CN114282737A - A method, device and electronic device for predicting short-term solar radiation intensity - Google Patents

Abstract

The application discloses a method and a device for predicting short-term solar irradiation intensity and electronic equipment, wherein the method and the device are used for acquiring meteorological data of a target area; and processing the meteorological data by using a hybrid prediction model which is formed by mixing and constructing a GCN model and a cavity convolution model to obtain the short-term solar radiation intensity of the target area in a future period. According to the method, the meteorological data of the modular photovoltaic power station and the surrounding area of the modular photovoltaic power station are processed in two dimensions of time and space, the short-term solar irradiation intensity is predicted by fusing the two parts, and a control basis is provided for stable operation of a power system.

Description

A method, device and electronic device for predicting short-term solar radiation intensity

technical field

The present application relates to the field of new energy technologies, and more particularly, to a method, device and electronic device for predicting short-term solar radiation intensity.

Background technique

In recent years, with the shortage of traditional resources and the need for environmental protection, the demand for renewable energy sources (RESs) has increased dramatically. Among all renewable energy sources, solar energy, as the most typical one, has attracted widespread attention due to its abundant resources and almost ubiquitous availability. At the same time, it has many advantages compared to other forms of power generation such as hydropower, so the scale of photovoltaic power generation has grown rapidly. In recent years, the global photovoltaic market has continued to grow. In 2018, the global installed capacity was 99.8 GW, of which the domestic new installed capacity was about 45 GW, and will continue to grow rapidly in the next few years.

However, due to its dependence on immediate meteorological factors such as atmospheric temperature, total cloud cover, and humidity, photovoltaic power generation suffers from problems such as randomness, volatility, and intermittency. These uncertainties degrade the real-time control performance and endanger the stable operation of the power system. In order to solve the above problems, it is necessary to predict the short-term irradiance intensity of the sun, so as to predict the power generated by the photovoltaic power generation system in the short term, and provide a control basis for the stable operation of the power system. Prediction of short-term irradiance intensity.

SUMMARY OF THE INVENTION

In view of this, the present application provides a short-term solar irradiance prediction method, device and electronic device for predicting the sun's short-term irradiance intensity.

In order to achieve the above purpose, the proposed scheme is as follows:

A prediction method of short-term solar radiation intensity, applied to electronic equipment, the prediction method comprises the steps:

Obtain the meteorological data of the target area;

The meteorological data is processed based on a pre-built hybrid prediction model, and the short-term solar radiation intensity of the target area in the future period is obtained. The hybrid prediction model is constructed by mixing the GCN model and the hole convolution model.

Optionally, also include steps:

constructing the GCN model based on the meteorological factor data of the surrounding area of the combined photovoltaic power station in space;

The dilated convolution model is constructed by temporally modeling solar irradiance time series data for each of the surrounding regions.

Optionally, the construction of the GCN model based on the meteorological factor data of the surrounding area of the combined photovoltaic power station in space includes the steps of:

First, the distribution and structured meteorological time series of the photovoltaic power station are defined as an undirected graph G _t =(V _t ,ε,W), where V _t is a finite vertex, and each vertex represents the weather at time t factor, ε is the set of edges, and W is the adjacency matrix of the graph;

Then, the graph convolution network is calculated by using graph Fourier transform in the spectral domain, and the CNN is extended to the graph domain to obtain the GCN model.

Optionally, the construction of the hollow convolution model by modeling the solar irradiance time series data of each of the surrounding areas in time includes the steps of:

的输出，其中A和B分别通过sigmoid函数和融合运算，最后进行Hadamard乘积来获得结果；First, perform two dilated convolution operations respectively to obtain two equal-sized

The output of , where A and B pass the sigmoid function and fusion operation respectively, and finally perform Hadamard product to obtain the result;

得到所述空洞卷积模型，其中，σ是sigmoid函数，

是元素级的Hadamard乘积运算。Then, the temporal convolution can be defined as

to obtain the atrous convolution model, where σ is the sigmoid function,

is an element-wise Hadamard product operation.

A short-term solar radiation intensity prediction device, applied to electronic equipment, the prediction device comprising:

a data acquisition module, configured to acquire meteorological data of the target area;

The forecast execution module is configured to process the meteorological data based on a pre-built mixed forecast model to obtain the short-term solar radiation intensity of the target area in the future period, the mixed forecast model is mixed by the GCN model and the hole convolution model constructed.

