A method for predicting surface solar radiation energy by combining ground-based cloud images and radiation observations

Abstract

The present application discloses a method for predicting surface solar radiation energy by combining ground-based cloud maps and radiation observations. The method includes: extracting the effective area of ​​cloud pixels according to the clear sky index and solar zenith angle corresponding to the target ground-based cloud map; determining the motion vector of the cloud cluster according to the effective area of ​​cloud pixels, and determining the solar occlusion coefficient based on the motion vector; obtaining the prediction result of surface radiation in the future period based on the solar occlusion coefficient and the machine learning model. The method extracts the effective area of ​​cloud pixels from the target ground-based cloud map by the clear sky index and solar zenith angle corresponding to the target ground-based cloud map, so that the recognition efficiency of cloud pixels is high and more accurate. At the same time, when the solar occlusion coefficient and historical surface radiation values ​​determined based on the motion vector are input into the machine learning model, the prediction result is generated to improve the accuracy and speed of the prediction. The method does not rely on external meteorological data as an auxiliary, and can be applied to ultra-short-term photovoltaic radiation prediction.

Description

Technical Field

[0001] The present application relates to the technical field of surface radiation analysis, and in particular to a method for predicting surface solar radiation energy by combining ground-based cloud maps and radiation observations. Background Art

[0002] Electricity resources are essential resources for human life. For a long time, human beings have relied heavily on the use of fossil fuels to generate electricity, resulting in the consumption of a large amount of non-renewable resources, exacerbating the process of global warming and causing great damage to the ecological environment on which human beings depend for survival. The sun is a huge natural fusion nuclear reactor, which will continue to deliver stable energy to the earth for a long time in the future. Since its energy is mainly concentrated in the short-wave range, solar radiation is also called short-wave radiation. The use of solar energy for power generation has great advantages such as safety, efficiency, easy access and no pollution. In recent years, the steady development of photovoltaic power generation technology at home and abroad has greatly reduced the cost of photovoltaic power generation and greatly improved the competitiveness of photovoltaic power generation technology.

[0003] However, the instability of photovoltaic power generation is a major obstacle to its large-scale use. The instability of photovoltaic power generation technology is mainly due to the intermittent and uncertain solar radiation received by the ground. The formation and movement of clouds is one of the fundamental reasons for the uncertainty of surface radiation changes. Due to the blocking of clouds, solar shortwave radiation will be attenuated to varying degrees. If these situations are not handled in a timely manner, it may lead to partial paralysis of the power supply system. Therefore, accurate prediction of solar radiation has become one of the key technologies to achieve large-scale integration of photovoltaic power generation into the power grid.

[0004] In related technologies, surface radiation prediction is usually solved by using the radiation transfer equation, which requires a large number of meteorological parameters as input, resulting in slow calculation speed and room for improvement in prediction accuracy. Summary of the invention

[0005] In view of this, the present application provides a method for predicting surface solar radiation energy by combining ground-based cloud maps and radiation observations, which can improve the accuracy of surface radiation prediction and increase the prediction speed.

[0006] In a first aspect, the present application provides a method for predicting surface solar radiation energy by combining ground-based cloud images and radiation observations, comprising:

[0007] According to the clear sky index and solar zenith angle corresponding to the target ground-based cloud image, the effective area of ​​cloud pixels is extracted from the target ground-based cloud image;

[0008] Determine a motion vector of the cloud cluster according to the effective area of ​​the cloud pixel, and determine a solar shading coefficient based on the motion vector;

[0009] Based on the machine learning model, the prediction results of surface radiation in the future period are obtained according to the solar shading coefficient and historical surface radiation values.

[0010] Optionally, extracting a cloud pixel effective area from the target ground-based cloud image according to a clear sky index and a solar zenith angle corresponding to the target ground-based cloud image includes:

[0011] According to the clear sky index and solar zenith angle corresponding to the target ground-based cloud image, based on the predefined correspondence between the clear sky index, solar zenith angle, and cloud recognition method, the target cloud recognition method applicable to the target ground-based cloud image is determined;

[0012] The cloud pixel effective area is extracted from the target ground-based cloud image according to the target cloud recognition method.

[0013] Optionally, when the target cloud recognition method is an adaptive threshold method, the target cloud recognition method includes:

[0014] The blue-red-blue-green ratio characteristic value BRBG of the target cloud image is determined by the following formula:

[0015] ;

[0016] In the above formula, R The red band DN value, B The blue band DN value, G Green band DN value;

[0017] The classification threshold T is determined based on the maximum inter-class variance method, and the pixels whose BRBG does not exceed the threshold T are established as cloud pixels, and then the effective area of ​​cloud pixels is determined.

