Method, device and equipment for predicting influence of clouds on intensity of solar radiation on ground

Abstract

The application relates to a cloud ground solar radiation intensity influence prediction method, device and equipment. The method comprises the following steps: acquiring initial ground images of a to-be-observed area at different moments, projecting the initial ground images on a ground plane after geometric transformation and distortion correction operations to obtain actual ground images at different moments; identifying a plurality of cloud shadows in each actual ground image, acquiring the shape, area and darkness of each cloud shadow and marking the same, and determining a plurality of target cloud shadows in each actual ground image; comparing the positions of the same target cloud shadow at adjacent moments to obtain the motion characteristics of the target cloud shadow, predicting the coverage area of the same target cloud shadow on the to-be-observed area within a preset time period in the future, and predicting the influence of the cloud on the ground solar radiation intensity based on the coverage area and the darkness. Thus, the problem that the cloud shadow information formed by the cloud on the ground is calculated according to the cloud information obtained by looking at the sky from the ground in the prior art, and the prediction effect is limited, is solved.

Description

Technical Field

[0001] This application relates to the field of solar energy application technology, and in particular to a method, apparatus and equipment for predicting the influence of clouds on the intensity of solar radiation on the ground. Background Technology

[0002] In ground-based centralized solar energy utilization projects (such as centralized photovoltaic and concentrated solar power), the actual intensity of solar radiation received by the ground is the main factor affecting energy output. However, ground-based solar radiation is easily affected by cloud cover in the short term, causing significant fluctuations. Therefore, ground-based centralized solar energy utilization projects have an urgent need for short-term forecasting of ground-based solar radiation intensity. In this regard, weather forecasts provide information with a large time scale and coarse spatial granularity, which cannot meet the needs of predicting short-term changes in illumination in localized areas.

[0003] Among the related technologies, it is proposed to use multiple cameras on the ground to collect sky images, identify cloud feature points based on the principle of binocular vision, and then combine spatial positioning information to calculate the cloud projection area on the ground and its movement speed towards the photovoltaic plant area. Based on the predicted proportion of the photovoltaic plant area covered by cloud shadows, the power generation capacity is predicted.

[0004] However, this method uses cloud information obtained from the ground looking up at the sky to infer the cloud shadow information formed on the ground. This results in a difference between the inferred cloud ground projection shape and the actual cloud shadow, which limits the prediction effect and urgently needs to be solved. Summary of the Invention

[0005] This application provides a method, apparatus, and equipment for predicting the influence of clouds on the intensity of solar radiation on the ground, in order to solve the problem that the existing technology relies on cloud information obtained from the ground looking into the sky to infer the cloud shadow information formed on the ground, which leads to limited prediction results, thereby improving the accuracy of prediction.

[0006] To achieve the above objectives, the first aspect of this application proposes a method for predicting the influence of clouds on the intensity of solar radiation on the ground, comprising the following steps:

[0007] Initial ground images of the area to be observed at different times are acquired, and after geometric transformation and distortion correction operations are performed on the initial ground images, they are projected onto the ground plane to obtain actual ground images at different times.

[0008] Based on the actual ground images at different times, identify at least one cloud shadow in the actual ground image at each time, obtain and mark the shape, area and darkness of the at least one cloud shadow, and determine at least one target cloud shadow in the actual ground image at each time.

[0009] By comparing the positions of the same target cloud shadow in two adjacent actual ground images, the motion characteristics of the same target cloud shadow can be obtained;

[0010] Based on the shape and motion characteristics of the same target cloud shadow, predict the coverage area of ​​the same target cloud shadow on the area to be observed within a preset time period in the future, and predict the influence of the cloud on the solar radiation intensity on the ground based on the coverage area and the darkness.

[0011] According to one embodiment of this application, before acquiring initial ground images of the area to be observed at different times, the method further includes:

[0012] Obtain the speed of cloud movement and the prediction duration;

[0013] Select a target photovoltaic power station and construct an observation area with the target photovoltaic power station as the center and the product of the cloud movement speed and the predicted duration as the radius.

[0014] According to one embodiment of this application, identifying at least one cloud shadow in the actual ground image at each time moment includes:

[0015] Obtain an initial reference ground image of the area to be observed during a preset period of clear, cloudless weather;

[0016] Based on the initial reference ground image, obtain a target reference ground image at a time close to the actual ground image at the current time;

[0017] Based on the target reference ground image and the actual ground image at the current moment, at least one cloud shadow in the actual ground image at each moment is identified using the principles of visible light imaging and infrared light imaging.

[0018] According to one embodiment of this application, predicting the impact of clouds on ground solar radiation intensity based on the coverage area and the darkness includes:

[0019] Acquire multiple three-dimensional coordinates of the photometric drone within the area of ​​the at least one target cloud shadow, and calculate multiple ground projection coordinates of the photometric drone in the direction of sunlight illumination based on the multiple three-dimensional coordinates;

[0020] Obtain multiple actual solar radiation intensities corresponding to the multiple ground projection coordinates, and match the multiple actual solar radiation intensities with the darkness of at least one target cloud shadow at the multiple ground projection coordinates based on the multiple ground projection coordinates and the multiple actual solar radiation intensities.

[0021] Obtain the theoretical solar radiation intensity at the location corresponding to the multiple ground projection coordinates, and calculate the degree of reduction in ground solar radiation intensity based on the relative difference between the theoretical solar radiation intensity and the actual solar radiation intensity.

[0022] Establish a correspondence between the darkness and the degree of reduction in the ground solar radiation intensity, and based on the correspondence, predict the impact of clouds on the ground solar radiation intensity according to the darkness.

[0023] According to one embodiment of this application, when predicting the impact of clouds on the ground solar radiation intensity based on the coverage area and the darkness, the method further includes:

[0024] Acquire the current sky image, current cloud image, current shooting date, current shooting time, and current camera pose of the area to be observed;

[0025] The current sky image, the current cloud image, the current shooting date, the current shooting time, and the current camera pose are input into a preset neural network model to obtain the current cloud shadow distribution and the current cloud shadow shading intensity.

[0026] Based on the current cloud shadow distribution and the current cloud shadow shading intensity, predict the impact of clouds on the solar radiation intensity on the ground.

[0027] The method for predicting the influence of clouds on the intensity of solar radiation on the ground, as proposed in this application, involves acquiring initial ground images of the area to be observed at different times, performing geometric transformations and distortion corrections on the initial ground images, and then projecting them onto the ground plane to obtain actual ground images at different times. Multiple cloud shadows are identified in the actual ground images at each time, and the shape, region, and darkness of each cloud shadow are obtained and labeled. Multiple target cloud shadows in the actual ground images at each time are determined, and their motion characteristics are obtained by comparing the positions of the same target cloud shadow at adjacent times. The coverage area of ​​the same target cloud shadow in the area to be observed within a preset time period is predicted, and the influence of clouds on the intensity of solar radiation on the ground is predicted based on the coverage area and darkness. This solves the problem in existing technologies where cloud shadow information is inferred from cloud information obtained from the ground looking towards the sky, leading to limited prediction results, and thus improves prediction accuracy.

