Irradiance prediction method and device, computer equipment and storage medium

Abstract

The invention relates to the technical field of photovoltaic power generation, and discloses an irradiance prediction method and device, computer equipment and a storage medium, and the method comprises the steps: carrying out the cloud displacement vector calculation of a foundation cloud picture, and obtaining the displacement vector of a cloud cluster; determining a sun center point and a plurality of first pixel points around the sun center point; establishing an adjacent matrix according to the relative position of each first pixel point and the sun center point and the displacement direction of the cloud cluster; determining the RGB value of each first pixel point according to the displacement vector; establishing a node feature matrix according to the RGB value of each first pixel point; inputting the adjacent matrix and the node feature matrix into a graph convolutional network prediction model to obtain an RGB value of a second pixel point located at the center point of the sun; and inputting the RGB value of the second pixel point into the irradiance prediction model to obtain an irradiance prediction result at the target moment. The cloud picture features are described through the picture structure data, and the rich feature information of the cloud layer is reflected through the RGB value, so that the prediction precision of the irradiance can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power generation, and in particular to an irradiance prediction method, device, computer equipment and storage medium. Background Art

[0002] Timely and accurate prediction of photovoltaic power generation can reduce the impact of its uncertainty to a certain extent, which is of great significance to grid dispatching and photovoltaic power station operation. As the most important factor affecting the power generation of photovoltaic power stations, accurate prediction of solar irradiance can effectively improve the accuracy of photovoltaic power prediction. However, in cloudy weather, the solar irradiance received by the surface is affected by the movement of clouds and their blocking of the sun, and will show violent fluctuations on the minute time scale. There is almost no clear causal mapping relationship between the irradiance fluctuation caused by the movement of clouds and the historical irradiation data, making the irradiance prediction on the minute time scale more difficult. Using sky images to track the movement of clouds and analyze their inherent characteristics can more accurately predict their blocking of the sun, thereby achieving high-precision minute-level irradiance prediction.

[0003] At present, the commonly used irradiance prediction methods are mainly based on two categories: numerical and image. The numerical type mainly analyzes historical irradiance data and various meteorological data, and uses regression models, machine learning or deep learning methods to construct a mapping relationship between numerical values, so as to realize the prediction of irradiance; the image type mainly uses satellite cloud images, radar images or ground-based cloud images, and uses image processing technology to analyze the distribution and movement of clouds, so as to realize the prediction of solar irradiance. At present, there are studies on using ground-based cloud images for minute-level irradiance prediction, but the feature extraction method for Euclidean spatial data is usually used, which is not applicable to the irregular graph structure data composed of multiple key pixels in the cloud image. At the same time, the existing research usually simply uses linear extrapolation based on the calculation results of cloud displacement vectors to obtain the cloud domain that will affect the irradiance fluctuation in the future, which is difficult to reflect the impact of complex cloud cluster changes on the solar cover in the future. At the same time, the single cloud domain obtained by linear extrapolation also ignores the common influence of multiple cloud domains on the center of the sun, resulting in low prediction accuracy. Summary of the invention

[0004] In view of this, the present invention provides an irradiance prediction method, apparatus, computer equipment and storage medium to solve the problem of low irradiance prediction accuracy in the prior art.

[0005] In a first aspect, the present invention provides an irradiance prediction method, which includes: calculating a cloud displacement vector of a ground-based cloud image to obtain a displacement vector of a cloud cluster, wherein the displacement vector includes a displacement direction and a displacement speed of the cloud cluster; determining the center point of the sun, and a plurality of first pixel points around the center point of the sun; establishing an adjacency matrix according to the relative position of each first pixel point to the center point of the sun, and the displacement direction of the cloud cluster; determining the RGB value of each first pixel point according to the displacement vector of the cloud cluster; establishing a node feature matrix according to the RGB value of each first pixel point; inputting the adjacency matrix and the node feature matrix as graph structure data into a pre-trained graph convolutional network prediction model to obtain the RGB value of a second pixel point located at the center point of the sun; inputting the RGB value of the second pixel point as an input feature into a pre-trained irradiance prediction model to obtain an irradiance prediction result at a target moment.