[0018] Optionally, when the target cloud recognition method is a clear sky background difference method, the target cloud recognition method includes:

[0019] According to the solar zenith angle corresponding to the target ground-based cloud image, a clear sky background cloud image is established from the predefined clear sky cloud library image;

[0020] The normalized ratio characteristics NRBR of the target ground-based cloud image and the clear sky background cloud image are determined by the following formulas:

[0021] ;

[0022] In the above formula, R The red band DN value, B The blue band DN value, G Green band DN value;

[0023] The difference between the NRBR of the target ground-based cloud image and the NRBR of the clear sky background cloud image is determined, and pixels whose difference exceeds a threshold are defined as cloud pixels, thereby determining a cloud pixel effective area.

[0024] Optionally, when the target cloud recognition method is a regional threshold method, the target cloud recognition method includes:

[0025] The blue-red-blue-green ratio characteristic value BRBG of the target ground-based cloud map partition is determined by the following formula:

[0026] ;

[0027] In the above formula, R The red band DN value, B The blue band DN value, G Green band DN value;

[0028] Determining classification threshold based on maximum inter-class variance method TH 1 , and according to the classification threshold TH 1 The cloud recognition threshold is determined by the clear sky value of the target ground-based cloud image TH 2 ;

[0029] Partition the target ground-based cloud image to obtain the area around the solar disk and other areas;

[0030] According to the cloud identification threshold TH 2 Identification of the region around the solar disk;

[0031] Adaptive threshold algorithm is used to identify other areas.

[0032] Optionally, when the target cloud recognition method is a fixed threshold method, the target cloud recognition method includes:

[0033] The blue-red-blue-green ratio characteristic value BRBG of the target cloud image is determined by the following formula:

[0034] ;

[0035] In the above formula, R The red band DN value, B The blue band DN value, G Green band DN value;

[0036] The pixels whose BRBG exceeds the threshold are established as cloud pixels, and then the cloud pixel valid area is determined.

[0037] Optionally, the motion vector of the cloud is determined according to the cloud pixel effective area, specifically:

[0038] Based on a block matching algorithm, a motion vector of the cloud is determined according to the effective area of ​​the cloud pixels.

[0039] Optionally, determining a solar shading coefficient based on the motion vector comprises:

[0040] Extract the main motion vector from the motion vector based on the k-means algorithm;

[0041] Determine a cloud grid around the sun according to the main motion vector;

[0042] Determining a cloud fraction at a predicted time in the cloud grid around the sun according to the length, width and pixel information of cloud pixels of the cloud grid around the sun;

[0043] Based on the cloud fraction and in combination with predefined rules, a solar obstruction coefficient is determined.

[0044] In a second aspect, the present application provides a surface radiation prediction device based on a ground-based cloud map, comprising:

[0045] An extraction module is used to extract the cloud pixel effective area from the target ground-based cloud image according to the clear sky index and solar zenith angle corresponding to the target ground-based cloud image;

[0046] A determination module, configured to determine a motion vector of the cloud cluster according to the effective area of ​​the cloud pixel, and determine a solar shading coefficient based on the motion vector;

[0047] The prediction module is used to obtain the prediction results of surface radiation in the future period based on the machine learning model, the solar shading coefficient and the historical surface radiation values.

[0048] In a third aspect, the present application provides an execution device, including a processor and a memory, wherein the processor is coupled to the memory;

[0049] The memory is used to store programs;

[0050] The processor is used to execute the program in the memory, so that the execution device executes the above method.

[0051] In the method disclosed in the present application, different recognition algorithms are selected through the combination relationship between the clear sky index and the solar zenith angle corresponding to the target ground-based cloud map, and the effective area of ​​cloud pixels is extracted from the target ground-based cloud map, so that the recognition efficiency of cloud pixels is high and more accurate. At the same time, when the solar shading coefficient determined based on the motion vector and the historical surface radiation value are input into the machine learning model, the prediction result of surface radiation is generated, which improves the accuracy and speed of the prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] The technical solution and other beneficial effects of the present application will be made apparent by describing in detail the specific implementation methods of the present application in conjunction with the accompanying drawings.

[0053] Figure 1 An operation flow chart of a method for predicting solar radiation energy on the ground surface by combining ground-based cloud images and radiation observations provided by an exemplary embodiment is shown;

[0054] Figure 2 A diagram showing the cloud recognition effect around the solar disk provided by an exemplary embodiment is shown.

[0055] Figure 3 A comparison result diagram of a reconstructed predicted image and a real image provided by an exemplary embodiment is shown.