[0028] To achieve the above objectives, a second aspect of this application provides a device for predicting the influence of clouds on the intensity of solar radiation on the ground, comprising:

[0029] The first acquisition module is used to acquire initial ground images of the area to be observed at different times, and to project the initial ground images onto the ground plane after performing geometric transformation and distortion correction operations to obtain actual ground images at different times.

[0030] The determination module is used to identify at least one cloud shadow in the actual ground image at each time based on the actual ground images at different times, obtain the shape, area and darkness of the at least one cloud shadow and mark it, and determine at least one target cloud shadow in the actual ground image at each time.

[0031] The second acquisition module is used to compare the position of the same target cloud shadow in two adjacent actual ground images to obtain the motion characteristics of the same target cloud shadow;

[0032] The prediction module is used to predict the coverage area of ​​the same target cloud shadow on the observation area within a preset time period in the future, based on the shape and motion characteristics of the same target cloud shadow, and to predict the influence of the cloud on the solar radiation intensity on the ground based on the coverage area and the darkness.

[0033] According to one embodiment of this application, before acquiring initial ground images of the area to be observed at different times, the first acquisition module is further configured to:

[0034] Obtain the speed of cloud movement and the prediction duration;

[0035] Select a target photovoltaic power station and construct an observation area with the target photovoltaic power station as the center and the product of the cloud movement speed and the predicted duration as the radius.

[0036] According to one embodiment of this application, the determining module is specifically used for:

[0037] Obtain an initial reference ground image of the area to be observed during a preset period of clear, cloudless weather;

[0038] Based on the initial reference ground image, obtain a target reference ground image at a time close to the actual ground image at the current time;

[0039] Based on the target reference ground image and the actual ground image at the current moment, at least one cloud shadow in the actual ground image at each moment is identified using the principles of visible light imaging and infrared light imaging.

[0040] According to one embodiment of this application, the prediction module is specifically used for:

[0041] Acquire multiple three-dimensional coordinates of the photometric drone within the area of ​​the at least one target cloud shadow, and calculate multiple ground projection coordinates of the photometric drone in the direction of sunlight illumination based on the multiple three-dimensional coordinates;

[0042] Obtain multiple actual solar radiation intensities corresponding to the multiple ground projection coordinates, and match the multiple actual solar radiation intensities with the darkness of at least one target cloud shadow at the multiple ground projection coordinates based on the multiple ground projection coordinates and the multiple actual solar radiation intensities.

[0043] Obtain the theoretical solar radiation intensity at the location corresponding to the multiple ground projection coordinates, and calculate the degree of reduction in ground solar radiation intensity based on the relative difference between the theoretical solar radiation intensity and the actual solar radiation intensity.

[0044] Establish a correspondence between the darkness and the degree of reduction in the ground solar radiation intensity, and based on the correspondence, predict the impact of clouds on the ground solar radiation intensity according to the darkness.

[0045] According to one embodiment of this application, when predicting the impact of clouds on the ground solar radiation intensity based on the coverage area and the darkness, the prediction module is further configured to:

[0046] Acquire the current sky image, current cloud image, current shooting date, current shooting time, and current camera pose of the area to be observed;

[0047] The current sky image, the current cloud image, the current shooting date, the current shooting time, and the current camera pose are input into a preset neural network model to obtain the current cloud shadow distribution and the current cloud shadow shading intensity.

[0048] Based on the current cloud shadow distribution and the current cloud shadow shading intensity, predict the impact of clouds on the solar radiation intensity on the ground.

[0049] The device for predicting the influence of clouds on the intensity of solar radiation on the ground, as proposed in this application, acquires initial ground images of the area to be observed at different times. After performing geometric transformation and distortion correction operations on the initial ground images, it projects them onto the ground plane to obtain actual ground images at different times. It identifies multiple cloud shadows in the actual ground images at each time, obtains and marks the shape, region, and darkness of each cloud shadow, determines multiple target cloud shadows in the actual ground images at each time, compares the positions of the same target cloud shadow at adjacent times to obtain its motion characteristics, and predicts the coverage area of ​​the same target cloud shadow in the area to be observed within a preset time period. Based on the coverage area and darkness, it predicts the influence of clouds on the intensity of solar radiation on the ground. This solves the problem in existing technologies where cloud shadow information is inferred from cloud information obtained from the ground looking towards the sky, leading to limited prediction results, thereby improving prediction accuracy.

[0050] To achieve the above objectives, a third aspect of this application provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method for predicting the influence of clouds on the intensity of solar radiation on the ground as described in the above embodiments.

[0051] To achieve the above objectives, a fourth aspect of this application provides a computer-readable storage medium having a computer program stored thereon, which is executed by a processor to implement the method for predicting the influence of clouds on the intensity of solar radiation on the ground as described in the above embodiments.

[0052] To achieve the above objectives, a fifth aspect of the present invention provides a computer program product, including a computer program that, when executed by a processor, is used to implement the method for predicting the influence of clouds on the intensity of solar radiation on the ground as described in the above embodiments.

[0053] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0054] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0055] Figure 1 This is a flowchart illustrating a method for predicting the influence of clouds on the intensity of solar radiation on the ground, according to an embodiment of this application.

[0056] Figure 2 This is a schematic diagram of an image taken by a system for predicting the influence of clouds on the intensity of solar radiation on the ground, according to an embodiment of this application.

[0057] Figure 3 This is a schematic diagram of an initial ground image captured by a camera according to an embodiment of this application and the corresponding projected (mapped) actual ground image;

[0058] Figure 4 This is a schematic diagram of the arrangement of a main UAV and a forward UAV according to an embodiment of this application;

[0059] Figure 5 A flowchart illustrating the identification of cloud shadows using a target-referenced ground image comparison strategy according to an embodiment of this application;

[0060] Figure 6 This is a block diagram of a device for predicting the influence of clouds on the intensity of solar radiation on the ground, according to an embodiment of this application.

[0061] Figure 7This is a schematic diagram of the structure of an electronic device provided according to an embodiment of this application. Detailed Implementation

[0062] The embodiments of this application are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0063] The following describes, with reference to the accompanying drawings, a method, apparatus, and device for predicting the influence of clouds on the intensity of solar radiation on the ground, according to embodiments of this application.

[0064] Figure 1 This is a flowchart of a method for predicting the influence of clouds on the intensity of solar radiation on the ground, according to one embodiment of this application.

[0065] Before introducing the method for predicting the influence of clouds on the intensity of solar radiation on the ground proposed in the embodiments of this application, let me briefly introduce the background of this application and the methods for predicting the influence of clouds on the intensity of solar radiation on the ground in related technologies.

[0066] In ground-mounted centralized solar energy utilization projects (such as centralized photovoltaic and concentrated solar power), the actual intensity of solar radiation received by the ground is the main factor affecting energy output. However, ground-mounted solar radiation is easily affected by cloud cover in the short term, causing significant fluctuations. Taking photovoltaic power plants as an example, the large short-term changes in power generation load caused by cloud cover significantly impact grid absorption and regulation, leading to "curtailment" of solar power. While solar thermal utilization systems (such as power generation or heating) are significantly more tolerant of light fluctuations than photovoltaic power generation, it is still necessary to predict solar radiation in advance to reduce the blind spots in the control and regulation of the production system. Therefore, there is an urgent need for short-term prediction of ground-mounted solar radiation intensity in ground-mounted centralized solar energy utilization projects.