[0006] The method provided in the embodiment of the present invention focuses on the cloud area covering the center of the sun which has a direct impact on the irradiance. On the basis of calculating the cloud displacement vector, the ground-based cloud map data is used to construct an adjacency matrix of a graph structure according to the relative position of each first pixel point and the center of the sun, and the displacement direction of the cloud group. The information of the cloud situation around the sun and whether the adjacent cloud range may cover the sun is creatively converted into graph structure data. A graph convolutional neural network is used to predict the RGB values ​​of the pixel points in the area that will cover the sun in the future, and the irradiance prediction result at the target moment is obtained by using the RGB values ​​of the pixels that will be located at the center of the sun at that moment in the future. Compared with the traditional irradiance prediction method, the method proposed in the present invention adopts an improved graph structure based on cloud displacement direction analysis to describe the characteristics of cloud maps, which fully considers the spatiotemporal correlation between the cloud conditions in the solar neighborhood and the solar center point; the graph structure is used to describe the characteristics of cloud maps, and a unique feature matrix can be constructed for the irregular structure formed by key nodes, which is conducive to improving the defect that traditional feature extraction methods can only extract features for Euclidean space data and are not suitable for feature extraction of key pixel points in cloud maps that affect irradiance fluctuations; existing methods often use linear extrapolation to simply describe complex cloud movement, and ignore the common influence of surrounding clouds on the solar center point, while using graph convolutional neural networks to predict pixel RGB values ​​can better integrate the common influence of multi-point cloud domains on the solar center point, thereby describing the complex cloud deformation, improving the accuracy of RGB feature prediction, and RGB values ​​can directly reflect the rich feature information of the cloud layer, which helps to improve the prediction accuracy of irradiance and provide data support for the optimized operation and maintenance of photovoltaic power stations.

[0007] In an optional embodiment, the step of determining multiple first pixel points around the center point of the sun includes: in multiple directions centered on the center point of the sun, respectively determining the first pixel points at a preset distance from the center point of the sun, and the preset distance is determined based on the displacement speed of the cloud and the unit time length.

[0008] In an optional embodiment, an adjacency matrix is ​​established based on the relative position of each first pixel point and the center of the sun, and the displacement direction of the cloud, including: respectively calculating the angle between the direction of each first pixel point pointing to the center of the sun and the displacement direction of the cloud; determining the correlation between each first pixel point and the target event based on the angle, the target event being that the first pixel point covers the center of the sun at the target time; and establishing the adjacency matrix based on the correlation.

[0009] In an optional embodiment, the relevance of each first pixel point to the occurrence of the target event is determined based on the angle, including: if the angle between the direction of the first pixel point pointing to the center of the sun and the displacement direction of the cloud is less than a correlation threshold, it is determined that the first pixel point is related to the target event; if the angle between the direction of the first pixel point pointing to the center of the sun and the displacement direction of the cloud is greater than the correlation threshold, it is determined that the first pixel point is not related to the target event.

[0010] In an optional implementation, establishing a node feature matrix according to the RGB value of each first pixel point includes: establishing a node feature matrix according to the R value, G value, and B value of each first pixel point.

[0011] In an optional embodiment, the input features also include relative humidity at the current moment, aerosol optical depth of 500 nm, and historical irradiance data.

[0012] In a second aspect, the present invention provides an irradiance prediction device, which includes: a displacement vector calculation device, which is used to calculate the cloud displacement vector of the ground-based cloud map to obtain the displacement vector of the cloud group, and the displacement vector includes the displacement direction and displacement speed of the cloud group; a solar center point determination module, which is used to determine the solar center point and multiple first pixel points around the solar center point; an adjacency matrix establishment module, which is used to establish an adjacency matrix according to the relative position of each first pixel point and the solar center point, and the displacement direction of the cloud group; a first RGB value determination module, which is used to determine the RGB value of each first pixel point according to the displacement vector of the cloud group; a node feature matrix calculation module, which is used to establish a node feature matrix according to the RGB value of each first pixel point; a second RGB value determination module, which is used to input the adjacency matrix and the node feature matrix as graph structure data into a pre-trained graph convolutional network prediction model to obtain the RGB value of the second pixel point located at the solar center point; an irradiance prediction module, which is used to input the RGB value of the second pixel point as an input feature into a pre-trained irradiance prediction model to obtain the irradiance prediction result at the target time.