[0056] Figure 4 A graph showing surface radiation prediction results in cloudy weather provided by an exemplary embodiment is shown.

[0057] Figure 5 A graph showing surface radiation prediction results in clear weather provided by an exemplary embodiment is shown.

[0058] Figure 6 A block diagram of a surface solar radiation energy prediction device combining ground-based cloud images and radiation observations provided by an exemplary embodiment is shown.

[0059] Figure 7 A block diagram of the execution device provided by an exemplary embodiment is shown. DETAILED DESCRIPTION

[0060] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present application. However, the present application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present application, so the present application is not limited by the specific embodiments disclosed below.

[0061] The terms "first", "second", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described here can be implemented in an order other than that illustrated or described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or modules is not necessarily limited to those steps or modules clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products or devices. The naming or numbering of steps in this application does not mean that the steps in the method flow must be executed in the time / logical sequence indicated by the naming or numbering. The process steps that have been named or numbered can change the execution order according to the technical purpose to be achieved, as long as the same or similar technical effects can be achieved. The division of units in this application is a logical division. There may be other division methods when it is implemented in actual applications. For example, multiple units can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection between units can be electrical or other similar forms, which are not limited in this application. In addition, the units or sub-units described as separate components may or may not be physically separated, may or may not be physical units, or may be distributed in multiple circuit units, and some or all of the units may be selected according to actual needs to achieve the purpose of the present application.

[0062] Please refer to Figure 1 ,in Figure 1 A flowchart of a method for predicting solar radiation energy on the ground surface by combining ground-based cloud images and radiation observations provided by an exemplary embodiment is shown. The method is implemented through steps 102-106.

[0063] In step 102, a cloud pixel effective area is extracted from the target ground-based cloud image according to the clear sky index and the solar zenith angle corresponding to the target ground-based cloud image.

[0064] The technical contribution of the use of ground-based cloud images in the method of the present application is: in the related technology, satellites can obtain large-scale cloud information, but their temporal resolution is often low, and short-term radiation prediction cannot be achieved, which cannot meet the requirements of photovoltaic power generation systems; while the ground-based cloud images taken by the ground-based all-sky imager in the method of the present application can provide sub-kilometer-level high-resolution cloud images, and the temporal resolution is relatively high, usually reaching the minute level, which can achieve ultra-short-term prediction of surface radiation.

[0065] Here, as a reference for determining the clear sky index and the solar zenith angle, the following are included:

[0066] Step 102A1: Calculate the solar zenith angle according to the latitude and longitude of the observation site and the observation time corresponding to the target ground-based cloud map. , the specific formula is:

[0067] ;

[0068] in, is the latitude of the observation site expressed in radians, is the solar declination angle, The hour angle.

[0069] The calculation formula of the solar declination angle is:

[0070] ;

[0071] in, The accumulated days are the years;

[0072] Among them, the hour angle The calculation formula is:

[0073] ;

[0074] in, t For the local time.

[0075] Step 102A2: Calculate the solar azimuth angle according to the solar zenith angle, the longitude and latitude of the observation site, and the observation time corresponding to the target ground-based cloud map. , the specific formula is:

[0076] ;

[0077] in, is the latitude of the observation site expressed in radians, is the solar declination angle, is the solar altitude angle.

[0078] Step 102A3: Calculate the astronomical radiation according to the latitude and longitude of the observation site and the observation time corresponding to the target ground-based cloud map. , the specific formula is:

[0079] ;

[0080] Step 102A4: Calculate the clear sky index using the surface shortwave radiation at the observation time measured by the ground-based radiometer and the calculated astronomical radiation. , the specific formula is:

[0081] ;

[0082] in, DSR is the surface shortwave radiation value at the measurement time.

[0083] Step 102A5: Calculate the air quality according to the observation site height and the observation time corresponding to the target ground-based cloud map m , the specific formula is:

[0084] ;

[0085] in, h To measure the height of a point from the ground surface, μ is the solar zenith angle.

[0086] Before extracting the effective area of ​​cloud pixels from the target ground-based cloud image, a clear sky cloud library image can be constructed first.

[0087] Here, the construction of the clear sky cloud library map can include:

[0088] Step 102B11, rotating the ground-based cloud image by the angle of the solar azimuth angle α in a clockwise direction, so that the sun is located on the vertical bisector of each image, and thus the position of the sun on the cloud image is determined only by the solar zenith angle μ;

[0089] Step 102B12: classify the historical clear sky images into one category according to the solar zenith angle according to a preset angle (including but not limited to 10°), and establish a clear sky cloud image library index.