[0067] Among the related technologies, some solutions have been proposed, such as (1) using multiple cameras on the ground to collect sky images, identifying cloud feature points based on the principle of binocular vision, and then combining spatial positioning information to calculate the cloud projection area on the ground and its speed of movement toward the photovoltaic plant area, and predicting the power generation capacity based on the predicted proportion of the area covered by cloud shadows in the photovoltaic plant area.

[0068] However, this method has two shortcomings: First, clouds have irregular three-dimensional shapes, and the shape of clouds seen from the ground camera looking up at the sky is different from the shape of clouds seen from the opposite direction of sunlight (or from the ground shadow of the clouds). Especially when the clouds are slightly farther away from the sun in terms of spherical angle, the calculated shape of the cloud's ground projection will differ from the actual cloud shadow. Second, this method essentially assumes that the power generation of photovoltaic plant areas blocked by clouds is reduced to zero, without considering that the "shading" intensity of clouds on sunlight is not always 100%. When the cloud layer is not very thick, a considerable portion of sunlight can still reach the ground.

[0069] (2) The time series of sky cloud images collected from the ground is used as one of the key input data to train the neural network to predict the photovoltaic power generation.

[0070] However, this method has a drawback: the cloud images it uses are ground-based photographs, which do not include or calculate information about the distance and height of the clouds. According to basic imaging principles, small, nearby clouds and large, distant clouds may appear very similar in a two-dimensional image, but their projected areas on the ground and the distance between the projection and the photographer differ significantly. Therefore, using only two-dimensional cloud images without considering three-dimensional spatial information limits the accuracy of cloud shadow prediction and consequently, the prediction of photovoltaic power generation.

[0071] Based on the aforementioned problems, this application proposes a method for predicting the impact of clouds on the intensity of solar radiation on the ground. This method directly observes cloud shadows on the ground from the air and predicts their future impact on the ground-based solar radiation of photovoltaic power plants based on the cloud shadows' location, movement characteristics, and shading intensity, thereby predicting changes in photovoltaic power generation. The ground-based cloud shadows seen from the air directly reflect the three-dimensional shape of the clouds, their orientation in the sky, and their shading capacity on the ground, making the prediction of ground-based solar radiation more direct and accurate.

[0072] To facilitate further understanding, the following describes the prediction system for the influence of clouds on the ground solar radiation intensity involved in the prediction method of the embodiment of this application.

[0073] In this embodiment, the system for predicting the impact of clouds on ground solar radiation intensity consists of a drone, a communication network, and a computer. The drone includes a main drone, a forward-deployed drone, and a photometric drone. For example, Figure 2As shown, the main UAV hovers at an altitude of several hundred to several thousand meters above a solar energy utilization ground station (such as a photovoltaic power station). A camera (i.e., the top camera) that can be controlled by a gimbal is mounted on the top of the main UAV, facing the sun, and is used to capture images of the sky and clouds centered on the sun. A set of cameras (i.e., the bottom camera, which includes a visible light camera and an infrared thermal imaging camera) is mounted on the bottom of the main UAV, and is used to capture visible light and infrared light images simultaneously. The bottom camera can rotate 360° horizontally, and the camera's optical axis is at an acute angle to the direction of gravity, capturing ground images within a radius of several kilometers to tens of kilometers centered on the photovoltaic power station. In addition, the main UAV is equipped with satellite positioning and attitude sensors, and the cameras (top and bottom cameras) are also equipped with positioning sensors, so that the spatial coordinates (longitude, latitude, and altitude) of the UAV (and cameras) and the orientation of the camera axis can be obtained in real time. The forward UAV is configured with the same configuration as the main UAV (multiple units can be set up) and is set at a certain distance (several kilometers or more) in the direction of cloud flow. The images it captures can also be transmitted to the computer to further improve the accuracy and timeliness of the prediction. The photometric UAV carries a solar radiometer and a gimbal device (so that the radiometer normal always points to the sun) and flies around the area to be observed, covering the entire area in a grid pattern. The photometric UAV is used to transmit its coordinates and the measured solar radiation intensity to the computer.

[0074] Furthermore, the bottom camera in this system can also be a fisheye camera, capable of capturing a single image covering 360° horizontally and 180° vertically without horizontal rotation. Its advantage is its simple structure, but the captured images require pre-processing for "fisheye" correction. Alternatively, the bottom camera can employ a strategy of multiple cameras arranged concentrically. The combined field of view of these cameras covers 360° horizontally, and the fields of view of adjacent cameras overlap. Therefore, all cameras do not need to rotate, and images captured simultaneously can be combined into a single image covering 360° horizontally. The advantages are no moving parts and high image resolution, but the disadvantages are higher cost and greater weight. The main drone can also be replaced by a weather balloon carrying a positioning and photographic device. The balloon can be fixed to the ground via cables, with power and communication cables provided. Compared to a drone, its advantages include longer flight time, greater payload capacity, and stronger resistance to wind and severe weather.

[0075] For large-scale photovoltaic power generation bases in "sand desert" areas, the base can be divided into multiple grid areas according to geographical location. Each grid area is equipped with the cloud influence prediction system of this application on the intensity of solar radiation on the ground. Since photovoltaic power generation plants in this type of base are generally distributed over a large area, the cloud shadow information observed by the system configured in each grid area can be shared by neighboring grid areas. This "equivalently" expands the observation range and prediction capability of a single grid area, and "shares" the cost by effectively reducing the number of observation hardware devices required.

[0076] For example, such as Figure 1 As shown, the method for predicting the impact of clouds on the intensity of solar radiation on the ground includes the following steps:

[0077] In step S101, initial ground images of the area to be observed at different times are acquired. After geometric transformation and distortion correction operations are performed on the initial ground images, they are projected onto the ground plane to obtain actual ground images at different times.

[0078] Specifically, in this application embodiment, the initial ground images of the area to be observed at different times can be captured by the bottom camera. Since the initial ground images captured by the bottom camera are captured when the camera optical axis is not perpendicular to the ground, the points on the initial ground images do not conform to the actual scale of the ground. Therefore, the initial ground images can be geometrically transformed and the camera imaging system distortion correction can be performed (i.e., by calculating and applying an appropriate transformation matrix, the distorted image can be restored to the original distortion-free state). Then, combined with the map data in the GIS (Geographic Information System), the images can be projected onto the ground plane where the photovoltaic power station is located to obtain the actual ground images at different times (the geographical features on the actual ground images are consistent with the real Earth surface).

[0079] It is understandable that the dominant wind direction in the same area over a period of time leads to a certain directionality in cloud movement. Clouds moving towards the photovoltaic power station and their shadows are the main factors affecting ground solar radiation and should therefore be closely monitored. Wind direction can be obtained from weather forecasts or determined by the direction of cloud shadow movement. Due to limitations imposed by the Earth's curvature and the performance of drones and cameras, the cloud influence prediction system for the intensity of ground solar radiation described in this application can further enhance the accuracy and timeliness of predictions by deploying forward drones at a certain distance (several kilometers or more) in the direction of cloud flow, in addition to the aforementioned drone hovering above the center of the photovoltaic power station (i.e., the main drone). Multiple forward drones can be deployed, and the ground images they capture can be transmitted to a computer for processing.