Optionally, also include:

a first building module configured to build the GCN model based on meteorological factor data of the surrounding area of the combined photovoltaic power station in space;

A second building block configured to build the hole convolution model by temporally modeling solar irradiance time series data for each of the surrounding regions.

Optionally, the first building module includes:

The first building unit is used to define the distribution and structured meteorological time series of the photovoltaic power station as an undirected graph G _t =(V _t ,ε,W), where V _t is a finite vertex, each vertex Meteorological factor representing time t, ε is the set of edges, W is the adjacency matrix of the graph;

The second construction unit is used to calculate the graph convolution network by using graph Fourier transform in the spectral domain, and generalize the CNN to the graph domain to obtain the GCN model.

Optionally, the second building module includes:

的输出，其中A和B分别通过sigmoid函数和融合运算，最后进行Hadamard乘积来获得结果；The third building unit is used to perform two dilated convolution operations respectively to obtain two equal-sized

The output of , where A and B pass the sigmoid function and fusion operation respectively, and finally perform Hadamard product to obtain the result;

得到所述空洞卷积模型，其中，σ是sigmoid函数，

是元素级的Hadamard乘积运算。The fourth building block, used to define the temporal convolution as

to obtain the atrous convolution model, where σ is the sigmoid function,

is an element-wise Hadamard product operation.

An electronic device is provided with the above-mentioned prediction device.

An electronic device, characterized in that, at least one processor and a memory connected to the processor are provided, wherein:

the memory is used to store computer programs or instructions;

The processor is for the computer program or instructions to cause the electronic device to implement the prediction method as described above.

As can be seen from the above technical solutions, the present application discloses a method, device and electronic equipment for predicting short-term solar radiation intensity, the method and device are specifically to obtain meteorological data of a target area; The hybrid forecasting model constructed by mixing the models processes the meteorological data to obtain the short-term solar radiation intensity of the target area in the future. The present application provides a control basis for the stable operation of the power system by processing the meteorological data of the photovoltaic power station and its surrounding areas in the two dimensions of time and space, and by integrating the two parts to predict the short-term solar radiation intensity.

Description of drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only These are some embodiments of the present application. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings without any creative effort.

1 is a flowchart of a method for predicting short-term solar radiation intensity according to an embodiment of the application;

2 is a flowchart of another method for predicting short-term solar radiation intensity according to an embodiment of the application;

FIG. 3 is a single-sided amplitude spectrum diagram of a power Consumption series according to an embodiment of the application;

4 is a block diagram of a device for predicting short-term solar radiation intensity according to an embodiment of the application;

5 is a block diagram of another apparatus for predicting short-term solar radiation intensity according to an embodiment of the present application;

FIG. 6 is a block diagram of an electronic device according to an embodiment of the present application.

Detailed ways

The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, but not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.

In this application, while using the meteorological factor data of the surrounding area of the GCN combined photovoltaic power station, the hollow convolution is used to connect the time series data of the solar irradiance change in each area, so as to obtain a hybrid model with superior performance, which can be used for short-term solar Irradiance prediction. The short-term solar irradiance prediction method based on GCN and hole convolution hybrid model mainly includes modeling the solar irradiance data of photovoltaic power plants and their surrounding areas in two dimensions of time and space, and predicting the solar irradiance by fusing the two parts. Irradiance, see the description of the following embodiments for specific content.

Example 1

FIG. 1 is a flowchart of a method for predicting short-term solar radiation intensity according to an embodiment of the present application.

As shown in FIG. 1 , the prediction method provided in this embodiment is applied to an electronic device to predict the short-term solar radiation intensity of the target area where the photovoltaic power station is located. The electronic device can be understood as a device with data calculation and information processing capabilities. A computer or server, the forecasting method includes the following steps:

S1. Obtain the meteorological data of the target area.

The target area here refers to the area that needs to predict the short-term solar radiation intensity, there is a corresponding photovoltaic power station in this area, and the meteorological data refers to the temperature, humidity, wind, wind, etc. Data such as cloud cover, date, latitude and longitude.

S2, using meteorological data to predict the short-term solar radiation intensity of the target area.