[0090] As an exemplary embodiment, extracting a cloud pixel effective area from the target ground-based cloud image according to the clear sky index and the solar zenith angle corresponding to the target ground-based cloud image includes:

[0091] Step 102B1, according to the clear sky index and solar zenith angle corresponding to the target ground-based cloud image, and based on the predefined correspondence between the clear sky index, solar zenith angle, and cloud recognition method, determine the target cloud recognition method applicable to the target ground-based cloud image.

[0092] It should be noted that the above correspondence can be exemplarily represented as shown in Table 1:

[0093] Table 1

[0094]

[0095] The following is a detailed explanation of different cloud identification methods.

[0096] Ⅰ. When the target cloud recognition method is an adaptive threshold method, the target cloud recognition method includes:

[0097] Step 102B1A1, determine the blue-red-blue-green ratio characteristic value BRBG of the target cloud image by the following formula:

[0098]

[0099] In the above formula, R The red band DN value, B The blue band DN value, G Green band DN value;

[0100] Step 102B1A2: determine the classification threshold T based on the maximum inter-class variance method, and establish pixels whose BRBG does not exceed the threshold T as cloud pixels, and then determine the cloud pixel valid area.

[0101] The operation process of the maximum between-class variance method in step 102B2A2 is specifically demonstrated below.

[0102] Assume that the probability of a pixel being classified as a cloud pixel is , then the average grayscale of cloud pixels The specific calculation formula is:

[0103] ;

[0104] in, Pixel probability.

[0105] Average grayscale of sky pixels Similarly, the average grayscale of the entire image is M It is expressed as:

[0106] ;

[0107] According to the concept of variance, the inter-class variance of cloud pixels and sky pixels can be expressed as:

[0108] ;

[0109] The average gray M Substitute the above formula and simplify it to get:

[0110] ;

[0111] satisfy of k The optimal threshold for segmentation is selected when the BRBG feature value of the cloud image is less than k A pixel is defined as a cloud pixel, whereas a negative pixel is a sky pixel.

[0112] II. When the target cloud recognition method is a clear sky background difference method, the target cloud recognition method includes:

[0113] Step 102B1B1, establishing a clear sky background cloud image from the clear sky cloud library image according to the solar zenith angle corresponding to the target ground-based cloud image;

[0114] Step 102B1B2: Determine the normalized ratio feature NRBR of the target ground cloud image and the clear sky background cloud image respectively by the following formula:

[0115] ;

[0116] In the above formula, R The red band DN value, B The blue band DN value, G Green band DN value;

[0117] Step 102B1B3, determine the difference between the NRBR of the target ground-based cloud image and the NRBR of the clear sky background cloud image, and establish the pixels whose difference exceeds the threshold (including but not limited to 17) as cloud pixels; it can be understood that the pixels whose difference is less than the threshold (including but not limited to 17) are defined as sky pixels, thereby determining the effective area of ​​cloud pixels.

[0118] III. When the target cloud recognition method is a regional threshold method, the target cloud recognition method includes:

[0119] Step 102B1C1, determine the blue-red-blue-green ratio characteristic value BRBG of the target ground-based cloud image partition by the following formula:

[0120] ;

[0121] In the above formula, R The red band DN value, B The blue band DN value, G Green band DN value;

[0122] Step 102B1C2: Determine the classification threshold based on the maximum inter-class variance method and BRBG TH 1 , and according to the classification threshold TH 1 and the clear sky value of the target ground-based cloud image K t Determining the cloud identification threshold TH 2 ;

[0123] In this step 102B1C2, the cloud recognition threshold TH 2 The calculation formula can be as follows,

[0124] ;

[0125] Step 102B1C3, partition the target ground-based cloud image into regions around the solar disk and other regions;

[0126] In this step 102B1C3, the specific operation process is to divide the target cloud image into three parts: the sun position area, the area around the sun disk, and the other areas. The sun area is the actual position and size of the sun on the image, the area around the sun disk is the area with the center of the sun as the center and a radius of 500 pixels, and the area other than that is the other area.

[0127] Step 102B1C4: According to the cloud recognition threshold TH 2 For the identification of the area around the solar disk, the maximum inter-class variance method is used, and the adaptive threshold algorithm is combined to identify other areas.

[0128] It should be added that the technical contribution of the regional threshold method in this example is: the related technology only uses the spectral information of the image to extract the cloud area, and the recognition accuracy around thin clouds and the sun is not high. In order to improve the recognition accuracy around the solar disk, some people have proposed a recognition method that uses the clear sky background as a difference, but this method can only get good results when the solar disk is clear. When there are clouds on the solar disk, it will cause a large number of missed recognitions, resulting in large errors.

