Weather observation system, weather observation method, method for generating a learning model, and program

Abstract

To provide a meteorological observation system, meteorological observation method, generation method of learning model and program which can estimate a solar radiation amount or power generation amount in real time.SOLUTION: A meteorological observation system 1 comprises: an image acquisition unit 11 which acquires an image G1 including at least the sky in an imaging region Ar1 acquired by a camera 20; a pixel value acquisition unit 12 which acquires a pixel value G2 of a specific region S6 in the imaging region Ar1; and an estimation unit 10 (solar radiation amount estimation part 13, power generation amount estimation part 14) which calculates an estimation value (solar radiation amount estimation value G3, power generation amount estimation value G4) being any of a solar radiation amount and a power generation amount correlated to the pixel value G2 by using correlation data (solar radiation amount conversion formula D1, power generation amount conversion formula D2) expressing a relation between a measurement value being any of the solar radiation amount and the power generation amount and the pixel value G2.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present disclosure relates to a weather observation system, a weather observation method, a method for generating a learning model, and a program.

[0002] In photovoltaic power generation, it is important to estimate the solar irradiance or the power generation amount. Patent Document 1 discloses a system for estimating solar irradiance. Patent Document 2 discloses a system for estimating the power generation amount based on the movement of clouds.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0004] The present disclosure provides a weather observation system, a weather observation method, a method for generating a learning model, and a program capable of estimating the solar irradiance or the power generation amount in real time.

Means for Solving the Problems

[0005] The weather observation system of the present disclosure includes an image acquisition unit that acquires an image including at least the sky in an imaging region acquired by a camera, a pixel value acquisition unit that acquires pixel values of a specific region in the imaging region, and an estimation unit that calculates an estimated value that is either the solar irradiance or the power generation amount correlated with the pixel values using correlation data representing the relationship between a measurement value that is either the solar irradiance or the power generation amount and the pixel values.

[0006] This configuration uses correlation data representing the relationship between pixel values ​​and measured values ​​in a specific region of an image including the sky to calculate estimated values ​​correlated with the pixel values, making it possible to estimate solar radiation or power generation in real time. [Brief explanation of the drawing]

[0007] [Figure 1] A block diagram showing the configuration of the weather observation system of the first embodiment. [Figure 2] A diagram showing an example of an image captured by a camera. [Figure 3] An explanatory diagram regarding the imaging area in an image captured by a camera. [Figure 4] A perspective view showing the structure of the camera. [Figure 5] An explanatory diagram regarding the region in an image from which pixel values ​​are calculated. [Figure 6] A graph showing the change in the average brightness value over a given day. [Figure 7] A graph showing the change in solar radiation on the same day, including the period shown in Figure 6. [Figure 8] A graph showing the change in power generation on the same day, including the period shown in Figure 6. [Figure 9] A flowchart illustrating the weather observation process performed by the weather observation system of the first embodiment. [Figure 10] A flowchart illustrating the learning model generation process performed by a weather observation system of a modified embodiment of the first embodiment. [Figure 11] A block diagram showing the configuration of a weather observation system in a modified form of the first embodiment. [Modes for carrying out the invention]

[0008] <First Embodiment> Hereinafter, a weather observation system 1 of the first embodiment of this disclosure will be described with reference to the drawings. The weather observation system 1 of the first embodiment estimates solar radiation or power generation based on images captured by a camera.

[0009] As shown in Figure 1, the weather observation system 1 includes a camera 20 for imaging targets and a computer for processing the images captured by the camera 20. The camera 20 can be any camera that can photograph targets. In the first embodiment, the camera 20 uses an all-sky camera for weather observation, and the all-sky camera is pointed vertically upward so that the imaging area includes the sky. The all-sky camera is a camera using a fisheye lens that can photograph a wide area including targets around the camera installation point with a single camera.

[0010] Figure 2 shows an example of an image captured by camera 20. Figure 3 is an explanatory diagram of the imaging area Ar1 in the image captured by camera 20. As shown in Figures 2 and 3, the imaging area Ar1 of camera 20 is rectangular, and the sky area S1 is included in the center of the imaging area Ar1. The sky area S1 includes the sun S2, clouds S3, and blue sky S4 depending on the weather conditions. Outside the sky area S1 in the imaging area Ar1, the surrounding environment S5 of camera 20 is captured. The surrounding environment S5 includes natural objects such as trees, mountains, and the ground, as well as structures such as buildings. Outside the surrounding environment S5, a part of the housing 21 of camera 20 is captured in an annular shape. As shown in Figure 4, camera 20 has a lens unit 22 provided on the top surface of the housing 21 and a transparent hemispherical dome cover 23 that covers the lens unit 22. As shown by the arrows in Figure 4, light including sunlight is reflected off the top surface (top plate) of the housing 21 and reaches the lens unit 22, causing the housing 21 to appear as a ring shape in the image captured by the camera 20. The image G1 obtained from the camera 20 has known pixels and orientations within the image. In particular, the positional relationship between the pixel position and the target is known in areas other than the empty area S1.

[0011] As shown in Figure 1, the weather observation system 1 includes an image acquisition unit 11, a pixel value acquisition unit 12, and an estimation unit 10. The estimation unit 10 includes a solar radiation estimation unit 13 and a power generation estimation unit 14. The weather observation system 1 also has a storage unit 1a, which includes a solar radiation conversion formula D1 (correlation data) and a power generation conversion formula D2 (correlation data). These units are realized through the collaborative action of software and hardware, where the processor 1b executes a program pre-stored in memory in a computer equipped with a processor 1b such as a CPU, storage 1a such as memory, and various interfaces. As an example of a hardware configuration, each component 11-18 is implemented on the cloud or server, and the camera 20, pyranometer 24, and photovoltaic power generation system 25 are configured to communicate with the cloud or server.