The present application processes the above-mentioned meteorological data by using a hybrid prediction model composed of a GCN model and a hollow convolution model to obtain the short-term solar radiation intensity of the target area.

It can be seen from the above technical solutions that this embodiment provides a method for predicting short-term solar radiation intensity. The method is applied to electronic equipment, specifically to obtain meteorological data of a target area; using a mixture of GCN model and hole convolution model The constructed hybrid prediction model processes the meteorological data to obtain the short-term solar radiation intensity of the target area in the future period. The present application provides a control basis for the stable operation of the power system by processing the meteorological data of the photovoltaic power station and its surrounding areas in the two dimensions of time and space, and by integrating the two parts to predict the short-term solar radiation intensity.

In a specific embodiment of the present application, the following steps are further included, as shown in FIG. 2 .

S01 , constructing a GCN model based on the regional meteorological factor data around the combined photovoltaic power station in space.

First, the distribution and structured meteorological time series of photovoltaic power plants are defined as an undirected graph G _t =(V _t ,ε,W), where V _t is a finite vertex, each vertex represents the meteorological factor at time t, ε is the set of edges and W is the adjacency matrix of the graph. The meteorological factor data forecast on the graph can be written as Eq(11):

Then, a graph convolutional network (GCN) is generalized to the graph domain by computing it with graph Fourier transform in the spectral domain. It can be written as Eq(12),

g _θ (·)*x=Ug _θ (Λ)U ^T x (12)

where U is the eigenvector matrix of the normalized graph Laplacian L=I _N -D ^-1 / ₂ AD ^-1 / ₂ =UΛU, the diagonal matrix Λ of its eigenvalues, U ^T x is the graph of x Fourier transform.

Data is organized into graphs according to the distribution of photovoltaic power plants, which can effectively utilize spatial information. At the same time, we directly apply graph convolution operations on structured data to extract deep patterns and features in the spatial domain. However, matrix multiplication of eigenvectors in Eq(2) can be computationally expensive for large graphs, which can be overcome using Chebyshev polynomial approximations and hierarchical linear formulations.

In order to reduce the time complexity, the filter is approximated by a truncated extension of the Chebyshev polynomial _Tk (x) to degree ^Kth . Then we can rewrite the graph convolution as Eq(13),

可以通过(UΛU^T)^k＝UΛ^kU^T来计算。通过多项式近似计算K-局部卷积，可以降低Eq(13)的时间复杂度。in,

It can be calculated by ( ^UΛUT ) ^k = ^{UΛkUT .} The time complexity of Eq(13) can be reduced by computing K-local convolutions by polynomial approximation.

By restricting K=1, the graph convolution function can become linear on the graph Laplacian. Furthermore, since the neural network can adapt to scale changes, we can approximate λ _max =2. It can be written as Eq(14),

和

它可以写成Eq(15)，where θ ₀ and θ ₁ are the two shared filter parameters. In order to reduce the occurrence of overfitting and numerical manipulation, θ ₀ and θ ₁ can be exchanged for a parameter θ, let θ = θ ₀ =-;

and

It can be written as Eq(15),

S02, constructing a hollow convolution model by modeling the solar irradiance time series data of each target area in time.

We all know that RNN-like models have always been time-consuming and cannot cope with changing data due to complex gating mechanisms, while CNN has the advantage of fast training and parallel training processes can be achieved by stacking convolutional layers. Therefore, we apply a gated linear unit with a _Kt -width kernel and a 1D dilated convolution on the temporal dimension of the input solar irradiance data of PV plants and surrounding areas.

Atrous convolution can expand the receptive field. In order to increase the receptive field and reduce the amount of calculation in the deep network, downsampling is always performed. Although the receptive field can be increased, the spatial resolution is reduced. In order not to lose resolution and still expand the receptive field, atrous convolution can be used. This is useful in prediction tasks. On the one hand, if the sensory field is larger, it is possible to learn solar irradiance data in a wider range of time and space, and on the other hand, if the resolution is higher, the prediction time point can be precisely located.