[0080] The following explains how the region to be observed is determined in the embodiments of this application.

[0081] In some embodiments, before acquiring initial ground images of the area to be observed at different times, the method further includes: acquiring the cloud movement speed and predicted duration; selecting a target photovoltaic power station, and constructing an area to be observed with the target photovoltaic power station as the center and the product of the cloud movement speed and predicted duration as the radius.

[0082] Specifically, the target photovoltaic power station to be predicted is determined, and the observation area is a circular ground area with the target photovoltaic power station as the center. The radius of this circular ground area can be calculated by multiplying the prediction duration and the cloud movement speed, whereby the cloud movement speed can be obtained from historical meteorological data.

[0083] For example, suppose we need to predict the trend of cloud influence on the intensity of solar radiation on the ground within the next 15 minutes (i.e., the prediction period). Based on historical data, we can find that the speed of local clouds does not exceed 72 km / h, which corresponds to a distance of no more than 18 km in 15 minutes. Therefore, we can use drones to monitor the distribution and movement of cloud shadows in a circular ground area with a radius of 18 to 20 km centered on the target photovoltaic power station.

[0084] In step S102, based on the actual ground images at different times, at least one cloud shadow in the actual ground image at each time is identified, and the shape, area and darkness of at least one cloud shadow are obtained and marked, thereby determining at least one target cloud shadow in the actual ground image at each time.

[0085] It is understandable that if the area to be observed is looking towards the sun, and there are clouds in the sky, the clouds will cast a darker shadow on the ground, which is called a cloud shadow.

[0086] In other words, based on the actual ground images at different times, at least one cloud shadow can be identified from the actual ground images at each time. For the detected cloud shadow, its features such as shape, region and darkness are extracted and visualized, such as using different colors or markers to identify each cloud shadow region, thereby determining at least one target cloud shadow in the actual ground image at each time.

[0087] The following details how to identify at least one cloud shadow in the actual ground image at each moment.

[0088] As one possible implementation, in some embodiments, identifying at least one cloud shadow in the actual ground image at each moment includes: acquiring an initial reference ground image of the area to be observed during a preset period of clear, cloudless weather; acquiring a target reference ground image at a time close to the actual ground image at the current moment based on the initial reference ground image; and identifying at least one cloud shadow in the actual ground image at each moment based on the target reference ground image and the actual ground image at the current moment, using visible light imaging principles and infrared light imaging principles.

[0089] Specifically, an initial reference ground image of the area to be observed is obtained during a recent period of clear, cloudless weather. Based on the initial reference ground image, a target reference ground image is selected from the image and compared with the actual ground image at the current time. Using the principles of visible light imaging and infrared light imaging, if there is a cloud shadow in the actual ground image at the current time, the brightness of that area will decrease in the visible light image and the temperature of that area will also decrease in the infrared image. Furthermore, the thicker the cloud layer, the more difficult it is for sunlight to penetrate, and the greater the decrease in brightness and temperature in the cloud shadow area. Therefore, at least one cloud shadow can be obtained by analyzing the difference between the target reference ground image and the actual ground image at the current time, thereby obtaining the shape, area, and darkness of the cloud shadow.

[0090] In addition to the above methods, cloud shadows can also be identified by the following two conventional methods: (1) Identify the boundaries of brightness abrupt changes on the visible light ground image. If the brightness is lower on one side of the boundary, it is a cloud shadow area (if the brightness is higher on the other side, it is an illuminated area). The difference between the average brightness of the cloud shadow area and the average brightness of the area outside the cloud shadow represents the darkness of the cloud shadow. Due to the uneven color characteristics of the ground itself, this method may mistakenly identify some landform boundaries with obvious color changes as cloud shadow boundaries. To reduce this interference, the following method (2) can be combined for cloud shadow identification; (2) Identify the boundaries of temperature abrupt changes on the infrared light ground image. If the temperature is lower on one side of the boundary, it is a cloud shadow area (if the temperature is higher on the other side, it is an illuminated area). The difference between the average temperature of the cloud shadow area and the average temperature of the area outside the cloud shadow represents the darkness of the cloud shadow. Compared with visible light, infrared light images are less affected by the local landform color characteristics of the ground, which can further improve the accuracy of cloud shadow identification.

[0091] Understandably, since most of my country's large-scale photovoltaic power generation bases are located in the "sand and barren" regions of Northwest China, with sparse or no vegetation, the surface color is mostly light, such as sandy yellow or earthy yellow. When the surface is in cloud shadow, the brightness of its visible light image changes significantly compared to when it is under direct sunlight, making it easy to identify the areas and darkness of cloud shadows. However, the photovoltaic panels themselves are darker in color, even approaching black at certain viewing angles. Whether in cloud shadow or under direct sunlight, the differences in their visible light image characteristics are not significant. In this case, it is advisable to use the infrared thermal imaging camera mentioned earlier and the infrared ground images it captures to provide better differentiation than visible light images.

[0092] It should be noted that cloud shadow recognition can employ one of the methods mentioned above, or a combination of several methods can be selected to improve the accuracy of cloud shadow recognition.

[0093] In step S103, the positions of the same target cloud shadow in two adjacent actual ground images are compared to obtain the motion characteristics of the same target cloud shadow.

[0094] Specifically, taking a target cloud shadow as an example, by comparing the position of the same target cloud shadow in the actual ground images at adjacent times, the motion characteristics (including the direction of movement and the speed of movement) of the target cloud shadow can be calculated by relevant calculation methods (such as optical flow). According to this method, the motion characteristics of each target cloud shadow can be obtained. It should be noted that as time goes by, cloud shadows that disappear or drift away from the observation area will no longer be tracked.

[0095] In step S104, based on the shape and motion characteristics of the same target cloud shadow, the coverage area of ​​the same target cloud shadow in the observation area within a preset time period is predicted, and the influence of the cloud on the ground solar radiation intensity is predicted based on the coverage area and darkness.

[0096] In other words, based on the shape and movement characteristics of the same target cloud shadow, the location of the target cloud shadow in the future can be predicted, that is, the coverage of the target cloud shadow in the area to be observed (such as the coverage area). Thus, based on the coverage area and darkness, the influence of clouds on the intensity of solar radiation on the ground can be predicted, and then the power change of photovoltaic power station under cloud shading can be predicted.

[0097] To facilitate understanding, the following section provides a detailed explanation of how to predict the impact of clouds on the intensity of solar radiation on the ground based on their coverage area and darkness.