[0129] The inventors have noticed that clouds are one of the main factors that affect the radiation reaching the surface of the earth. The changes in clouds are random, so the changes in clouds cannot be accurately captured by relying solely on the image information of cloud maps. In addition, high-brightness solar glare often appears around the sun on ground-based cloud maps, which can easily cause confusion with clouds, leading to misidentification and affecting the accuracy of solar energy prediction. When the radiation value measured on the surface is higher, the cloud amount around the sun is generally smaller. In this example, the regional threshold method combined with the measured radiation value corrects the misidentification of solar glare, thereby effectively improving the accuracy of cloud recognition around the sun.

[0130] IV. When the target cloud recognition method is a fixed threshold method, the target cloud recognition method includes:

[0131] Step 102B1D1, determine the blue-red-blue-green ratio characteristic value BRBG of the target cloud image by the following formula,

[0132] ;

[0133] In the above formula, RThe red band DN value, B The blue band DN value, G Green band DN value;

[0134] Step 102B1D2: Pixels whose BRBG exceeds a threshold value (including but not limited to 80) are established as cloud pixels, and pixels whose BRBG is lower than a threshold value (including but not limited to 80) are established as sky pixels.

[0135] Step 102B2: extracting the cloud pixel effective area from the target ground-based cloud image according to the target cloud recognition method.

[0136] In the actual operation of this step 102B2, the cloud pixel valid area can be marked as 1, the sky pixel area (ie, non-cloud pixel).

[0137] In step 104, a motion vector of the cloud cluster is determined according to the effective area of ​​the cloud pixel, and a solar shading coefficient is determined based on the motion vector.

[0138] In an exemplary embodiment, the motion vector of the cloud is determined according to the effective area of ​​the cloud pixels, specifically: based on a block matching algorithm, the motion vector of the cloud is determined according to the effective area of ​​the cloud pixels. More specifically, the target area for motion vector extraction is clipped according to the inscribed rectangle of the cloud image effective area; the motion vector matrix of the target area is calculated using the block matching algorithm, and the motion vector is screened using the binary mask of the extracted cloud area to extract the main motion vector.

[0139] In an exemplary embodiment, determining a solar shading coefficient based on the motion vector includes:

[0140] Step 104A1, extracting motion vectors based on a block matching algorithm, and extracting main motion vectors from the motion vectors based on a k-means algorithm;

[0141] Specifically, the actual operation process of step 104A1 may include:

[0142] Step 104A11: Input predicted time The two previous moments T 1 , T 2 Image, select T 1 Start the search with a point in the image at that moment as the center;

[0143] Step 104A12: construct a large diamond template with the center point as the center, and calculate the unmatched points and The absolute error and function of the current point in the image at a certain moment are calculated as follows:

[0144] ;

[0145] in, For image 1 Brightness value, Image 2 Brightness value.

[0146] Step 104A13: If the absolute error and function of the center point are the smallest, a small diamond template is constructed with the center, and the unmatched points are searched. T 2 The absolute error and function of the current point in the image at that moment.

[0147] Step 104A14: The point with the minimum absolute error and function is identified as the matching point, and the current block search ends. u、v That is the corresponding motion vector;

[0148] Step 104A15: If the absolute error and function minimum point in step 104A13 is a non-center point, reconstruct the large diamond template with the minimum absolute error and function point as the center, and repeat the operations from step 104A12 to step 104A15 until the current search is completed and the motion vector of each pixel is obtained. The points that have been matched are no longer involved in the calculation;

[0149] Step 104A16, using the binary mask of the cloud generated above to perform mask calculation on the calculated motion vector, to obtain the motion vector of the cloud area;

[0150] Step 104A17: Perform K-means clustering on the motion vectors in the cloud area to calculate the main motion vector.

[0151] Step 104A2: Determine the cloud grid around the sun according to the main motion vector.

[0152] Specifically, the specific operation process of this step 104A2 is to establish a cloud grid of 20×20 pixels in size, with the sun position at the previous moment as the origin, the direction being the opposite direction of the cloud movement, and the distance being the predetermined time interval of the main motion vector.

[0153] Step 104A3: According to the length of the cloud grid around the sun M ,width N and pixel information of Cloud Pixel , determine the predicted time in the cloud grid around the sun of cloud fractions.

[0154] Specifically, determine the predicted time in the cloud grid around the sun Cloud score This can be done using the following formula:

[0155] ;

[0156] in, The length and width of the grid, are the points in the grid identified as cloud pixels in the previous step.