有C_i个通道，核大小为

At the same time, it can also capture multi-scale context information: the hole convolution has a parameter that can be set, and the specific meaning is to fill multiple 0s in the convolution kernel. Therefore, when different scales are set, the receptive field will be different, that is, the acquisition of multi-scale solar irradiance data information. The input to this model can be viewed as a sequence of solar irradiance data of length M

There are C _i channels and the kernel size is

的输出。它可以写成Eq(21)，First, after entering the model, perform two dilated convolution operations respectively to obtain two equal-sized

Output. It can be written as Eq(21),

是核，x_t是序列x的第t个值。where d is the dilution parameter controlling the skip distance,

is the kernel, and x _t is the t-th value of the sequence x.

Then, pass A and B through the sigmoid function and fusion operation respectively, and finally perform the Hadamard product to obtain the result. The sigmoid function helps filter inputs that help to discover dynamically changing patterns in the data, while the nonlinear gate captures general information about the data. Finally, temporal convolution can be defined as

是元素级的Hadamard乘积运算。where σ is the sigmoid function;

is an element-wise Hadamard product operation.

After the solar irradiance data is modeled separately in the space-time domain, the meteorological data can be calculated by the hybrid model.

The advantages and effects of the present application compared with the prior art are:

The present application is a method for short-term solar irradiance prediction based on a hybrid model of GCN and atrous convolution, which uses joint spatiotemporal features to predict the solar irradiance in the missing area for the next five days. Since the solar irradiance can only be collected during the day, the forecasted time period is 8:00-18:00 every day, and the forecast results for the next five days are shown in Table 1. Among them, No-Spa is the method to remove the GCN model. The results of No-Spa are much more accurate than GRU and CNN, indicating that GCN model can extract more global spatial information. The prediction results of this method (STM) are better than that of No-Spa, which proves that capturing spatial and temporal features is very important and effective for predicting solar irradiance.

Table 1 Prediction results of solar irradiance

The visualization of the prediction results is shown in Figure 3. It can be seen from the figure that the prediction curve of this method (STM) is closest to the true value curve of solar irradiation, while the curves of other methods are in the region of fluctuation and oscillation and the true value curve Big difference.

Embodiment 2

FIG. 4 is a block diagram of an apparatus for predicting short-term solar radiation intensity according to an embodiment of the present application.

As shown in FIG. 4 , the prediction device provided in this embodiment is applied to an electronic device to predict the short-term solar radiation intensity of the target area where the photovoltaic power station is located. The electronic device can be understood as a device with data calculation and information processing capabilities. A computer or server, the prediction device can be regarded as the computer or server itself, or a hardware module of Li Jiwei's computer or server, the prediction device includes a data acquisition module 10 and a prediction execution module 20 .

The data acquisition module is used to acquire the meteorological data of the target area.

The target area here refers to the area that needs to predict the short-term solar radiation intensity, there is a corresponding photovoltaic power station in this area, and the meteorological data refers to the temperature, humidity, wind, wind, etc. Data such as cloud cover, date, latitude and longitude.

The forecast execution module is used for forecasting the short-term solar irradiance intensity of the target area by using the meteorological data.

The present application processes the above-mentioned meteorological data by using a hybrid prediction model composed of a GCN model and a hollow convolution model to obtain the short-term solar radiation intensity of the target area.

It can be seen from the above technical solutions that this embodiment provides a short-term solar radiation intensity prediction device, which is applied to electronic equipment, specifically to obtain meteorological data of a target area; using a mixture of a GCN model and a hole convolution model The constructed hybrid prediction model processes the meteorological data to obtain the short-term solar radiation intensity of the target area in the future period. The present application provides a control basis for the stable operation of the power system by processing the meteorological data of the photovoltaic power station and its surrounding areas in the two dimensions of time and space, and by integrating the two parts to predict the short-term solar radiation intensity.

In a specific embodiment of the present application, a first building module 30 and a second building module 40 are also included, as shown in FIG. 5 .

The first building module is used to build a GCN model based on the regional meteorological factor data around the combined photovoltaic power station in space. The module includes a first building unit and a second building unit.

The first building unit is used to define the distribution and structured meteorological time series of photovoltaic power plants as an undirected graph G _t =(V _t ,ε,W), where V _t is a finite vertex, and each vertex represents time t The meteorological factor of , ε is the set of edges, and W is the adjacency matrix of the graph. The meteorological factor data forecast on the graph can be written as Eq(11):

The second building unit is used to generalize the CNN to the graph domain by computing the graph convolutional network (GCN) with graph Fourier transform in the spectral domain. It can be written as Eq(12),

g _θ (·)*x=Ug _θ (Λ)U ^T x (12)

where U is the eigenvector matrix of the normalized graph Laplacian L=I _N -D ^-1/ ₂ AD ^-1/ ₂ =UΛU, the diagonal matrix Λ of its eigenvalues, U ^T x is the graph of x Fourier transform.