[0098] As one possible approach, in some embodiments, predicting the impact of clouds on ground solar radiation intensity based on coverage area and darkness includes: acquiring multiple three-dimensional coordinates of a photometric drone within the area of ​​at least one target cloud shadow; calculating multiple ground projection coordinates of the photometric drone in the direction of sunlight illumination based on the multiple three-dimensional coordinates; acquiring multiple actual solar radiation intensities at positions corresponding to the multiple ground projection coordinates; matching the multiple actual solar radiation intensities with the darkness of at least one target cloud shadow at the positions corresponding to the multiple ground projection coordinates based on the multiple ground projection coordinates and the multiple actual solar radiation intensities; acquiring the theoretical solar radiation intensity at the positions corresponding to the multiple ground projection coordinates; calculating the degree of reduction in ground solar radiation intensity based on the relative difference between the theoretical solar radiation intensity and the actual solar radiation intensity; establishing a correspondence between darkness and the degree of reduction in ground solar radiation intensity; and predicting the impact of clouds on ground solar radiation intensity based on the correspondence.

[0099] It should be noted that although the darkness of cloud shadows is positively correlated with the degree of light shading by cloud shadows, the accurate quantitative relationship between the darkness of cloud shadows and the degree of reduction in the intensity of solar radiation on the ground still needs to be calibrated.

[0100] Specifically, the system for predicting the impact of clouds on the intensity of solar radiation on the ground is equipped with a photometric drone, which carries a solar radiometer and a gimbal (so that the radiometer normal always points to the sun). The photometric drone circulates over the area to be observed and covers the entire area in a grid pattern. A photometric drone transmits multiple three-dimensional coordinates within the area of ​​at least one target cloud shadow, along with the measured actual solar radiation intensities at the corresponding coordinate locations, to a computer. The computer can calculate multiple ground projection coordinates of the photometric drone in the direction of sunlight based on these three-dimensional coordinates. Based on these ground projection coordinates, the darkness of at least one target cloud shadow at the corresponding location in multiple actual ground images obtained by the main drone can be determined. The multiple actual solar radiation intensities are matched with the darkness of at least one target cloud shadow at the corresponding coordinate locations obtained from the main drone images and stored in a database. Using relevant technologies, the theoretical solar radiation intensity at the location corresponding to the multiple ground projection coordinates can be calculated. The relative difference between the theoretical and actual solar radiation intensities (expressed as a percentage) represents the degree of reduction in ground solar radiation intensity. The darkness of at least one target cloud shadow is matched with the degree of reduction in ground solar radiation intensity at the corresponding location, establishing a correspondence. Based on this correspondence, the degree of reduction in ground solar radiation intensity at the corresponding location can be obtained from the darkness of the target cloud shadow, thus determining the reduction in photovoltaic power generation when the target cloud shadow obstructs the area where a photovoltaic power station is located—that is, predicting the impact of clouds on ground solar radiation intensity.

[0101] By performing the above process multiple times under different times and weather conditions, the correlation between the darkness of the target cloud shadow and the degree of reduction in ground solar radiation intensity can be obtained. Due to seasonal variations in solar altitude angle and ground conditions, the aforementioned calibration process should be performed every few weeks. After establishing the correlation between the darkness of the target cloud shadow and the degree of reduction in ground solar radiation intensity, the degree of reduction in ground solar radiation intensity can be calculated based on the darkness of the cloud shadow, which in turn allows for the calculation of the reduction in photovoltaic power generation when the cloud shadow blocks the area where the photovoltaic panels are located.

[0102] Furthermore, in some embodiments, when predicting the impact of clouds on the intensity of solar radiation on the ground based on coverage area and darkness, the method further includes: acquiring the current sky image, current cloud image, current shooting date, current shooting time, and current camera pose of the area to be observed; inputting the current sky image, current cloud image, current shooting date, current shooting time, and current camera pose into a preset neural network model to obtain the current cloud shadow distribution and the current cloud shadow shading intensity; and predicting the impact of clouds on the intensity of solar radiation on the ground based on the current cloud shadow distribution and the current cloud shadow shading intensity.

[0103] In other words, based on the above-mentioned scheme principle for predicting the impact of clouds on the intensity of solar radiation on the ground, a preset neural network model can be constructed. That is, the historical sky images, historical cloud images, historical shooting dates, historical shooting times, and historical camera attitudes of the area to be observed, taken by the top camera of the main UAV, are used as inputs, and the corresponding historical cloud shadow distribution and historical cloud shadow shading intensity are used as outputs (prediction targets). The preset neural network is trained to obtain a preset neural network model that meets the preset conditions.

[0104] After obtaining the preset neural network model, it can be used to predict the current cloud shadow distribution and the current cloud shadow shading intensity. That is, by inputting the current sky image, the current cloud image, the current shooting date, the current shooting time, and the current camera pose into the preset neural network model, the current cloud shadow distribution and the current cloud shadow shading intensity can be obtained. Then, based on the current cloud shadow distribution and the current cloud shadow shading intensity, the influence of clouds on the solar radiation intensity on the ground can be predicted.

[0105] To help those skilled in the art further understand the prediction method for the influence of clouds on the intensity of solar radiation on the ground proposed in the embodiments of this application, the following detailed explanation is provided in conjunction with specific embodiments.

[0106] Suppose a photovoltaic power station is located in the Gobi Desert region of Northwest my country. The terrain around the area where the photovoltaic panels are located is relatively flat. The photovoltaic panels are arranged within a relatively regular square area of ​​2km east-west and 2km north-south. Accordingly, the actual distribution location of the photovoltaic panels can be marked in a GIS. Now, it is necessary to predict the impact of clouds on the ground solar radiation intensity in the next 10-15 minutes. According to historical meteorological data, the local cloud movement speed generally does not exceed 72km / h, corresponding to a movement distance of no more than 18km in 15 minutes. Therefore, a drone can be used to monitor the cloud shadow distribution and movement within a circular ground area (called the limit boundary circle) with a radius of 20km centered on the photovoltaic power station. A 3D photogrammetry model of the topography of this circular ground area is pre-built and incorporated into the GIS system.

[0107] Based on actual measurements of the local cloud base height, the main UAV was set to fly at an altitude of 1.5 km. The visible light and infrared cameras on the UAV's bottom both had a focal length of 3.5 mm, with horizontal and vertical field-of-view angles of 94° and 78° respectively. The cameras rotated around an axis perpendicular to the horizontal plane and captured multiple initial ground images, thus obtaining an initial ground image within a 360° horizontal field of view. Assuming the longer side (horizontal side) of the camera image also corresponds to the horizontal plane, the radius of the observed area depends on its vertical field of view (78°). At an altitude of 1.5 km, the radius of the observed ground area is approximately 7.05 km (≈1.5 × tan78°).

[0108] Because the drone is equipped with satellite positioning, its latitude, longitude, and altitude can be determined. The drone also has an attitude sensor, allowing the calculation of the relationship between its rigid body coordinate system and the geodetic coordinate system. The camera has a positioning sensor to determine the relationship between its rigid body coordinate system and the drone's rigid body coordinate system. If necessary, a gyroscope can be added to the camera to improve its attitude stability. Therefore, based on the drone's absolute position and the relative transformation of the coordinate systems, the camera's axis orientation can be obtained in real time. In other words, for each image captured by the camera, the latitude, longitude, altitude, and optical axis orientation (which can be represented by Euler angles relative to the geodetic coordinate system or other methods) of the camera at the time of capture are all determined. Therefore, each initial ground image captured by the camera can be geometrically transformed and mapped (projected) onto the geodetic coordinate system. Figure 3 A simple example is given of mapping (projecting) an initial ground image onto the ground to obtain an actual ground image. Figure 3 (a) is a schematic diagram of the initial ground image taken by the camera. Figure 3 (b) is a schematic diagram of the actual ground image after mapping (projection) onto the ground. In the aforementioned image mapping, feature point matching (i.e., for some points with obvious features in the terrain and landform in the initial ground image, finding their "corresponding points" in the three-dimensional terrain and landform model of the GIS system) can also be used to improve the accuracy of the mapping. The identification of target cloud shadows and the estimation of darkness are both performed on the mapped image (i.e., the actual ground image).

