the inclined surface irradiation prediction model to evaluate the linearity and accuracy of the model to ensure the accuracy of the prediction results.
It realizes minute-level prediction of the irradiation intensity of the inclined surface, improves prediction accuracy, is suitable for efficient operation of photovoltaic power plants and grid management, and enhances the economic and reliability of the power grid.
Smart Images

Figure CN119066410A_ABST
Abstract
Description
Technical Field
[0001] The present application belongs to the field of artificial intelligence technology and relates to a training method, and in particular to a training method, medium and electronic equipment for a minute-level inclined surface irradiation prediction model. Background Art
[0002] The power generation of photovoltaic power stations is affected by climate and geographical location, and its fluctuation is large and unstable. When photovoltaic power stations are connected to the large power grid, they will cause significant harmonic interference to the power grid, which is not conducive to the dispatch and planning of electric energy. From the perspective of energy system management of the main grid or microgrid, achieving high-precision prediction of the short-term irradiance of the photovoltaic array inside the photovoltaic power station can improve the overall economy and operational reliability of the main grid or microgrid. In addition, in practical applications, in order to improve efficiency and avoid problems such as dust accumulation, photovoltaic modules are often installed on inclined surfaces. Furthermore, some application scenarios, such as solar tracking photovoltaic systems and photovoltaic power station intelligent operation and maintenance systems, require high-precision ultra-short-term irradiance prediction data for photovoltaic power stations. Therefore, how to achieve high-precision prediction of minute-level irradiance data on inclined surfaces has become one of the technical problems that technicians in related fields need to solve urgently. Summary of the invention
[0003] The embodiments of the present application provide a training method, medium and electronic device for a minute-level inclined surface irradiation prediction model, which are used to solve the above-mentioned problems existing in the prior art.
[0004] In a first aspect, an embodiment of the present application provides a training method for a minute-level inclined surface irradiation prediction model, the training method comprising: obtaining training data, the training data being minute-level data, the training data comprising input parameters and their corresponding measured data; using the training data to train the minute-level inclined surface irradiation prediction model; and obtaining the linearity and accuracy of the prediction model.
[0005] In an implementation of the first aspect, the training data includes first training data and second training data, the input parameters of the first training data include the irradiation intensity of the inclined surface at a first reference angle, and the input parameters of the second training data include the irradiation intensity of the inclined surface at a second reference angle; using the training data to train the minute-level inclined surface irradiation prediction model includes: using the first training data to train the minute-level inclined surface irradiation prediction model to obtain a first prediction model, and using the second training data to train the minute-level inclined surface irradiation prediction model to obtain a second prediction model; obtaining the linearity and accuracy of the prediction model includes: obtaining the linearity and accuracy of the first prediction model and the second prediction model.
[0006] In an implementation manner of the first aspect, the training method further includes: obtaining a contribution of input parameters to a prediction result, the input parameters including input parameters of the first training data and input parameters of the second training data.
[0007] In an implementation of the first aspect, the training data includes third training data and fourth training data, the input parameter of the third training data is data of a first time scale, and the input parameter of the fourth training data is data of a second time scale; using the training data to train the minute-level inclined surface irradiation prediction model includes: using the third training data to train the minute-level inclined surface irradiation prediction model to obtain a third prediction model, and using the fourth training data to train the minute-level inclined surface irradiation prediction model to obtain a fourth prediction model; obtaining the linearity and accuracy of the prediction model includes: obtaining the linearity and accuracy of the third prediction model and the fourth prediction model.
[0008] In an implementation of the first aspect, the training method further includes: averaging the one-minute scale data at the first time scale to obtain the input parameters of the third training data, and averaging the one-minute scale data at the second time scale to obtain the input parameters of the fourth training data; or sampling the one-minute scale data at intervals of the first time scale to obtain the input parameters of the third training data, and sampling the one-minute scale data at intervals of the second time scale to obtain the input parameters of the fourth training data.
[0009] In an implementation of the first aspect, the training data includes fifth training data and sixth training data, and the input parameters of the fifth training data and the sixth training data correspond to different weather types; using the training data to train the minute-level inclined surface irradiation prediction model includes: using the fifth training data to train the minute-level inclined surface irradiation prediction model to obtain a fifth prediction model, and using the sixth training data to train the minute-level inclined surface irradiation prediction model to obtain a sixth prediction model; obtaining the linearity and accuracy of the prediction model includes: obtaining the linearity and accuracy of the fifth prediction model and the sixth prediction model.
[0010] In an implementation of the first aspect, the training data includes seventh training data, and the seventh training data is data of the first region. The method also includes: using the seventh training data to train the minute-level inclined surface irradiation prediction model to obtain a seventh prediction model; using the seventh prediction model to predict the target angle inclined surface irradiation intensity of the first region and the second region respectively to obtain the prediction results of the first region and the prediction results of the second region; determining the generalization ability of the seventh prediction model according to the prediction results of the first region and the prediction results of the second region.