Data is organized into graphs according to the distribution of photovoltaic power plants, which can effectively utilize spatial information. At the same time, we directly apply graph convolution operations on structured data to extract deep patterns and features in the spatial domain. However, matrix multiplication of eigenvectors in Eq(2) can be computationally expensive for large graphs, which can be overcome using Chebyshev polynomial approximations and hierarchical linear formulations.

In order to reduce the time complexity, the filter is approximated by a truncated extension of the Chebyshev polynomial _Tk (x) to degree ^Kth . Then we can rewrite the graph convolution as Eq(13),

可以通过(UΛU^T)^k＝UΛ^kU^T来计算。通过多项式近似计算K-局部卷积，可以降低Eq(13)的时间复杂度。in,

It can be calculated by ( ^UΛUT ) ^k = ^{UΛkUT .} The time complexity of Eq(13) can be reduced by computing K-local convolutions by polynomial approximation.

By restricting K=1, the graph convolution function can become linear on the graph Laplacian. Furthermore, since the neural network can adapt to scale changes, we can approximate λ _max =2. It can be written as Eq(14),

和

它可以写成Eq(15)，where θ ₀ and θ ₁ are the two shared filter parameters. In order to reduce the occurrence of overfitting and numerical manipulation, θ ₀ and θ ₁ can be exchanged for a parameter θ, let θ = θ ₀ =-;

and

It can be written as Eq(15),

The second building block is used to build an atrous convolution model by temporally modeling solar irradiance time series data for each target area.

We all know that RNN-like models have always been time-consuming and cannot cope with changing data due to complex gating mechanisms, while CNN has the advantage of fast training and parallel training processes can be achieved by stacking convolutional layers. Therefore, we apply a gated linear unit with a _Kt -width kernel and a 1D dilated convolution on the temporal dimension of the input solar irradiance data of PV plants and surrounding areas.

Atrous convolution can expand the receptive field. In order to increase the receptive field and reduce the amount of calculation in the deep network, downsampling is always performed. Although the receptive field can be increased, the spatial resolution is reduced. In order not to lose resolution and still expand the receptive field, atrous convolution can be used. This is useful in prediction tasks. On the one hand, if the sensory field is larger, it is possible to learn solar irradiance data in a wider range of time and space, and on the other hand, if the resolution is higher, the prediction time point can be precisely located.

有C_i个通道，核大小为

At the same time, it can also capture multi-scale context information: the hole convolution has a parameter that can be set, and the specific meaning is to fill multiple 0s in the convolution kernel. Therefore, when different scales are set, the receptive field will be different, that is, the acquisition of multi-scale solar irradiance data information. The input of this model can be regarded as the solar irradiance sequence data of length M

There are C _i channels and the kernel size is

的输出。它可以写成Eq(21)，The module includes a third building unit and a fourth building unit. The third building unit is used to perform two dilated convolution operations after entering the model to obtain two

Output. It can be written as Eq(21),

是核，x_t是序列x的第t个值。where d is the dilution parameter controlling the skip distance,

is the kernel, and x _t is the t-th value of the sequence x.

The fourth building unit is used to pass A and B through the sigmoid function and fusion operation respectively, and finally perform the Hadamard product to obtain the result. The sigmoid function helps filter inputs that help to discover dynamically changing patterns in the data, while the nonlinear gate captures general information about the data. Finally, temporal convolution can be defined as

是元素级的Hadamard乘积运算。where σ is the sigmoid function;

is an element-wise Hadamard product operation.

After the solar irradiance data is modeled separately in the space-time domain, the meteorological data can be calculated by the hybrid model.