[0109] Because the area of ​​the boundary circle is relatively large, a single UAV cannot cover the entire area without translation. Furthermore, the UAV needs to be below the bottom of the clouds; otherwise, its observation of the ground may be obstructed by clouds, thus limiting its altitude and observation range. Therefore, in addition to the main UAV, three forward-deployed UAVs are deployed. Forward-deployed UAV 1 is positioned in the direction of the cloud flow (prevailing wind direction), meaning the line connecting forward-deployed UAV 1 and the main UAV points in the same direction as the prevailing wind. Forward-deployed UAVs 2 and 3 are symmetrically positioned on either side of this line. The distance between forward-deployed UAV 1 and the main UAV is 14 km. The perpendicular distances of forward-deployed UAVs 2 and 3 from the line connecting forward-deployed UAV 1 and the main UAV are both 7 km, and the perpendicular distance between the main UAV and the line connecting forward-deployed UAVs 2 and 3 is also 7 km. The top view of the total observation range covered by the four UAVs is shown below. Figure 4 The four dashed circles in (a) are shown. Alternatively, only two forward-deployed UAVs can be configured, i.e., forward-deployed UAVs 1 and 2 are on a straight line parallel to the prevailing wind direction and passing through the main UAV. The distance between forward-deployed UAV 1 and the main UAV, and the distance between forward-deployed UAV 2 and forward-deployed UAV 1, are both 7 km. The top view of the total observation range covered by the three UAVs is shown below. Figure 4(b) shows the three dashed circles. Compared to the previous configuration of three forward-deployed UAVs, this configuration reduces the number of forward-deployed UAVs by one, thus reducing system complexity. However, it also reduces the observation coverage area and decreases the adaptability to wind direction changes, making it more suitable for weather with relatively stable wind direction.

[0110] All initial ground images captured by drones at the same time can be transmitted to a ground computer. First, the aforementioned image mapping is performed to obtain the actual ground image. Then, multiple actual ground images are stitched together. Visible light images and infrared images complete the aforementioned process independently. The stitched image is then incorporated into the GIS system.

[0111] Next, the cloud shadows will be identified using the method described above, such as... Figure 5 As shown, the current actual ground image obtained by the camera is compared with the target reference ground image obtained at a similar time when it is clear and cloudless recently. In advance, under clear and cloudless weather (only the area within the limit boundary circle needs to be cloudless, not the entire sky), the main UAV and multiple forward UAVs are used to carry out the image shooting and mapping process described above in the local short-term prevailing wind direction (upwind of the photovoltaic power station) to obtain the target reference ground image. Since the solar declination angle does not change much within a few to ten days, the operation of shooting and obtaining the target reference ground image is not performed frequently and will not cause a large workload.

[0112] Then, the actual monitoring begins. The main UAV and multiple forward-deployed UAVs continuously capture initial ground images in the prevailing wind direction on a given day. The ground computer processes the images and compares them with target reference ground images taken at similar times. For visible light images, the focus is on comparing grayscale. Areas with significantly lower grayscale values ​​compared to the target reference ground image are considered cloud shadow areas (assuming no recent rainfall or other factors causing significant changes in surface color). The degree of lower grayscale value corresponds to the image darkness in the visible light image. For infrared images, the image values ​​directly represent temperature. Areas with significantly lower temperatures compared to the target reference ground image are considered cloud shadow areas. The degree of lower temperature corresponds to the image darkness in the infrared image. For areas in the target reference ground image that are predominantly characterized by light-colored "sandbar" features, the cloud shadow identification results obtained from the visible light image are prioritized. For areas in the target reference ground image that are darker in color (lower grayscale values), such as photovoltaic panels that appear bluish-black from a certain viewing angle, the cloud shadow identification results obtained from the infrared image are prioritized.

[0113] The relationship between the cloud shadow intensity and the reduction in ground solar radiation at various locations within the observation area is determined using a photometric drone equipped with a solar radiometer and a gimbal. The gimbal ensures that the radiometer normal always points towards the sun (the sun's position in the sky at any given moment, or its azimuth and elevation angles relative to the Earth's coordinate system, can be accurately calculated, thus ensuring the radiometer normal always points towards the sun regardless of whether it is obscured by clouds). A ground computer directs the photometric drone to the identified target cloud shadow area. Within this area, the drone circulates at a certain height above the ground (not necessarily too high, a few meters to tens of meters is sufficient), transmitting the three-dimensional coordinates and the measured actual solar radiation intensity to the computer in real time. The theoretical solar radiation intensity at a given ground location can be accurately calculated; the difference between the actual solar radiation intensity and the theoretical value represents the reduction in ground solar radiation intensity. The computer calculates the ground projection coordinates of the photometric drone in the direction of sunlight, matches the actual radiation intensity with the cloud shadow darkness at the corresponding coordinate location, and stores it in a database. By performing the above process multiple times at different times and under different sky and cloud conditions, the correspondence between cloud shadow darkness and the degree of reduction in ground solar radiation intensity can be obtained.

[0114] The workload of calibration depends on the richness of the local topography. If the topography within the boundary circle is uniform, the correspondence between cloud shadow darkness and the degree of reduction in ground solar radiation intensity is basically the same at each location. In this case, calibration only needs to be performed on a representative area, without needing to calibrate the entire region. After establishing the correspondence between cloud shadow darkness and the degree of reduction in ground solar radiation intensity, the degree of reduction in ground solar radiation intensity can be calculated based on cloud shadow darkness, which in turn allows for the calculation of the reduction in photovoltaic power generation when the cloud shadow blocks the area where the photovoltaic panel is located.

[0115] By comparing the ground position of cloud shadows at different nearby times, the direction and speed of cloud shadow movement can be calculated. This allows us to predict the position of cloud shadows and their overlap with photovoltaic panels in the future. Combined with the degree of reduction in ground solar radiation intensity caused by cloud shadows, we can predict the magnitude of change in photovoltaic power generation.

[0116] Furthermore, based on the aforementioned principle of predicting the impact of clouds on ground solar radiation intensity, a pre-defined neural network model is constructed. Using images of the sky and clouds, the date and time of image capture, and the camera's orientation as input, this model can predict cloud shadow distribution and its shading intensity. After practical verification demonstrating sufficient accuracy, the structure and implementation process of the prediction system for cloud impact on ground solar radiation intensity can be significantly simplified. For example, only multiple cameras observing the sky need to be set up on the ground, eliminating the need for drones observing the ground, to predict the distribution of cloud shadows and their shading intensity. Although the "Shagolan" region is vast, its climate, geography, and landforms are relatively simple and similar. A pre-defined neural network model obtained in one photovoltaic power station can be applied to photovoltaic power stations in other similar areas, thus facilitating large-scale promotion both technically and economically.