[0011] In an implementation of the first aspect, the seventh training data is data of the first region in the first weather type, and the seventh prediction model is used to predict the target angle inclined surface irradiance of the first region and the second region respectively to obtain the prediction result of the first region and the prediction result of the second region, including: using the seventh prediction model to predict the target angle inclined surface irradiance of the first region in the first weather type to obtain the prediction result of the first region; using the seventh prediction model to predict the target angle inclined surface irradiance of the second region in the first weather type to obtain the prediction result of the second region.
[0012] In a second aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method described in any one of the first aspects of the present application is implemented.
[0013] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a memory storing a computer program; and a processor communicatively connected to the memory, for executing any of the methods described in the first aspect of the present application when the computer program is called.
[0014] The training method of the minute-level inclined surface irradiation prediction model provided in the embodiment of the present application can train the minute-level inclined surface irradiation prediction model using minute-level training data. The prediction model after training can predict the inclined surface irradiation intensity at the minute level. In addition, the training method can also evaluate the linearity and accuracy of the prediction model to ensure the accuracy of the prediction results. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] Figure 1 Shown is a schematic diagram of an application scenario of the training method for the minute-level inclined surface irradiation prediction model described in an embodiment of the present application.
[0016] Figure 2 Shown is a flow chart of the training method of the minute-level inclined surface irradiation prediction model described in an embodiment of the present application.
[0017] FIG. 3A to FIG. 7C Shown is a scatter plot of the predicted values and the measured values in the examples of this application.
[0018] Fig. 8A Shown are several step flow charts in the embodiments of the present application.
[0019] FIG. 8B to FIG. 9D Shown is a scatter plot of the predicted values and the measured values in the examples of this application.
[0020] Fig.10 Shown is a schematic diagram of the structure of an electronic device in an embodiment of the present application.
[0021] Component number description
[0022] 1 Electronic equipment
[0023] 11 Memory
[0024] 12 Display
[0025] 13 General Purpose Processor
[0026] 131 CPU
[0027] 132 Neural Network Processor
[0028] 10 Electronic devices
[0029] 101 Memory
[0030] 102 Processor
[0031] 103 Display
[0032] Steps S21 to S23
[0033] Steps S81 to S83 DETAILED DESCRIPTION
[0034] The following describes the embodiments of the present application through specific examples, and those skilled in the art can easily understand other advantages and effects of the present application from the contents disclosed in this specification. The present application can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present application. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.
[0035] It should be noted that the illustrations provided in the following embodiments are only used to illustrate the basic concept of the present application in a schematic manner, and therefore the illustrations only show components related to the present application rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.
[0036] The present application embodiment provides a training method for a minute-level inclined surface irradiation prediction model. The training method can be applied, for example, to Figure 1 The electronic device 1 shown. Figure 1 As shown, the electronic device 1 includes a memory 11 , a display 12 and at least one general purpose processor 13 .
[0037] The memory 11 may include a volatile memory, such as a random access memory (RAM), a cache. The memory 11 may also include a non-volatile memory, such as a read-only memory (ROM), a flash memory, a hard disk drive (HDD), or a solid-state drive (SSD). The memory 11 may be used to store program instructions for the processor to call and execute at least a portion of the training method for the minute-level inclined surface irradiation prediction model provided in the embodiment of the present application.
[0038] The display 12 may specifically include a display screen (display panel). In some implementations, the display panel may be configured in the form of a liquid crystal display (LCD), an organic light-emitting diode (OLED), or the like. In addition, the display 12 may also be a touch panel (touch screen, touch screen), which may include a display screen and a touch-sensitive surface. When the touch-sensitive surface detects a touch operation on or near it, it is transmitted to the general processor 13 to determine the type of touch event, and then the general processor 13 provides a corresponding visual output on the display device according to the type of touch event.
[0039] The general processor 13 can be any type of device capable of processing electronic instructions. In the embodiment of the present application, the electronic device 1 may include one or more general processors 13, such as a central processing unit (CPU) 131 and a neural network processor (NPU) 132. One or two. In addition, it may also include a graphics processor (GPU), a microprocessor, a microcontroller, a main processor, a controller, and an ASIC (Application Specific Integrated Circuit). The general processor 13 is configured to execute various types of digital storage instructions, such as software or firmware programs stored in the memory 11, which enables the electronic device 1 to provide a variety of services. For example, the processor 11 is capable of executing programs or processing data to execute at least a portion of the training method for the minute-level inclined surface irradiation prediction model provided in the embodiment of the present application.
[0040] The main function of the central processing unit 131 is to parse computer instructions and process data in computer software, realize overall control of the electronic device 1, and control and allocate all hardware resources of the electronic device 1 (such as storage resources, communication resources, I / O interfaces, etc.).
[0041] Neural network processor is a general term for new processors based on neural network algorithms and acceleration. It is specially designed for artificial intelligence and is used to accelerate neural network calculations and solve the problem of low efficiency of traditional chips in neural network calculations.