The advantages and effects of the present application compared with the prior art are:

The present application is a method for short-term solar irradiance prediction based on a hybrid model of GCN and atrous convolution, which uses joint spatiotemporal features to predict the solar irradiance in the missing area for the next five days. Since the solar irradiance can only be collected during the day, the forecasted time period is 8:00-18:00 every day, and the forecast results for the next five days are shown in Table 1. Among them, No-Spa is the method to remove the GCN model. The results of No-Spa are much more accurate than GRU and CNN, indicating that GCN model can extract more global spatial information. The prediction results of this method (STM) are better than that of No-Spa, which proves that capturing spatial and temporal features is very important and effective for predicting solar irradiance.

Table 2 Prediction results of solar irradiance

The visualization of the prediction results is shown in Figure 3. It can be seen from the figure that the prediction curve of this method (STM) is closest to the true value curve of solar radiation, while the curves of other methods are different from the true value curve in the areas where fluctuations and oscillations occur. larger.

Embodiment 3

This embodiment provides an electronic device, which can be understood as a computer or server having data computing and information processing. The electronic device is provided with the device for predicting the short-term solar radiation intensity of the above-mentioned embodiment. The device is used to obtain the meteorological data of the target area; the meteorological data is processed by a hybrid prediction model constructed by mixing the GCN model and the hollow convolution model to obtain the short-term solar radiation intensity of the target area in the future period. The present application provides a control basis for the stable operation of the power system by processing the meteorological data of the photovoltaic power station and its surrounding areas in the two dimensions of time and space, and by integrating the two parts to predict the short-term solar radiation intensity.

Embodiment 4

FIG. 6 is a block diagram of an electronic device according to an embodiment of the present application.

As shown in FIG. 6 , the electronic device provided in this embodiment can be understood as a computer or server having data computing and information processing. The electronic device includes at least a processor 101 and a memory 102 , which are connected through a data bus 103 .

The memory is used to store corresponding computer programs or instructions, and the processor is used to execute the above computer programs or instructions, so that the electronic device implements the method for predicting the short-term solar irradiance intensity described in the embodiment. The method is specifically used to obtain the meteorological data of the target area; the meteorological data is processed by a hybrid prediction model constructed by mixing the GCN model and the hollow convolution model to obtain the short-term solar radiation intensity of the target area in the future period. The present application provides a control basis for the stable operation of the power system by processing the meteorological data of the photovoltaic power station and its surrounding areas in the two dimensions of time and space, and by integrating the two parts to predict the short-term solar radiation intensity.

The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same and similar parts between the various embodiments may be referred to each other.

It should be understood by those skilled in the art that the embodiments of the embodiments of the present invention may be provided as a method, an apparatus, or a computer program product. Accordingly, embodiments of the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, embodiments of the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.

Embodiments of the present invention are described with reference to flowcharts and/or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the present invention. It will be understood that each process and/or block in the flowchart illustrations and/or block diagrams, and combinations of processes and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to the processor of a general purpose computer, special purpose computer, embedded processor or other programmable data processing terminal equipment to produce a machine that causes the instructions to be executed by the processor of the computer or other programmable data processing terminal equipment Means are created for implementing the functions specified in a flow or flows of the flowcharts and/or a block or blocks of the block diagrams.

These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing terminal equipment to operate in a particular manner, such that the instructions stored in the computer readable memory result in an article of manufacture comprising instruction means, the The instruction means implement the functions specified in the flow or flows of the flowcharts and/or the block or blocks of the block diagrams.

These computer program instructions can also be loaded on a computer or other programmable data processing terminal equipment, so that a series of operational steps are performed on the computer or other programmable terminal equipment to produce a computer-implemented process, thereby executing on the computer or other programmable terminal equipment The instructions executed on the above provide steps for implementing the functions specified in the flowchart or blocks and/or the block or blocks of the block diagrams.

While preferred embodiments of the embodiments of the present invention have been described, additional changes and modifications to these embodiments may be made by those skilled in the art once the basic inventive concepts are known. Therefore, the appended claims are intended to be construed to include the preferred embodiments as well as all changes and modifications that fall within the scope of the embodiments of the present invention.

Finally, it should also be noted that in this document, relational terms such as first and second are used only to distinguish one entity or operation from another, and do not necessarily require or imply these entities or there is any such actual relationship or sequence between operations. Moreover, the terms "comprising", "comprising" or any other variation thereof are intended to encompass non-exclusive inclusion such that a process, method, article or terminal device that includes a list of elements includes not only those elements, but also a non-exclusive list of elements. other elements, or also include elements inherent to such a process, method, article or terminal equipment. Without further limitation, an element defined by the phrase "comprises a..." does not preclude the presence of additional identical elements in the process, method, article, or terminal device that includes the element.