[0117] It should be noted that it is precisely because of the complete system and method proposed in this application that it is possible to construct and train a pre-defined neural network model. The latter is built on the foundation of the former, and the optimization and updating of the latter model cannot be separated from the continuous operation of the former and the continuous accumulation of images and data.

[0118] Therefore, by directly observing cloud shadows on the ground from the air, rather than observing clouds in the sky from the ground and then inferring the cloud shadows formed on the ground, many error factors and complex calculation processes caused by the latter indirect observation are eliminated. Based on the principle that the ground temperature decreases due to reduced solar radiation in the cloud shadow area, an infrared thermal imaging camera and the information it obtains are used to identify the cloud shadow area and estimate the darkness of the cloud shadow, which provides a powerful assistance to visible light-based observation methods and helps to reduce errors in cloud shadow identification and calculation.

[0119] The method for predicting the influence of clouds on the intensity of solar radiation on the ground, as proposed in this application, involves acquiring initial ground images of the area to be observed at different times, performing geometric transformations and distortion corrections on the initial ground images, and then projecting them onto the ground plane to obtain actual ground images at different times. Multiple cloud shadows are identified in the actual ground images at each time, and the shape, region, and darkness of each cloud shadow are obtained and labeled. Multiple target cloud shadows in the actual ground images at each time are determined, and their motion characteristics are obtained by comparing the positions of the same target cloud shadow at adjacent times. The coverage area of ​​the same target cloud shadow in the area to be observed within a preset time period is predicted, and the influence of clouds on the intensity of solar radiation on the ground is predicted based on the coverage area and darkness. This solves the problem in existing technologies where cloud shadow information is inferred from cloud information obtained from the ground looking towards the sky, leading to limited prediction results, and thus improves prediction accuracy.

[0120] Next, referring to the accompanying drawings, a device for predicting the influence of clouds on the intensity of solar radiation on the ground, according to an embodiment of this application, is described.

[0121] Figure 6 This is a block diagram of a device for predicting the influence of clouds on the intensity of solar radiation on the ground, according to an embodiment of this application.

[0122] like Figure 6 As shown, the cloud-based device 10 for predicting the influence of solar radiation intensity on the ground includes: a first acquisition module 100, a determination module 200, a second acquisition module 300, and a prediction module 400.

[0123] The first acquisition module 100 is used to acquire initial ground images of the area to be observed at different times, and after performing geometric transformation and distortion correction operations on the initial ground images, it projects them onto the ground plane to obtain actual ground images at different times.

[0124] The determination module 200 is used to identify at least one cloud shadow in the actual ground image at each time based on the actual ground image at different times, obtain and identify the shape, area and darkness of at least one cloud shadow, and determine at least one target cloud shadow in the actual ground image at each time.

[0125] The second acquisition module 300 is used to compare the position of the same target cloud shadow in two adjacent actual ground images to obtain the motion characteristics of the same target cloud shadow;

[0126] The prediction module 400 is used to predict the coverage area of ​​the same target cloud shadow in the observation area within a preset time period based on the shape and motion characteristics of the same target cloud shadow, and to predict the impact of the cloud on the solar radiation intensity on the ground based on the coverage area and darkness.

[0127] Furthermore, in some embodiments, before acquiring initial ground images of the area to be observed at different times, the first acquisition module 100 is further configured to:

[0128] Obtain the speed of cloud movement and the prediction duration;

[0129] Select a target photovoltaic power station and construct an observation area with the target photovoltaic power station as the center and the product of the cloud movement speed and the prediction duration as the radius.

[0130] Furthermore, in some embodiments, the determining module 200 is specifically used for:

[0131] Acquire an initial reference ground image of the area to be observed during a preset period of clear, cloudless weather;

[0132] Based on the initial reference ground image, obtain the target reference ground image at a time close to the current actual ground image;

[0133] Based on the target reference ground image and the actual ground image at the current moment, at least one cloud shadow in the actual ground image at each moment is identified using the principles of visible light imaging and infrared light imaging.

[0134] Furthermore, in some embodiments, the prediction module 400 is specifically used for:

[0135] Acquire multiple three-dimensional coordinates of the photometric drone within the area of ​​at least one target cloud shadow, and calculate multiple ground projection coordinates of the photometric drone in the direction of sunlight illumination based on the multiple three-dimensional coordinates;

[0136] Obtain multiple actual solar radiation intensities corresponding to multiple ground projection coordinates, and match the multiple actual solar radiation intensities with the darkness of at least one target cloud shadow at the multiple ground projection coordinates based on the multiple ground projection coordinates and the multiple actual solar radiation intensities.

[0137] The theoretical solar radiation intensity at locations corresponding to multiple ground projection coordinates is obtained. Based on the relative difference between the theoretical solar radiation intensity and the actual solar radiation intensity, the degree of reduction in ground solar radiation intensity is calculated.

[0138] Establish a correlation between darkness and the degree of reduction in ground solar radiation intensity, and based on this correlation, predict the impact of cloud cover on ground solar radiation intensity according to darkness.

[0139] Furthermore, in some embodiments, when predicting the impact of clouds on the intensity of solar radiation on the ground based on coverage area and darkness, the prediction module 400 is also used for:

[0140] Acquire the current sky image, current cloud image, current shooting date, current shooting time, and current camera pose of the area to be observed;

[0141] Input the current sky image, current cloud image, current shooting date, current shooting time, and current camera pose into the preset neural network model to obtain the current cloud shadow distribution and the current cloud shadow shading intensity;

[0142] Based on the current cloud shadow distribution and the current shading intensity of the cloud shadow, predict the impact of clouds on the intensity of solar radiation on the ground.

[0143] It should be noted that the explanation of the aforementioned method embodiment for predicting the influence of clouds on the intensity of solar radiation on the ground also applies to the device for predicting the influence of clouds on the intensity of solar radiation on the ground in this embodiment, and will not be repeated here.

[0144] The device for predicting the influence of clouds on the intensity of solar radiation on the ground, as proposed in this application, acquires initial ground images of the area to be observed at different times. After performing geometric transformation and distortion correction operations on the initial ground images, it projects them onto the ground plane to obtain actual ground images at different times. It identifies multiple cloud shadows in the actual ground images at each time, obtains and marks the shape, region, and darkness of each cloud shadow, determines multiple target cloud shadows in the actual ground images at each time, compares the positions of the same target cloud shadow at adjacent times to obtain its motion characteristics, and predicts the coverage area of ​​the same target cloud shadow in the area to be observed within a preset time period. Based on the coverage area and darkness, it predicts the influence of clouds on the intensity of solar radiation on the ground. This solves the problem in existing technologies where cloud shadow information is inferred from cloud information obtained from the ground looking towards the sky, leading to limited prediction results, thereby improving prediction accuracy.

[0145] Figure 7 A schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device may include:

[0146] The memory 701, the processor 702, and the computer program stored on the memory 701 and executable on the processor 702.