[0157] Step 104A4: Determine the solar shielding coefficient according to the cloud fraction and in combination with predefined rules.

[0158] In this step 104A4, the predefined rules are as shown in Table 2 below:

[0159] Table 2

[0160]

[0161] In step 106, based on the machine learning model, the prediction result of the surface radiation in the future period is obtained according to the solar shielding coefficient and the historical surface radiation value.

[0162] Here, the machine learning model includes but is not limited to a random forest model or other forms.

[0163] Here, the specific value of the future period can be four hours, or as needed.

[0164] Specifically, taking the random forest model as an example to demonstrate the execution process of this step 106, it may include:

[0165] Step 1061, inputting the solar shielding coefficient;

[0166] Step 1062: Input the predicted time Calculate the solar zenith angle at the predicted time ;

[0167] Step 1063: Input the predicted time Calculation of astronomical radiation at the predicted moment ;

[0168] Step 1064: Input the predicted time Calculated air quality at the forecast time ;

[0169] Step 1065: Enter the predicted time The radiation values ​​in the historical period (including but not limited to the first 5 minutes, first 6 minutes, and first 7 minutes) are used as reference radiation values;

[0170] Step 1066: construct a random forest regression model with decision trees (including but not limited to 30 trees), and based on the above input data, predict the radiation at the prediction moment, so as to realize the radiation prediction for the future period (including but not limited to 4 hours on a scale of 5 minutes).

[0171] Please refer to Figure 6 , which shows a block diagram of the surface solar radiation energy prediction device provided by the present application in combination with ground-based cloud maps and radiation observation. The device 200 includes:

[0172] An extraction module 202 is used to extract a cloud pixel effective area from the target ground-based cloud image according to the clear sky index and the solar zenith angle corresponding to the target ground-based cloud image;

[0173] A determination module 204 is used to determine a motion vector of the cloud cluster according to the cloud pixel effective area, and determine a solar shading coefficient based on the motion vector;

[0174] The prediction module 206 is used to obtain the prediction result of the surface radiation in the future period based on the machine learning model according to the solar shielding coefficient and the historical surface radiation value.

[0175] Since the above methods have been discussed in detail, the specific implementation of the above modules will not be repeated here.

[0176] Next, an execution device provided by an embodiment of the present application is introduced. Figure 7 , Figure 7 This is a schematic diagram of a structure of an execution device provided in an embodiment of the present application. The execution device 300 can be specifically manifested as an autonomous driving vehicle, a mobile phone, a tablet, a laptop computer, a desktop computer, a monitoring data processing device, etc., which is not limited here. Figure 1 The functions of the execution device in the corresponding embodiment. Specifically, the execution device 300 includes: a receiver 301, a transmitter 302, a processor 303 and a memory 304 (wherein the number of the processor 303 in the execution device 300 can be one or more, Figure 7 In the example of FIG. 3 , the processor 303 may include an application processor 3031 and a communication processor 3032. In some embodiments of the present application, the receiver 301, the transmitter 302, the processor 303 and the memory 304 may be connected via a bus or other means.

[0177] The memory 304 may include a read-only memory and a random access memory, and provides instructions and data to the processor 303. A portion of the memory 304 may also include a non-volatile random access memory (NVRAM). The memory 304 stores processor and operation instructions, executable modules or data structures, or subsets thereof, or extended sets thereof, wherein the operation instructions may include various operation instructions for implementing various operations.

[0178] The processor 303 controls the operation of the execution device. In a specific application, the various components of the execution device are coupled together through a bus system, wherein the bus system includes not only a data bus but also a power bus, a control bus, and a status signal bus, etc. However, for the sake of clarity, various buses are referred to as bus systems in the figure.

[0179] The method disclosed in the above embodiment of the present application can be applied to the processor 303, or implemented by the processor 303. The processor 303 can be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit or software instructions in the processor 303. The above processor 303 can be a general-purpose processor, a digital signal processor (digital signal processing, DSP), a microprocessor or a microcontroller, and can further include an application specific integrated circuit (application specific integrated circuit, ASIC), a field programmable gate array (field-programmable gate array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The processor 303 can implement or execute the various methods, steps and logic block diagrams disclosed in the embodiments of the present application. The general processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the embodiment of the present application can be directly embodied as a hardware decoding processor to be executed, or the hardware and software modules in the decoding processor can be combined and executed. The software module may be located in a storage medium mature in the art, such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 304, and the processor 303 reads the information in the memory 304 and completes the steps of the above method in combination with its hardware.