[0042] It should be noted that the name of the neural network processor does not constitute a limitation of the present application. For example, in other application scenarios, the neural network processor can also be transformed and replaced by other processors with similar functions, such as a tensor processor (Tensor Processing Unit, TPU), a deep learning processor (Deep learning Processing Unit, DPU), and so on.
[0043] Figure 2 The flowchart of the training method of the minute-level inclined surface irradiation prediction model provided by the embodiment of the present application is shown. Figure 2 As shown, the training method of the minute-level inclined surface irradiation prediction model provided in the embodiment of the present application includes the following steps S21 to S23.
[0044] S21, obtaining training data. The training data is minute-level data, including input parameters and their corresponding measured data.
[0045] In some possible implementations, the input parameters may include horizontal total irradiance, solar altitude angle, declination angle, solar incidence angle, atmospheric mass number, and clarity index, but the present application is not limited thereto.
[0046] S22, train the minute-level inclined surface irradiation prediction model using the training data. The minute-level inclined surface irradiation prediction model is a neural network model, which is used to predict the minute-level irradiation intensity of the target angle inclined surface. The target angle can be configured according to actual needs. For example, in some embodiments, the target angle can be 30°. In these embodiments, the minute-level inclined surface irradiation prediction model is used to predict the irradiation intensity of the 30° inclined surface.
[0047] In some possible implementations, the minute-level inclined surface irradiation prediction model includes 1 input layer, 1 hidden layer, and 1 output layer. The nodes in the input layer are only used to input data and do not perform calculations. Both the hidden layer and the output layer perform data calculations. The output result of the output layer is the irradiation intensity of the inclined surface at the target angle.
[0048] S23, obtaining the linearity and accuracy of the prediction model. The linearity of the prediction model refers to its linear property or fitting ability. If the prediction model shows a linear relationship between the input and output, that is, the output of the model can be obtained by a linear combination of inputs, then the prediction model has a high linearity. The accuracy of the prediction model is used to evaluate the correctness and precision of the model prediction results. The higher the accuracy of the prediction model, the closer its prediction results are to the measured values.
[0049] After training, the prediction model can be used to process input data to predict the minute-level irradiation intensity of the target angle inclined surface. Compared with traditional methods, the prediction model after training can obtain the irradiation intensity of the target angle inclined surface at a scale of 1 minute, and the prediction result has a higher accuracy.
[0050] In one embodiment of the present application, the training data includes first training data and second training data. The input parameters of the first training data include the irradiation intensity of the inclined surface at the first reference angle. The input parameters of the second training data include the irradiation intensity of the inclined surface at the second reference angle. In an embodiment of the present application, using the training data to train the minute-level inclined surface irradiation prediction model includes: using the first training data to train the minute-level inclined surface irradiation prediction model to obtain a first prediction model, and using the second training data to train the minute-level inclined surface irradiation prediction model to obtain a second prediction model. Obtaining the linearity and accuracy of the prediction model includes: obtaining the linearity and accuracy of the first prediction model and the second prediction model.
[0051] In some possible implementations, the irradiance intensity of the inclined surface at the first reference angle is the total horizontal irradiance, and the input parameters of the first training data include the total horizontal irradiance, the solar altitude angle, the solar incidence angle, the declination angle, the atmospheric mass number, and the clarity index. The first training data is used to train the minute-level inclined surface irradiance prediction model to obtain the first prediction model. The output parameter of the first prediction model is the irradiance intensity of the 30° inclined surface. Figure 3A The scatter plot of the predicted value and the measured value of the first prediction model in the embodiment of the present application is shown. Figure 3A It can be seen that the linearity and accuracy of the first prediction model are good. In addition, R2, NRMSE (Normalized Root Mean Squared Error, standardized root mean square error) and the like can be used to quantitatively evaluate the linearity and accuracy of the first prediction model.
[0052] In these implementations, the training method of the minute-level inclined surface irradiance prediction model may also include: obtaining the contribution of the input parameters of the first training data to the prediction result. For example, the random forest RF algorithm can be used to rank the contribution of the input parameters of the first training data. In some instances, the contribution of the input parameters of the first training data is, for example: the contribution of the clarity index is 35.4%, the contribution of the solar altitude angle is 21.3%, the contribution of the solar incidence angle is 15.8%, the contribution of the atmospheric mass number is 12.1%, the contribution of the horizontal total irradiance is 9.3%, and the contribution of the declination angle is 6.1%.
[0053] In some possible implementations, the irradiance intensity of the second reference angle inclined surface is the irradiance intensity of the 10° inclined surface. The input parameters of the second training data include horizontal total irradiance, irradiance intensity of the 10° inclined surface, solar altitude angle, solar incidence angle, declination angle, atmospheric mass number, and clarity index. The minute-level inclined surface irradiance prediction model is trained using the second training data to obtain a second prediction model. The output parameter of the second prediction model is the irradiance intensity of the 30° inclined surface. Figure 3B The scatter plot shows the predicted value and the measured value of the second prediction model in the embodiment of the present application. Figure 3A and Figure 3B It can be seen that the linearity and accuracy of the second prediction model are better than those of the first prediction model. In addition, R2, NRMSE, etc. can be used to quantitatively evaluate the linearity and accuracy of the second prediction model.