[0147] When the processor 702 executes the program, it implements the method for predicting the influence of clouds on the intensity of solar radiation on the ground provided in the above embodiments.

[0148] Furthermore, electronic devices also include:

[0149] Communication interface 703 is used for communication between memory 701 and processor 702.

[0150] The memory 701 is used to store computer programs that can run on the processor 702.

[0151] The memory 701 may include high-speed RAM (Random Access Memory) memory, and may also include non-volatile memory, such as at least one disk storage.

[0152] If the memory 701, processor 702, and communication interface 703 are implemented independently, then the communication interface 703, memory 701, and processor 702 can be interconnected via a bus to complete communication between them. The bus can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 7 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0153] Optionally, in a specific implementation, if the memory 701, processor 702, and communication interface 703 are integrated on a single chip, then the memory 701, processor 702, and communication interface 703 can communicate with each other through an internal interface.

[0154] The processor 702 may be a CPU (Central Processing Unit), an ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of this application.

[0155] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for predicting the influence of clouds on the intensity of solar radiation on the ground.

[0156] This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method for predicting the influence of clouds on the intensity of solar radiation on the ground.

[0157] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0158] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0159] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A method for predicting the influence of clouds on the intensity of solar radiation on the ground, characterized in that, Includes the following steps: Initial ground images of the area to be observed at different times are acquired, and after geometric transformation and distortion correction operations are performed on the initial ground images, they are projected onto the ground plane to obtain actual ground images at different times. Based on the actual ground images at different times, identify at least one cloud shadow in the actual ground image at each time, obtain and mark the shape, area and darkness of the at least one cloud shadow, and determine at least one target cloud shadow in the actual ground image at each time. By comparing the positions of the same target cloud shadow in two adjacent actual ground images, the motion characteristics of the same target cloud shadow can be obtained; Based on the shape and motion characteristics of the same target cloud shadow, predict the coverage area of ​​the same target cloud shadow on the area to be observed within a preset time period in the future, and predict the influence of the cloud on the solar radiation intensity on the ground based on the coverage area and the darkness. The method of predicting the impact of clouds on ground solar radiation intensity based on the coverage area and the darkness includes: acquiring multiple three-dimensional coordinates of a photometric drone within the area of ​​at least one target cloud shadow; calculating multiple ground projection coordinates of the photometric drone in the direction of sunlight illumination based on the multiple three-dimensional coordinates; acquiring multiple actual solar radiation intensities at positions corresponding to the multiple ground projection coordinates; matching the multiple actual solar radiation intensities with the darkness of at least one target cloud shadow at the positions corresponding to the multiple ground projection coordinates based on the multiple ground projection coordinates and the multiple actual solar radiation intensities; acquiring the theoretical solar radiation intensity at the positions corresponding to the multiple ground projection coordinates; calculating the degree of reduction in ground solar radiation intensity based on the relative difference between the theoretical solar radiation intensity and the actual solar radiation intensity; establishing a correspondence between the darkness and the degree of reduction in ground solar radiation intensity; and predicting the impact of clouds on ground solar radiation intensity based on the darkness based on the correspondence.

2. The method according to claim 1, characterized in that, Before acquiring initial ground images of the area to be observed at different times, the method further includes: Obtain the speed of cloud movement and the prediction duration; Select a target photovoltaic power station and construct an observation area with the target photovoltaic power station as the center and the product of the cloud movement speed and the predicted duration as the radius.

3. The method according to claim 2, characterized in that, The identification of at least one cloud shadow in the actual ground image at each time moment includes: Obtain an initial reference ground image of the area to be observed during a preset period of clear, cloudless weather; Based on the initial reference ground image, obtain a target reference ground image at a time close to the current actual ground image; Based on the target reference ground image and the actual ground image at the current moment, at least one cloud shadow in the actual ground image at each moment is identified using the principles of visible light imaging and infrared light imaging.

4. The method according to claim 1, characterized in that, When predicting the impact of clouds on ground solar radiation intensity based on the coverage area and the darkness, the method further includes: Acquire the current sky image, current cloud image, current shooting date, current shooting time, and current camera pose of the area to be observed; The current sky image, the current cloud image, the current shooting date, the current shooting time, and the current camera pose are input into a preset neural network model to obtain the current cloud shadow distribution and the current cloud shadow shading intensity. Based on the current cloud shadow distribution and the current cloud shadow shading intensity, predict the impact of clouds on the solar radiation intensity on the ground.

5. A device for predicting the influence of clouds on the intensity of solar radiation on the ground, characterized in that, include The first acquisition module is used to acquire initial ground images of the area to be observed at different times, and to project the initial ground images onto the ground plane after performing geometric transformation and distortion correction operations to obtain actual ground images at different times. The determination module is used to identify at least one cloud shadow in the actual ground image at each time based on the actual ground images at different times, obtain the shape, area and darkness of the at least one cloud shadow and mark it, and determine at least one target cloud shadow in the actual ground image at each time. The second acquisition module is used to compare the position of the same target cloud shadow in two adjacent actual ground images to obtain the motion characteristics of the same target cloud shadow; The prediction module is used to predict the coverage area of ​​the same target cloud shadow on the observation area within a preset time period in the future, based on the shape and motion characteristics of the same target cloud shadow, and to predict the influence of the cloud on the solar radiation intensity on the ground based on the coverage area and the darkness. Specifically, the prediction module is used to: acquire multiple three-dimensional coordinates of the photometric drone within the area of ​​the at least one target cloud shadow; calculate multiple ground projection coordinates of the photometric drone in the direction of sunlight illumination based on the multiple three-dimensional coordinates; acquire multiple actual solar radiation intensities at positions corresponding to the multiple ground projection coordinates; match the multiple actual solar radiation intensities with the darkness of the at least one target cloud shadow at the positions corresponding to the multiple ground projection coordinates based on the multiple ground projection coordinates and the multiple actual solar radiation intensities; acquire the theoretical solar radiation intensity at the positions corresponding to the multiple ground projection coordinates; calculate the degree of reduction in ground solar radiation intensity based on the relative difference between the theoretical solar radiation intensity and the actual solar radiation intensity; establish a correspondence between the darkness and the degree of reduction in ground solar radiation intensity; and predict the impact of clouds on ground solar radiation intensity based on the darkness according to the correspondence.

6. The apparatus according to claim 5, characterized in that, Before acquiring initial ground images of the area to be observed at different times, the first acquisition module is further configured to: Obtain the speed of cloud movement and the prediction duration; Select a target photovoltaic power station and construct an observation area with the target photovoltaic power station as the center and the product of the cloud movement speed and the predicted duration as the radius.

7. An electronic device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the program to implement the method for predicting the influence of clouds on the intensity of solar radiation on the ground as described in any one of claims 1-4.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by a processor to implement the method for predicting the influence of clouds on the intensity of solar radiation on the ground as described in any one of claims 1-4.

9. A computer program product, characterized in that, The method includes a computer program, which, when executed by a processor, is used to implement the method for predicting the influence of clouds on the intensity of solar radiation on the ground as described in any one of claims 1-4.

Images