[0180] The receiver 301 can be used to receive inputted digital or symbol information, and generate signal input related to the relevant settings and function control of the execution device. The transmitter 302 can be used to output the digital or symbol information through the first interface; the transmitter 302 can also be used to send instructions to the disk group through the first interface to modify the data in the disk group; the transmitter 302 can also include a display device such as a display screen.

[0181] In the embodiment of the present application, the processor 303 is used to execute Figure 1 The specific manner in which the application processor 3031 in the processor 303 performs the above steps is the same as that in the present application. Figure 1 The corresponding method embodiments are based on the same concept, and the technical effects they bring are the same as those in this application. Figure 1 The corresponding method embodiments are the same. For specific contents, please refer to the description in the method embodiments shown above in this application, which will not be repeated here.

[0182] The above is only a preferred specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed in the present application should be covered within the protection scope of the present application.

Claims

1. A method for predicting surface solar radiation energy by combining ground-based cloud images and radiation observations, characterized in that: include: According to the clear sky index and solar zenith angle corresponding to the target ground-based cloud image, the effective area of ​​cloud pixels is extracted from the target ground-based cloud image; Determine a motion vector of the cloud cluster according to the effective area of ​​the cloud pixel, and determine a solar shading coefficient based on the motion vector; Based on the machine learning model, the prediction results of surface radiation in the future period are obtained according to the solar shielding coefficient and historical surface radiation values; The effective area of ​​cloud pixels is extracted from the target ground-based cloud image according to the clear sky index and solar zenith angle corresponding to the target ground-based cloud image, including: According to the clear sky index and solar zenith angle corresponding to the target ground-based cloud image, and based on the predefined correspondence between the clear sky index, solar zenith angle, and cloud recognition method, the target cloud recognition method applicable to the target ground-based cloud image is determined, including: According to the latitude and longitude of the observation site and the observation time corresponding to the target ground-based cloud map, calculate the solar zenith angle: ； in, is the latitude of the observation site expressed in radians, is the solar declination angle, is the hour angle, is the solar zenith angle; Solar declination The calculation formula is: ； in, The accumulated days are the years; Hour angle The calculation formula is: ； in, t For local places; Calculate the solar azimuth angle according to the solar zenith angle, the longitude and latitude of the observation site, and the observation time corresponding to the target ground-based cloud map: ； in, is the latitude of the observation site expressed in radians, is the solar declination angle, is the solar altitude angle, is the solar azimuth; Calculate the astronomical radiation according to the latitude and longitude of the observation site and the observation time corresponding to the target ground-based cloud map: ； in, For astronomical radiation; The clear sky index is calculated using the shortwave radiation from the ground at the time of observation measured by the ground-based radiometer and the calculated astronomical radiation: ； in, DSR is the surface shortwave radiation value at the measurement time, is the clear sky index; Calculate the atmospheric mass based on the observation site height and the observation time corresponding to the target ground-based cloud map: in, h To measure the height of a point from the ground surface, μ is the solar zenith angle, m for air quality; Extracting the cloud pixel effective area from the target ground-based cloud image according to the target cloud recognition method, wherein before extracting the cloud pixel effective area, it also includes constructing a clear sky cloud library image; Determining a solar shading coefficient based on the motion vector includes: Extract the main motion vector from the motion vector based on the k-means algorithm; Determine a cloud grid around the sun according to the main motion vector; Determining a cloud fraction at a predicted time in the cloud grid around the sun according to the length, width and pixel information of cloud pixels of the cloud grid around the sun; Based on the cloud fraction and in combination with predefined rules, a solar obstruction coefficient is determined.

2. The method according to claim 1, characterized in that: When the target cloud recognition method is an adaptive threshold method, the target cloud recognition method includes: The blue-red-blue-green ratio characteristic value BRBG of the target cloud image is determined by the following formula: ； In the above formula, R The red band DN value, B The blue band DN value, G Green band DN value; The classification threshold T is determined based on the maximum inter-class variance method, and the pixels whose BRBG does not exceed the threshold T are established as cloud pixels, and then the effective area of ​​cloud pixels is determined.

3. The method according to claim 1, characterized in that When the target cloud recognition method is a clear sky background difference method, the target cloud recognition method includes: According to the solar zenith angle corresponding to the target ground-based cloud image, a clear sky background cloud image is established from the predefined clear sky cloud library image; The normalized ratio characteristics NRBR of the target ground-based cloud image and the clear sky background cloud image are determined by the following formulas: ； In the above formula, R The red band DN value, B The blue band DN value, G Green band DN value; The difference between the NRBR of the target ground-based cloud image and the NRBR of the clear sky background cloud image is determined, and pixels whose difference exceeds a threshold are established as cloud pixels, thereby determining a cloud pixel effective area.