[0054] In these implementations, the training method of the minute-level inclined surface irradiance prediction model may also include: obtaining the contribution of the input parameters of the second training data to the prediction result. For example, the Random Forests (RF) algorithm can be used to sort the contribution of the input parameters of the second training data. In some instances, the contribution of the input parameters of the second training data is, for example: the contribution of the clarity index is 28.8%, the contribution of the solar incidence angle is 20.8%, the contribution of the atmospheric mass number is 19.3%, the contribution of the solar altitude angle is 18.8%, the contribution of the total irradiance of the 10° inclined surface is 8.9%, the contribution of the horizontal total irradiance is 3.5%, and the contribution of the declination angle is 0%.
[0055] In some possible implementations, the number of the second reference angle inclined surfaces is 2. In these implementations, the irradiance intensity of the second reference angle inclined surface is the irradiance intensity of the 10° inclined surface and the irradiance intensity of the 20° inclined surface. The input parameters of the second training data include horizontal total irradiance, total irradiance of the 10° inclined surface, total irradiance of the 20° inclined surface, solar altitude angle, solar incidence angle, declination angle, atmospheric mass number, and clarity index. The output parameter of the second prediction model is the irradiance intensity of the 30° inclined surface. Figure 3C The scatter plot shows the predicted value and the measured value of the second prediction model in the embodiment of the present application. Figure 3A , Figure 3B and Figure 3C It can be seen that these implementations have better linearity and accuracy.
[0056] In these implementations, the training method of the minute-level inclined surface irradiance prediction model may also include: obtaining the contribution of the input parameters of the second training data to the prediction result. For example, the random forest RF algorithm can be used to rank the contribution of the input parameters of the second training data. In some instances, the contribution of the input parameters of the second training data is, for example: the contribution of the total irradiance of the 20° inclined surface is 35.8%, the contribution of the clarity index is 23.4%, the contribution of the solar incidence angle is 11.5%, the contribution of the atmospheric mass number is 9.4%, the contribution of the solar altitude angle is 7.9%, the contribution of the total irradiance of the 10° inclined surface is 4.7%, the contribution of the horizontal total irradiance is 4.3%, and the contribution of the declination angle is 3.0%.
[0057] In summary, the accuracy of each model's prediction of the irradiance intensity of the 30° inclined surface is shown in Table 1. According to Table 1, adding the total irradiance of the 10° inclined surface and the 20° inclined surface to the input parameters can significantly improve the accuracy of the final prediction data. In addition, according to the data contribution, the parameters that have the greatest impact on the prediction results are the clarity index and the irradiance intensity of the inclined surface, and the closer the reference angle in the input parameters is to the target angle, the higher its importance.
[0058] Table 1. Summary of model evaluation parameters
[0059]
[0060] In one embodiment of the present application, the training data includes third training data and fourth training data. The input parameter of the third training data is data of the first time scale. The input parameter of the fourth training data is data of the second time scale. Using the training data to train the minute-level inclined surface irradiation prediction model includes: using the third training data to train the minute-level inclined surface irradiation prediction model to obtain a third prediction model, and using the fourth training data to train the minute-level inclined surface irradiation prediction model to obtain a fourth prediction model. Obtaining the linearity and accuracy of the prediction model includes: obtaining the linearity and accuracy of the third prediction model and the fourth prediction model.
[0061] In some possible implementations, the input parameter of the third training data is data of 1 minute scale. The minute-scale inclined surface irradiance prediction model is trained using the third training data to obtain a third prediction model. The scatter plot of the predicted value and the measured value of the third prediction model is as follows: Figure 3A shown.
[0062] In some possible implementations, the input parameter of the fourth training data is 5-minute scale data. In these implementations, the 1-minute scale data is averaged every 5 minutes as the input parameter of the fourth training data. The minute-level inclined surface irradiance prediction model is trained using the fourth training data to obtain the fourth prediction model. Figure 4A Shown are scatter plots of predicted and observed values of the fourth prediction model in these implementations.
[0063] In some other possible implementations, the input parameter of the fourth training data is data at a scale of 15 minutes. In these implementations, the average value of the data at a scale of 1 minute is taken for 15 minutes as the input parameter of the fourth training data. The minute-level inclined surface irradiance prediction model is trained using the fourth training data to obtain a fourth prediction model. Figure 4B Shown are scatter plots of predicted and observed values of the fourth prediction model in these implementations.
[0064] Table 2 shows a summary table of evaluation parameters of the above model in the embodiment of the present application. As shown in Table 2, the input parameters of the training data are obtained by taking the average of the 1-minute data within the interval, and the minute-level inclined surface irradiation prediction model is trained with these input parameters. Since the random errors generated in the process of partial measurement data can be reduced in the process of mean calculation, more accurate training data can be obtained, and the prediction accuracy of the prediction model can be improved.