[0013] In a third aspect, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor are communicatively connected to each other, computer instructions are stored in the memory, and the processor executes the irradiance prediction method of the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0014] In a fourth aspect, the present invention provides a computer-readable storage medium, characterized in that computer instructions are stored on the computer-readable storage medium, and the computer instructions are used to enable a computer to execute the irradiance prediction method of the above-mentioned first aspect or any corresponding embodiment thereof.

[0015] In a fifth aspect, the present invention provides a computer program product, comprising computer instructions for causing a computer to execute the irradiance method of the first aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 is a schematic flow chart of an irradiance prediction method according to an embodiment of the present invention;

[0018] Figure 2 is a structural block diagram of an irradiance prediction device according to an embodiment of the present invention;

[0019] Figure 3 It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0021] According to an embodiment of the present invention, an embodiment of an irradiance prediction method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0022] In this embodiment, a method for calculating the interaction between particles is provided. Figure 1 is a flow chart of an irradiance prediction method according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0023] Step S101, calculate the cloud displacement vector of the ground-based cloud image to obtain the displacement vector of the cloud cluster, where the displacement vector includes the displacement direction and displacement speed of the cloud cluster.

[0024] In an optional embodiment, ground-based cloud images at at least two moments are continuously acquired, and the displacement vector of the cloud cluster is calculated based on the continuously acquired ground-based cloud images.

[0025] In an optional embodiment, the cloud displacement vector may be calculated using an optical flow method.

[0026] Using the optical flow method to calculate the cloud displacement vector requires three assumptions:

[0027] (1) The grayscale values ​​of corresponding pixels in adjacent images remain unchanged;

[0028] (2) The relative motion of corresponding objects in adjacent images is small;

[0029] (3) The pixels in the neighborhood have the same displacement. Based on assumption (1), we can get equation (1):

[0030] f(x,y,t)=f(x+dx,y+dy,t+dt) (1)

[0031] Where: f(x,y,t) represents the grayscale function of the image, (x,y) represents the coordinates of the pixel on the image, and t represents time. Taylor decomposition of the right side of the equation (1) yields:

[0032]

[0033] According to assumption (2), the remainder R1(x, y, t) in equation (2) is ignored, and we can obtain:

[0034]

[0035] make Formula (3) is expressed as:

[0036]

[0037] According to assumption (3), if the neighborhood contains n pixels, the following n (n>2) equations can be obtained:

[0038]

[0039] Among them: the footnotes 1, 2, ..., n represent the sequence numbers of the pixels in the neighborhood.

[0040] Solving the above linear equations using the least squares method yields:

[0041]

[0042] Finally, the average value of all pixel motion vectors is taken as the overall motion vector of the image, which includes the cloud displacement direction d and cloud displacement speed v.

[0043] Step S102, determining the sun center point and a plurality of first pixel points around the sun center point.

[0044] In an optional embodiment, the coordinates of the center point of the sun are obtained according to the solar zenith angle, the solar azimuth angle and the image radius information, as shown in equations (7) and (8):

[0045] X sun =r+rcos(θ s )cos(φ s ) (7)

[0046] Y sun =r-rcos(θ s )sin(μ s ) (8)

[0047] Among them, θs is the solar zenith angle, μs is the solar azimuth angle, and r is the image radius, which is the radius of the circular sky area in the sky image photo, in pixels.

[0048] Step S103, establishing an adjacency matrix according to the relative position of each first pixel point and the center point of the sun, and the displacement direction of the cloud cluster.

[0049] In the embodiment of the present invention, the adjacency matrix established according to the relative position of the first pixel point and the center point of the sun and the displacement direction of the cloud group can represent whether the cloud group will block the sun in the future.

[0050] Step S104, determining the RGB value of each first pixel point according to the displacement vector of the cloud.

[0051] Step S105: establishing a node feature matrix according to the RGB value of each first pixel.

[0052] In an optional embodiment, the RGB value of each first pixel at a certain moment in the future is determined according to the displacement direction and displacement vector of the cloud.

[0053] In an optional embodiment, if it is predicted that the first pixel will be blue sky at a certain moment in the future based on the displacement direction and displacement vector of the cloud, the RGB value of the first pixel will take the RGB value of blue; if it is predicted that there will be clouds at the first pixel at a certain moment in the future based on the displacement direction and displacement vector of the cloud, the RGB value of the first pixel will take the RGB value of the cloud pixel.

[0054] Step S106, input the adjacency matrix and the node feature matrix as graph structure data into a pre-trained graph convolutional network prediction model to obtain the RGB value of the second pixel point located at the center of the sun.