4. The method according to claim 2, characterized in that: When the target cloud recognition method is a regional threshold method, the target cloud recognition method includes: The blue-red-blue-green ratio characteristic value BRBG of the target ground-based cloud map partition is determined by the following formula: ； In the above formula, R The red band DN value, B The blue band DN value, G Green band DN value; Determine the classification threshold based on the maximum inter-class variance method and BRBG TH 1, and according to the classification threshold TH 1 and the clear sky value of the target ground-based cloud image to determine the cloud recognition threshold TH 2; Partition the target ground-based cloud image to obtain the area around the solar disk and other areas; According to the cloud identification threshold TH 2. Identification of the area around the solar disk; Adaptive threshold algorithm is used to identify other areas.

5. The method according to claim 1, characterized in that When the target cloud recognition method is a fixed threshold method, the target cloud recognition method includes: The blue-red-blue-green ratio characteristic value BRBG of the target cloud image is determined by the following formula: ； In the above formula, R The red band DN value, B The blue band DN value, G Green band DN value; The pixels whose BRBG exceeds the threshold are established as cloud pixels, and then the effective area of ​​cloud pixels is determined.

6. The method according to claim 1, characterized in that The motion vector of the cloud is determined according to the effective area of ​​the cloud pixel, specifically: Based on a block matching algorithm, a motion vector of the cloud is determined according to the effective area of ​​the cloud pixels.

7. A surface solar radiation energy prediction device combining ground-based cloud maps and radiation observations, characterized in that: include: An extraction module is used to extract the cloud pixel effective area from the target ground-based cloud image according to the clear sky index and solar zenith angle corresponding to the target ground-based cloud image; A determination module, configured to determine a motion vector of the cloud cluster according to the effective area of ​​the cloud pixel, and determine a solar shading coefficient based on the motion vector; The prediction module is used to obtain the prediction results of surface radiation in the future period based on the machine learning model according to the solar shielding coefficient and historical surface radiation values; The extraction module is specifically used for: The effective area of ​​cloud pixels is extracted from the target ground-based cloud image according to the clear sky index and solar zenith angle corresponding to the target ground-based cloud image, including: According to the clear sky index and solar zenith angle corresponding to the target ground-based cloud image, and based on the predefined correspondence between the clear sky index, solar zenith angle, and cloud recognition method, the target cloud recognition method applicable to the target ground-based cloud image is determined, including: According to the latitude and longitude of the observation site and the observation time corresponding to the target ground-based cloud map, calculate the solar zenith angle: ； in, is the latitude of the observation site expressed in radians, is the solar declination angle, is the hour angle, is the solar zenith angle; Solar declination The calculation formula is: ； in, The accumulated days are the years; Hour angle The calculation formula is: ； in, t For local places; Calculate the solar azimuth angle according to the solar zenith angle, the longitude and latitude of the observation site, and the observation time corresponding to the target ground-based cloud map: ； in, is the latitude of the observation site expressed in radians, is the solar declination angle, is the solar altitude angle, is the solar azimuth; Calculate the astronomical radiation according to the latitude and longitude of the observation site and the observation time corresponding to the target ground-based cloud map: ； in, For astronomical radiation; The clear sky index is calculated using the shortwave radiation from the surface at the time of observation measured by the ground-based radiometer and the calculated astronomical radiation; ； in, DSR is the surface shortwave radiation value at the measurement time, is the clear sky index; Calculate the atmospheric mass based on the observation site height and the observation time corresponding to the target ground-based cloud map; in, h To measure the height of a point from the ground surface, μ is the solar zenith angle, m for air quality; Extracting the cloud pixel effective area from the target ground-based cloud image according to the target cloud recognition method, wherein before extracting the cloud pixel effective area, it also includes constructing a clear sky cloud library image; The determining module determines the sun shielding coefficient based on the motion vector, including: Extract the main motion vector from the motion vector based on the k-means algorithm; Determine a cloud grid around the sun according to the main motion vector; Determining a cloud fraction at a predicted time in the cloud grid around the sun according to the length, width and pixel information of cloud pixels of the cloud grid around the sun; Based on the cloud fraction and in combination with predefined rules, a solar obstruction coefficient is determined.

8. An execution device, characterized in that: comprising a processor and a memory, wherein the processor is coupled to the memory; The memory is used to store programs; The processor is configured to execute the program in the memory so that the execution device executes the method according to any one of claims 1 to 6.

Images