[0065] Table 2. Summary of model evaluation parameters
[0066] Model R2 NRMSE Prediction model corresponding to 1 minute data 0.976 0.133 The prediction model corresponding to the 5-minute average 0.980 0.122 The prediction model corresponding to the 15-minute average 0.985 0.106
[0067] In some possible implementations, the input parameters of the fourth training data are 5-minute scale data. In these implementations, the input parameters of the fourth training data are obtained by sampling the 1-minute scale data at intervals of every 5 minutes. The minute-level inclined surface irradiance prediction model is trained using the fourth training data to obtain the fourth prediction model. Figure 5A Shown are scatter plots of predicted and observed values of the fourth prediction model in these implementations.
[0068] In some other possible implementations, the input parameters of the fourth training data are data of 15-minute scale. In these implementations, the input parameters of the fourth training data are obtained by sampling the data of 1-minute scale at intervals of 15 minutes. The minute-level inclined surface irradiance prediction model is trained using the fourth training data to obtain the fourth prediction model. Figure 5B Shown are scatter plots of predicted and observed values of the fourth prediction model in these implementations.
[0069] Table 3 shows a summary table of evaluation parameters of the above model in the embodiment of the present application. As shown in Table 3, the input parameters of the fourth training data are obtained by sampling the 1-minute data at intervals of a longer time scale, and the random error of the data points does not change, while the sample size of the training data is reduced.
[0070] Table 3. Summary of model evaluation parameters
[0071] Model R2 NRMSE Prediction model corresponding to 1 minute data 0.976 0.133 The prediction model corresponding to the 5-minute interval points 0.968 0.155 The prediction model corresponding to the 15-minute interval points 0.966 0.160
[0072] In one embodiment of the present application, the training data includes fifth training data and sixth training data. The input parameters of the fifth training data and the sixth training data correspond to different weather types. Using the training data to train the minute-level inclined surface irradiation prediction model includes: using the fifth training data to train the minute-level inclined surface irradiation prediction model to obtain a fifth prediction model, and using the sixth training data to train the minute-level inclined surface irradiation prediction model to obtain a sixth prediction model; obtaining the linearity and accuracy of the prediction model includes: obtaining the linearity and accuracy of the fifth prediction model and the sixth prediction model.
[0073] In some possible implementations, the clarity Kt value is used to distinguish weather types. In these implementations, the minute-level input parameters can be divided into three separate data sets based on the Kt value. The input parameters of the first data set correspond to low Kt values (for example, Kt values below 0.3), which are the input parameters of the first weather type. The input parameters of the second data set correspond to medium Kt values (for example, Kt values between 0.3 and 0.7), which are the input parameters of the second weather type. The input parameters of the third data set correspond to high Kt values (for example, Kt values above 0.7), which are the input parameters of the third weather type. The minute-level inclined surface irradiance prediction model is trained with the input parameters of the above three weather types, and the scatter plots corresponding to the obtained prediction models are as follows. Fig. 6A , 6B and 6C.
[0074] Table 4 shows a summary table of evaluation parameters of the above model in the embodiment of the present application. As shown in Table 4, compared with the prediction model that does not distinguish the Kt value, the prediction accuracy and linearity of the prediction model corresponding to the low Kt value data set are improved to a certain extent, and the prediction accuracy of the prediction model corresponding to the medium Kt value data set and the high Kt value data set is improved to a certain extent.
[0075] Table 4. Summary of model evaluation parameters
[0076] Model R2 NRMSE Prediction model that does not distinguish between Kt values 0.976 0.133 Prediction model corresponding to low Kt value 0.979 0.128 Prediction model corresponding to medium Kt value 0.970 0.091 Prediction model corresponding to high Kt value 0.802 0.088
[0077] In some possible implementations, the Kt' value is used to distinguish weather types. The Kt' value is a weighted Kt value. The specific calculation formula of the Kt' value is as follows: Formula 1.
[0078]
[0079] Where Kt is the clarity index and m is the atmospheric mass number.
[0080] In these implementations, the minute-level input parameters can be divided into three separate data sets based on the Kt' value. The input parameters of the first data set correspond to low Kt' values (for example, Kt' values are below 0.3), which are the input parameters of the first weather type. The input parameters of the second data set correspond to medium Kt' values (for example, Kt' values are between 0.3 and 0.7), which are the input parameters of the second weather type. The input parameters of the third data set correspond to high Kt' values (for example, Kt' values are above 0.7), which are the input parameters of the third weather type. The minute-level inclined surface irradiance prediction model is trained with the input parameters of the above three weather types, and the scatter plots corresponding to the obtained prediction models are as follows: Fig. 7A , 7B and 7C.
[0081] Table 5 shows a summary table of evaluation parameters of the above model in the embodiment of the present application. As shown in Table 5, compared with the prediction model that does not distinguish the Kt' value, the prediction accuracy and linearity of the prediction model corresponding to the low Kt' value and medium Kt' value data sets are improved to a certain extent, and the prediction accuracy of the prediction model corresponding to the high Kt' value data set is improved to a certain extent.