[0055] In an optional embodiment, the graph convolutional network prediction model takes the graph structure data including the adjacency matrix and the node feature matrix as input, normalizes the RGB data of each pixel in the node feature matrix in the graph structure data, and takes the RGB value of the pixel located at the center of the sun at the corresponding moment as output. If there are clouds covering the sun area, the RGB value of the cloud pixel is taken. If it is a blue sky, that is, the sun is not covered, the corresponding output RGB value takes the blue RGB value, that is, (0,0,255).

[0056] Step S107: input the RGB value of the second pixel as an input feature into a pre-trained irradiance prediction model to obtain an irradiance prediction result at the target time.

[0057] In an optional embodiment, graph structure data is generated according to the RGB value of the second pixel, the graph structure data is input into a neural network to obtain a normalized irradiance prediction result, and finally the irradiance prediction result is denormalized to obtain the actual irradiance.

[0058] In an optional embodiment, the irradiance prediction model is obtained based on Long Short Term Memory (LSTM) recurrent neural network training.

[0059] The method provided in the embodiment of the present invention focuses on the cloud area covering the center of the sun which has a direct impact on the irradiance. On the basis of calculating the cloud displacement vector, the ground-based cloud map data is used to construct an adjacency matrix of a graph structure according to the relative position of each first pixel point and the center point of the sun, and the displacement direction of the cloud group. The information of the cloud situation around the sun and whether the adjacent cloud area may cover the sun is creatively converted into graph structure data. A graph convolutional neural network is used to predict the RGB values ​​of the pixels in the area that will cover the sun in the future, and the irradiance prediction result at the target moment is obtained by using the RGB values ​​of the pixels that will be located at the center point of the sun at that moment in the future. Compared with the traditional irradiance prediction method, the method proposed in the present invention adopts an improved graph structure based on cloud displacement direction analysis to describe the characteristics of cloud maps, which fully considers the spatiotemporal correlation between the cloud conditions in the solar neighborhood and the solar center point; the graph structure is used to describe the characteristics of cloud maps, and a unique feature matrix can be constructed for the irregular structure formed by key nodes, which is conducive to improving the defect that traditional feature extraction methods can only extract features for Euclidean space data and are not suitable for feature extraction of key pixel points in cloud maps that affect irradiance fluctuations; existing methods often use linear extrapolation to simply describe complex cloud movement, and ignore the common influence of surrounding clouds on the solar center point, while using graph convolutional neural networks to predict pixel RGB values ​​can better integrate the common influence of multi-point cloud domains on the solar center point, thereby describing the complex cloud deformation, improving the accuracy of RGB feature prediction, and RGB values ​​can directly reflect the rich feature information of the cloud layer, which helps to improve the prediction accuracy of irradiance and provide data support for the optimized operation and maintenance of photovoltaic power stations.

[0060] In an optional embodiment, in the above step S102, in multiple directions centered on the center of the sun, the first pixel point at a preset distance from the center of the sun is determined respectively, and the preset distance is determined according to the displacement speed of the cloud group and the unit time length. After the first pixel point is obtained through this embodiment, the irradiance prediction result at the target moment is obtained in step S107, and the target moment refers to the moment in the future that is a unit time length away from the current moment.

[0061] In an optional embodiment, the multiple first pixel points around the solar center point refer to points that are centered on the solar center point and extend in eight directions of east, south, west, north, northeast, northwest, southeast, and southwest, and are at a distance of vt from the solar center point, wherein v represents the cloud displacement speed and t represents the unit time. The irradiance prediction result after the unit time is obtained in step S107. If the unit time is one minute, the irradiance prediction at the minute level can be achieved.

[0062] In an optional embodiment, the above step S103 specifically includes:

[0063] Step a1, respectively calculating the angle between the direction of each first pixel point pointing to the center of the sun and the displacement direction of the cloud.

[0064] Step a2, determining the correlation between each first pixel point and the target event according to the included angle, the target event being that the first pixel point covers the center point of the sun at the target time.

[0065] In an optional embodiment, if the angle between the direction of the first pixel pointing to the center of the sun and the displacement direction of the cloud is less than a correlation threshold, the first pixel is determined to be related to the target event; if the angle between the direction of the first pixel pointing to the center of the sun and the displacement direction of the cloud is greater than the correlation threshold, the first pixel is determined to be unrelated to the target event. Exemplarily, the correlation threshold can be 90°.