The technical solutions provided by the present invention have been described in detail above, and specific examples are used to illustrate the principles and implementations of the present invention. The descriptions of the above embodiments are only used to help understand the methods and core ideas of the present invention; At the same time, for those skilled in the art, according to the idea of the present invention, there will be changes in the specific embodiments and application scope. To sum up, the content of this specification should not be construed as a limitation of the present invention.

Claims (10)

1. A method for predicting short-term solar radiation intensity is applied to electronic equipment and is characterized by comprising the following steps:

acquiring meteorological data of a target area;

and processing the meteorological data based on a pre-constructed hybrid prediction model to obtain the short-term solar radiation intensity of the target area in a future period, wherein the hybrid prediction model is constructed by mixing a GCN model and a cavity convolution model.

2. The prediction method of claim 1, further comprising the steps of:

constructing the GCN model based on meteorological factor data of surrounding areas of the spatially combined photovoltaic power station;

the hole convolution model is constructed by modeling temporally the solar irradiance time series data for each of the surrounding regions.

3. The prediction method of claim 2, wherein said constructing the GCN model based on spatially integrating meteorological factor data for a surrounding area of a photovoltaic power plant comprises the steps of:

firstly, the distribution and the structured meteorological time series of the photovoltaic power station are defined as an undirected graph G_t＝(V_tε, W) in which V_tIs a finite vertex, each vertex represents a meteorological factor at time t, ε is a set of edges, and W is the adjacency matrix of the graph;

and then, calculating the graph convolution network by using graph Fourier transform in a spectrum domain, and popularizing the CNN into the graph domain to obtain the GCN model.

4. The prediction method of claim 2, wherein said constructing said hole convolution model by modeling temporally solar irradiance time series data for each of said surrounding regions comprises the steps of:

firstly, two expanding convolution operations are respectively carried out to obtain two products with the same size

The A and the B respectively obtain results through a sigmoid function and fusion operation, and finally Hadamard product is carried out;

then, the time convolution can be defined as

Obtaining the cavity convolution model, wherein sigma is sigmoid function,

is an element-level Hadamard product operation.

5. A device for predicting short-term solar radiation intensity is applied to electronic equipment, and is characterized by comprising:

a data acquisition module configured to acquire meteorological data of a target area;

and the prediction execution module is configured to process the meteorological data based on a pre-constructed hybrid prediction model to obtain the short-term solar radiation intensity of the target area in a future period, and the hybrid prediction model is constructed by mixing a GCN model and a void convolution model.

6. The prediction apparatus of claim 5, further comprising:

a first construction module configured to construct the GCN model based on meteorological factor data spatially combined with surrounding areas of a photovoltaic power plant;

a second construction module configured to construct the hole convolution model by modeling temporally the solar irradiance time series data for each of the surrounding regions.

7. The prediction apparatus of claim 6, wherein the first construction module comprises:

a first construction unit for defining the distribution and the structured meteorological time series of the photovoltaic power station as an undirected graph G_t＝(V_tε, W) in which V_tIs a finite vertex, each vertex represents a meteorological factor at time t, ε is a set of edges, and W is the adjacency matrix of the graph;

and the second construction unit is used for calculating the graph convolution network by using graph Fourier transform in a spectrum domain, and popularizing the CNN into the graph domain to obtain the GCN model.

8. The prediction apparatus of claim 6, wherein the second construction module comprises:

a third construction unit for performing two times of dilation convolution operations to obtain two signals with the same size

The A and the B respectively obtain results through a sigmoid function and fusion operation, and finally Hadamard product is carried out;

a fourth construction unit for defining the time convolution as

Obtaining the cavity convolution model, wherein sigma is sigmoid function,

is an element-level Hadamard product operation.

9. An electronic device, characterized in that a prediction apparatus according to any one of claims 5 to 8 is provided.

10. An electronic device, characterized in that at least one processor and a memory connected to the processor are provided, wherein:

the memory is for storing a computer program or instructions;

the processor is for the computer program or instructions to cause the electronic device to implement the prediction method of any one of claims 1 to 4.

Images