[0082] Table 5. Summary of model evaluation parameters
[0083] Model R2 NRMSE Prediction model that does not distinguish between Kt' values 0.976 0.133 Prediction model corresponding to low Kt' value 0.979 0.124 Prediction model for medium Kt' value 0.990 0.067 Prediction model corresponding to high Kt' value 0.930 0.088
[0084] In one embodiment of the present application, the training data includes seventh training data. The seventh training data is data of the first region, and may include, for example, total horizontal irradiance, solar altitude angle, declination angle, atmospheric mass number, clarity index, etc. Fig. 8A The training method of the minute-level inclined surface irradiation prediction model provided in the embodiment of the present application may also include steps S81 to S83.
[0085] S81, using the seventh training data to train the minute-level inclined surface irradiation prediction model to obtain a seventh prediction model.
[0086] S82, using the seventh prediction model to predict the target angle inclined surface irradiance intensity of the first region and the second region respectively, to obtain a prediction result of the first region and a prediction result of the second region.
[0087] S83, determining the generalization ability of the seventh prediction model according to the prediction result of the first region and the prediction result of the second region.
[0088] In some possible implementations, the first region is region A with low altitude and high scattering ratio, and the second region is region B with high altitude and low scattering ratio. In these implementations, the minute-level inclined surface irradiance prediction model is trained using the data of region A to obtain the seventh prediction model. The scatter plot of the target angle inclined surface irradiance intensity prediction of region A using the seventh prediction model is shown in FIG. Figure 8B As shown in the figure, the scatter plot of the seventh prediction model for predicting the irradiation intensity of the target angle inclined surface in area B (cross prediction) is as follows Figure 8C shown.
[0089] In some other possible implementations, the first region is region B with high altitude and low scattering ratio, and the second region is region A with low altitude and high scattering ratio. In these implementations, the minute-level inclined surface irradiation prediction model is trained using the data of region B to obtain the seventh prediction model. The scatter plot of the target angle inclined surface irradiation intensity prediction (cross prediction) of region A using the seventh prediction model is shown in the figure below: Fig.8D As shown in the figure, the scatter plot of the seventh prediction model for predicting the irradiation intensity of the target angle inclined surface in area B is as follows Fig. 8E shown.
[0090] Table 6 shows the evaluation parameters of the seventh prediction model. As shown in Table 6, the seventh prediction model trained according to the data of region A predicts the target angle tilt surface irradiation intensity of region A, and the seventh prediction model trained according to the data of region B predicts the target angle tilt surface irradiation intensity of region B, which is higher than the cross prediction (including the seventh prediction model trained according to the data of region A predicts the target angle tilt surface irradiation intensity of region B, and the seventh prediction model trained according to the data of region B predicts the target angle tilt surface irradiation intensity of region A).
[0091] Table 6. Summary of model evaluation parameters
[0092]
[0093] In one embodiment of the present application, the seventh training data is data of the first region in the first weather type. Using the seventh prediction model to predict the target angle tilt surface irradiance intensity of the first region and the second region respectively to obtain the prediction result of the first region and the prediction result of the second region includes: using the seventh prediction model to predict the target angle tilt surface irradiance intensity of the first region in the first weather type to obtain the prediction result of the first region; using the seventh prediction model to predict the target angle tilt surface irradiance intensity of the second region in the first weather type to obtain the prediction result of the second region.
[0094] In some possible implementations, Kt' values are used to distinguish weather types. In these implementations, the minute-level input parameters can be divided into three separate data sets based on the Kt' value. The input parameters of the first data set correspond to low Kt' values (for example, Kt' values below 0.3), which are the input parameters of the first weather type. The input parameters of the second data set correspond to medium Kt' values (for example, Kt' values between 0.3 and 0.7), which are the input parameters of the second weather type. The input parameters of the third data set correspond to high Kt' values (for example, Kt' values above 0.7), which are the input parameters of the third weather type.
[0095] In some possible implementations, the minute-level inclined surface irradiation prediction model is trained using the data of the first region (region A) at low Kt' values to obtain the seventh prediction model. The seventh prediction model is used to predict the inclined surface irradiation intensity at the target angle of region A at low Kt' values to obtain the prediction result of region A corresponding to the low Kt' value, and the seventh prediction model is used to predict the inclined surface irradiation intensity at the target angle of region B at low Kt' values (cross prediction) to obtain the prediction result of region B corresponding to the low Kt' value. The scatter plots of the above two prediction results are shown as follows: Fig. 9A and Fig. 9Bshown.
[0096] In some other possible implementations, the minute-level inclined surface irradiation prediction model is trained using the data of the second region (region B) at low Kt' values to obtain the seventh prediction model. The seventh prediction model is used to predict the inclined surface irradiation intensity at the target angle of region A at low Kt' values (cross prediction) to obtain the prediction result of region A corresponding to the low Kt' value, and the seventh prediction model is used to predict the inclined surface irradiation intensity at the target angle of region B at low Kt' values to obtain the prediction result of region B corresponding to the low Kt' value. The scatter plots of the above two prediction results are shown as follows: Fig. 9C and Fig.9D shown.