[0066] Step a3: Establish an adjacency matrix based on the correlation.

[0067] The values ​​of each item in the adjacency matrix can be:

[0068]

[0069] Among them, θ represents the angle between the direction of the first pixel point pointing to the center of the sun and the displacement direction of the cloud, θ i Represents the correlation threshold.

[0070] Based on the correlation analysis between the cloud displacement direction and the sun direction, a graph structure is constructed to describe the characteristics of the cloud map. The spatiotemporal correlation between the cloud conditions in the solar neighborhood and the solar center is fully considered. On this basis, the adjacency matrix and the node feature matrix are input as graph structure data into the pre-trained graph convolutional network prediction model to obtain the RGB value of the second pixel located at the solar center. This can better integrate the common influence of multiple cloud domains on the solar center to describe the complex cloud deformation, improve the accuracy of RGB feature prediction, and help improve the prediction accuracy of irradiance.

[0071] In an optional embodiment, the node feature matrix is ​​established based on the R value, G value, and B value of each first pixel point. For example, if eight first pixel points are determined with the sun as the center and extending in eight directions of east, south, west, north, northeast, northwest, southeast, and southwest, then the R, G, and B values ​​of the pixel points extending in the eight directions constitute the node feature matrix, and the matrix size is 8*3.

[0072] In an optional embodiment, the input features also include relative humidity at the current moment, aerosol optical depth of 500 nm, and historical irradiance data.

[0073] In addition to the situation where the sun is covered by clouds, which will significantly affect the fluctuation of irradiance, relative humidity and aerosol optical depth of 500nm are also meteorological factors that are highly correlated with irradiance changes and are used as one of the inputs of the prediction model.

[0074] The pixel RGB features at the center of the sun, relative humidity, aerosol optical depth of 500nm, and historical irradiance data of the previous two time intervals are taken as input, and the corresponding irradiance data per minute is taken as output. The long short-term memory network model is trained to obtain the irradiance prediction model.

[0075] Finally, the predicted pixel features at the center of the sun every minute in the future, relative humidity, aerosol optical depth of 500nm, and historical irradiance data of the previous two time intervals are input into the trained long short-term memory network model to obtain the minute-level irradiance prediction results with a resolution of one minute in the next ten minutes.

[0076] In this embodiment, an irradiance prediction device is also provided, which is used to implement the above-mentioned embodiments and preferred embodiments, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware of a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0077] This embodiment provides an irradiance prediction device, such as Figure 2 As shown, including:

[0078] The displacement vector calculation device 201 is used to calculate the cloud displacement vector of the ground-based cloud image to obtain the displacement vector of the cloud cluster, where the displacement vector includes the displacement direction and displacement speed of the cloud cluster;

[0079] A sun center point determination module 202, used to determine the sun center point and a plurality of first pixel points around the sun center point;

[0080] The adjacency matrix establishment module 203 is used to establish an adjacency matrix according to the relative position of each first pixel point and the center point of the sun, and the displacement direction of the cloud cluster;

[0081] A first RGB value determination module 204, used to determine the RGB value of each first pixel point according to the displacement vector of the cloud;

[0082] A node feature matrix calculation module 205 is used to establish a node feature matrix according to the RGB value of each first pixel;

[0083] The second RGB value determination module 206 is used to input the adjacency matrix and the node feature matrix as graph structure data into a pre-trained graph convolution network prediction model to obtain the RGB value of the second pixel point located at the center of the sun;

[0084] The irradiance prediction module 207 is used to input the RGB value of the second pixel as an input feature into a pre-trained irradiance prediction model to obtain an irradiance prediction result at a target moment.

[0085] The further functional description of each of the above modules is the same as that of the above corresponding embodiments and will not be repeated here.

[0086] The irradiance prediction device in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0087] The embodiment of the present invention also provides a computer device having the above Figure 2 The irradiance prediction device shown.

[0088] See also Figure 3 , Figure 3 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Figure 3 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 3 A processor 10 is taken as an example.

[0089] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0090] The memory 20 stores instructions executable by at least one processor 10, so that at least one processor 10 executes the method shown in the above embodiment.

[0091] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0092] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0093] The computer device also includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 may be connected via a bus or other means. Figure 3 The example of connecting through bus is taken in the following.

[0094] The input device 30 can receive input digital or character information, and generate key signal input related to the user settings and function control of the computer device, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, an indicator bar, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 may include a display device, an auxiliary lighting device (e.g., an LED) and a tactile feedback device (e.g., a vibration motor), etc. The above-mentioned display device includes but is not limited to a liquid crystal display, a light emitting diode, a display and a plasma display. In some optional embodiments, the display device can be a touch screen.

[0095] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0096] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A method for predicting irradiance, characterized in that: The method comprises: Calculating the cloud displacement vector of the ground-based cloud image to obtain the displacement vector of the cloud cluster, wherein the displacement vector includes the displacement direction and displacement speed of the cloud cluster; Determine the sun center point and a plurality of first pixel points around the sun center point; Establishing an adjacency matrix according to the relative position of each of the first pixel points and the center point of the sun, and the displacement direction of the cloud cluster; Determine the RGB value of each first pixel point according to the displacement vector of the cloud; Establishing a node feature matrix according to the RGB values ​​of each first pixel point; Inputting the adjacency matrix and the node feature matrix as graph structure data into a pre-trained graph convolutional network prediction model to obtain the RGB value of a second pixel point located at the center of the sun; The RGB value of the second pixel is input as an input feature into a pre-trained irradiance prediction model to obtain an irradiance prediction result at the target time.

2. The method according to claim 1, characterized in that The step of determining a plurality of first pixel points around the sun center point includes: In a plurality of directions centered at the center of the sun, first pixel points at a preset distance from the center of the sun are respectively determined, and the preset distance is determined according to the displacement speed and unit time of the cloud cluster.

3. The method according to claim 1 or 2, characterized in that: According to the relative position of each of the first pixel points and the center point of the sun, and the displacement direction of the cloud cluster, an adjacency matrix is ​​established, including: respectively calculating the angle between the direction of each first pixel point pointing to the sun center point and the displacement direction of the cloud cluster; Determine the correlation between each first pixel point and a target event according to the angle, wherein the target event is that the first pixel point covers the center point of the sun at the target time; The adjacency matrix is ​​established according to the correlation.

4. The method according to claim 3, characterized in that Determining the relevance of each first pixel point to the occurrence of the target event according to the angle includes: If the angle between the direction of the first pixel pointing to the sun center point and the displacement direction of the cloud is less than a correlation threshold, it is determined that the first pixel is related to the target event; If the angle between the direction of the first pixel point pointing to the sun center point and the displacement direction of the cloud is greater than the correlation threshold, it is determined that the first pixel point is irrelevant to the target event.

5. The method according to claim 3, characterized in that: Establishing a node feature matrix according to the RGB values ​​of each first pixel point includes: The node feature matrix is ​​established according to the R value, G value, and B value of each of the first pixel points.

6. The method according to claim 1, characterized in that The input features also include current relative humidity, aerosol optical depth at 500 nm, and historical irradiance data.

7. An irradiance prediction device, characterized in that: The device comprises: A displacement vector calculation device, used to calculate the cloud displacement vector of the ground-based cloud image to obtain the displacement vector of the cloud cluster, wherein the displacement vector includes the displacement direction and displacement speed of the cloud cluster; A solar center point determination module, used to determine the solar center point and a plurality of first pixel points around the solar center point; An adjacency matrix establishment module, used to establish an adjacency matrix according to the relative position of each of the first pixel points and the center point of the sun, and the displacement direction of the cloud cluster; A first RGB value determination module, used to determine the RGB value of each first pixel point according to the displacement vector of the cloud; A node feature matrix calculation module, used to establish a node feature matrix according to the RGB values ​​of each first pixel point; A second RGB value determination module is used to input the adjacency matrix and the node feature matrix as graph structure data into a pre-trained graph convolution network prediction model to obtain the RGB value of a second pixel point located at the center of the sun; The irradiance prediction module is used to input the RGB value of the second pixel point as an input feature into a pre-trained irradiance prediction model to obtain an irradiance prediction result at a target moment.

8. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the irradiance prediction method according to any one of claims 1 to 6 by executing the computer instructions.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the irradiance prediction method according to any one of claims 1 to 6.

10. A computer program product, characterized in that The method comprises computer instructions for causing a computer to execute the irradiance method according to any one of claims 1 to 6.

Images