[0097] In some possible implementations, the data of region A at medium Kt' values can be used to obtain the seventh prediction model, and the seventh prediction model can be used to predict the prediction results of region A at medium Kt' values and region B at medium Kt' values (cross prediction).
[0098] In some possible implementations, the data of region B at medium Kt' value can be used to obtain the seventh prediction model, and the seventh prediction model can be used to predict the prediction results of region A at medium Kt' value (cross prediction) and the prediction results of region B at medium Kt' value.
[0099] In some possible implementations, the data of region A at high Kt' values can be used to obtain the seventh prediction model, and the seventh prediction model can be used to predict the prediction results of region A at high Kt' values and the prediction results of region B at high Kt' values (cross prediction).
[0100] In some possible implementations, the data of region B at high Kt' values can be used to obtain the seventh prediction model, and the seventh prediction model can be used to predict the prediction results of region A at high Kt' values (cross prediction) and the prediction results of region B at high Kt' values.
[0101] Table 7 shows a summary of the evaluation parameters of the above predictions. As shown in Table 7, the generalization ability of the piecewise cross-prediction of the minute-level inclined surface irradiation prediction model in different regions is better than that of the holistic cross-prediction. Among them, piecewise cross-prediction refers to the cross-prediction that distinguishes weather types, and holistic cross-prediction refers to the cross-prediction that does not distinguish weather types. In addition, the differences in the values of R2 and NRMSE show that using local data for model training can achieve better accuracy results. Furthermore, further subdividing the cross-prediction of the Kt' interval can further improve the accuracy based on the original high Kt' interval prediction.
[0102] Table 7. Summary of model evaluation parameters
[0103]
[0104] According to the above description, the embodiment of the present application provides a training method for a minute-level inclined surface irradiation prediction model. The minute-level inclined surface irradiation prediction model trained by the training method can be applied to the prediction of the irradiation intensity of the target angle inclined surface at a one-minute scale, and the prediction result obtained is higher than the traditional model. In addition, based on the sensitivity analysis of the parameters, adding reference angle inclined surface data to the training data is conducive to improving the accuracy of the prediction. Based on the granularity analysis of the data, taking the average value of the intervals of the training data can reduce the random error of the data. Based on the prediction analysis of weather types, predicting the irradiation intensity of the target angle inclined surface separately with different weather types can further improve the prediction accuracy. Based on the prediction analysis of different regional types, using the data of the local region to predict the irradiation intensity of the target angle inclined surface in the local area has higher accuracy. In addition, the weighted clarity index Kt' is used to further divide the weather types for cross-prediction, which can improve the cross-regional generalization ability of the model.
[0105] The protection scope of the training method for the minute-level inclined surface irradiation prediction model provided in the embodiment of the present application is not limited to the execution order of the steps listed in this embodiment. All solutions implemented by adding, reducing or replacing steps in the prior art based on the principles of the present application are included in the protection scope of the present application.
[0106] The embodiment of the present application also provides a computer-readable storage medium on which a computer program is stored. The computer program is executed to implement the training method of the minute-level inclined surface irradiation prediction model provided in any embodiment of the present application.
[0107] A person of ordinary skill in the art can understand that all or part of the steps in the method for implementing the above-mentioned embodiment can be completed by instructing a processor through a program, and the program can be stored in a computer-readable storage medium, and the storage medium is a non-transitory medium, such as a random access memory, a read-only memory, a flash memory, a hard disk, a solid-state hard disk, a magnetic tape, a floppy disk, an optical disc, and any combination thereof. The above-mentioned storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that includes one or more available media. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a digital video disc (DVD)), or a semiconductor medium (for example, a solid-state disk (SSD)), etc.
[0108] The present application also provides an electronic device. Fig.10The structure diagram of the electronic device 10 in one embodiment of the present application is shown. Fig.10 As shown, in this embodiment, the electronic device 10 includes a memory 101 and a processor 102 .
[0109] The memory 101 is used to store computer programs. In some possible implementations, the memory 101 may include: ROM, RAM, a disk, a USB flash drive, a memory card, or an optical disk, etc., various media that can store program codes.
[0110] In this embodiment, the memory 101 may include a computer system readable medium in the form of a volatile memory, such as a RAM and / or a cache memory. The electronic device 10 may further include other removable / non-removable, volatile / non-volatile computer system storage media. The memory 101 may include at least one program product, which has a set (e.g., at least one) of program modules, which are configured to perform the functions of each embodiment of the present application.
[0111] The processor 102 is connected to the memory 101 and is used to execute the computer program stored in the memory 101 so that the electronic device 10 executes the training method of the minute-level inclined surface irradiation prediction model.
[0112] In this embodiment, the processor 102 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.
[0113] In some possible implementations, the electronic device 10 provided in the embodiment of the present application may further include a display 103. The display 103 is communicatively connected to the memory 101 and the processor 102, and is used to display a graphical user interface (GUI) related to the training method of the minute-level inclined surface irradiation prediction model.
[0114] In this embodiment, the display 103 may include a display screen (display panel). In some implementations, the display panel may be configured in the form of a liquid crystal display (LCD), an organic light-emitting diode (OLED), etc. In addition, the display 103 may also be a touch panel (touch screen, touch screen), and the touch panel may include a display screen and a touch-sensitive surface. When the touch-sensitive surface detects a touch operation on or near it, it is transmitted to the processor 102 to determine the type of touch event, and then the processor 102 provides a corresponding visual output on the display device according to the type of touch event.
[0115] The above embodiments are merely illustrative of the principles and effects of the present application and are not intended to limit the present application. Anyone familiar with the technology may modify or change the above embodiments without violating the spirit and scope of the present application. Therefore, all equivalent modifications or changes made by a person of ordinary skill in the art without departing from the spirit and technical ideas disclosed in the present application shall still be covered by the claims of the present application.
Claims
1. A training method for a minute-level inclined surface irradiation prediction model, characterized in that: include: Acquire training data, where the training data is minute-level data and includes input parameters and their corresponding measured data; Using the training data to train a minute-level inclined surface irradiation prediction model; The linearity and accuracy of the prediction model are obtained.
2. The method according to claim 1, characterized in that The training data includes first training data and second training data, the input parameter of the first training data includes the irradiance intensity of the inclined surface at a first reference angle, and the input parameter of the second training data includes the irradiance intensity of the inclined surface at a second reference angle; Using the training data to train the minute-level inclined surface irradiation prediction model includes: using the first training data to train the minute-level inclined surface irradiation prediction model to obtain a first prediction model, and using the second training data to train the minute-level inclined surface irradiation prediction model to obtain a second prediction model; Acquiring the linearity and accuracy of the prediction model includes: acquiring the linearity and accuracy of the first prediction model and the second prediction model.
3. The method according to claim 2, characterized in that Also includes: Contribution of input parameters to a prediction result is obtained, where the input parameters include input parameters of the first training data and input parameters of the second training data.
4. The method according to claim 1, characterized in that: The training data includes third training data and fourth training data, the input parameter of the third training data is data of a first time scale, and the input parameter of the fourth training data is data of a second time scale; Using the training data to train the minute-level inclined surface irradiation prediction model includes: using the third training data to train the minute-level inclined surface irradiation prediction model to obtain a third prediction model, and using the fourth training data to train the minute-level inclined surface irradiation prediction model to obtain a fourth prediction model; Acquiring the linearity and accuracy of the prediction model includes: acquiring the linearity and accuracy of the third prediction model and the fourth prediction model.
5. The method according to claim 4, characterized in that Also includes: Taking an average value of the one-minute scale data at the first time scale to obtain the input parameters of the third training data, and taking an average value of the one-minute scale data at the second time scale to obtain the input parameters of the fourth training data; or The one-minute scale data is sampled at intervals of the first time scale to obtain input parameters of the third training data, and the one-minute scale data is sampled at intervals of the second time scale to obtain input parameters of the fourth training data.
6. The method according to claim 1, characterized in that The training data includes fifth training data and sixth training data, and input parameters of the fifth training data and the sixth training data correspond to different weather types; Using the training data to train the minute-level inclined surface irradiation prediction model includes: using the fifth training data to train the minute-level inclined surface irradiation prediction model to obtain a fifth prediction model, and using the sixth training data to train the minute-level inclined surface irradiation prediction model to obtain a sixth prediction model; Acquiring the linearity and accuracy of the prediction model includes: acquiring the linearity and accuracy of the fifth prediction model and the sixth prediction model.
7. The method according to claim 1, characterized in that The training data includes seventh training data, and the seventh training data is data of the first region. The method further includes: Using the seventh training data to train the minute-level inclined surface irradiation prediction model to obtain a seventh prediction model; Using the seventh prediction model to predict the target angle inclined surface irradiance intensity of the first region and the second region respectively, so as to obtain a prediction result of the first region and a prediction result of the second region; The generalization ability of the seventh prediction model is determined according to the prediction result of the first region and the prediction result of the second region.
8. The method according to claim 7, characterized in that The seventh training data is data of the first region in the first weather type, and using the seventh prediction model to predict the target angle inclined surface irradiance of the first region and the second region respectively to obtain the prediction result of the first region and the prediction result of the second region includes: Using the seventh prediction model, predicting the target angle inclined surface radiation intensity of the first region in the first weather type, to obtain a prediction result for the first region; The seventh prediction model is used to predict the radiation intensity of the target angle inclined surface of the second area in the first weather type to obtain a prediction result for the second area.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.
10. An electronic device, characterized in that: The electronic device comprises: A memory storing a computer program; A processor is communicatively connected to the memory, and executes the method according to any one of claims 1 to 8 when calling the computer program.
Citation Information
Patent Citations
Photovoltaic power prediction method, system and device and storage medium
CN114626622A
Photovoltaic power prediction method, system and device based on LSTM combined prediction framework, and storage medium
CN116029442A