[0012] The image acquisition unit 11 acquires an image G1 obtained by the camera 20, in which the imaging area Ar1 includes at least the sky. The imaging area Ar1 of image G1 includes the housing 21 of the camera 20. Image G1 is a color image and includes RGB components or components that can be converted to RGB components.

[0013] The pixel value acquisition unit 12 acquires the pixel values ​​G2 of a specific region S6 within the imaging region Ar1 of the image G1 acquired by the image acquisition unit 11. The specific region S6 is at least a part of the ring-shaped housing 21 portion in the image G1. As an example, as shown in Figure 5, the average value (a value based on statistical values) of multiple pixels at the lower end P1 of the ring-shaped housing 21 portion in the image G1 is used. Furthermore, in the first embodiment, the brightness value is obtained as the pixel value G2. The brightness value can be calculated using the following formula. Brightness value = Red component (R) × 0.299 + Green component (G) × 0.587 + Blue component (B) × 0.114

[0014] The solar radiation amount estimation unit 13 (estimation unit 10) calculates an estimated solar radiation amount value G3 correlated with the pixel value G2 acquired by the pixel value acquisition unit 12 using the solar radiation amount conversion formula D1 (correlation data) stored in the storage 1a. Specifically, the solar radiation amount estimation unit 13 calculates the estimated solar radiation amount value G3 by inputting the pixel value G2 (luminance value) into the solar radiation amount conversion formula D1. The solar radiation amount conversion formula D1 can be said to be correlation data representing the relationship between the solar radiation amount and the pixel value (luminance value). As an example of the solar radiation amount conversion formula D1, for example, there is a conversion formula such as solar radiation amount = luminance value × α1 + α2. α1 and α2 are constants.

[0015] The power generation amount estimation unit 14 (estimation unit 10) calculates an estimated power generation amount value G4 correlated with the pixel value G2 acquired by the pixel value acquisition unit 12 using the power generation amount conversion formula D2 (correlation data) stored in the storage 1a. Specifically, the power generation amount estimation unit 14 calculates the estimated power generation amount value G4 by inputting the pixel value G2 (luminance value) into the power generation amount conversion formula D2. The power generation amount conversion formula D2 can be said to be correlation data representing the relationship between the power generation amount and the pixel value (luminance value). As an example of the power generation amount conversion formula D2, for example, there is a conversion formula such as power generation amount = luminance value × α3 + α4. α3 and α4 are constants.

[0016] As shown in FIG. 1, the weather observation system 1 may store and use the correlation data (solar radiation amount conversion formula D1 and power generation amount conversion formula D2) generated by another correlation data generation system in the storage 1a and may not have the function of generating correlation data. However, the weather observation system 1 of the first embodiment has a measurement value acquisition unit 15 and a correlation data generation unit 16 and can generate correlation data.

[0017] The measurement value acquisition unit 15 acquires measurement values ​​measured by a device capable of measuring either solar radiation or power generation. As shown in Figure 1, the measurement value acquisition unit 15 can acquire solar radiation measurement values ​​G5 measured by a device capable of measuring solar radiation (pyranometer 24). The measurement value acquisition unit 15 can also acquire power generation measurement values ​​G6 measured by a device capable of measuring power generation (part of the photovoltaic power generation system 25). The measurement values ​​G5 and G6 acquired by the measurement value acquisition unit 15 are stored in storage 1a as solar radiation time series data D4 and power generation time series data D5, respectively. In addition, the pixel value G2 (luminance value) acquired by the pixel value acquisition unit 12 is stored in storage 1a as pixel value time series data D3.

[0018] The correlation data generation unit 16 generates correlation data (solar radiation conversion formula D1, power generation conversion formula D2) based on the measured values ​​(solar radiation measured value G5, power generation measured value G6) acquired by the measurement value acquisition unit 15 and the pixel value G2 acquired by the pixel value acquisition unit 12. Figure 6 is a graph showing the change in the average brightness value on a given day. The elapsed time from around 8:30 is expressed in minutes. Figure 7 is a graph showing the change in solar radiation on the same day, including the period shown in Figure 6. The vertical axis represents solar radiation, and the horizontal axis represents elapsed time. Figure 8 is a graph showing the change in power generation on the same day, including the period shown in Figure 6. The vertical axis represents power generation, and the horizontal axis represents elapsed time. The period from time 0 to 180 in Figure 6 corresponds to the period from approximately 220 to approximately 580 in Figures 7 and 8. The changes in luminance values ​​shown in Figure 6, the changes in solar irradiance shown in Figure 7, and the changes in power generation shown in Figure 8 are clearly similar in shape to the human eye, indicating a high correlation between luminance values ​​and solar irradiance, and between luminance values ​​and power generation.

[0019] The correlation data generation unit 16 generates correlation data (solar radiation conversion formula D1, power generation conversion formula D2) by comparing the measured values ​​G5 and G6 with the pixel value G2 based on the image G1 captured at the same time as the measurement time of the measured values ​​G5 and G6. For example, the solar radiation measurement value G5 measured at 9:00 on December 20, 2021, is compared with the pixel value G2 (luminance value) based on the image captured at the same time. In the examples of Figures 6 to 8, examples include the comparison of the data shown in Figure 6 with the data shown in Figure 7, and the comparison of the data shown in Figure 6 with the data shown in Figure 8. The same time period here preferably refers to the same time, but it does not need to be exactly the same time. The same time period preferably refers to the same time when the camera time and the measuring instrument time are synchronized, but it is not limited to this. For example, the same time period may have a time difference of about 10 minutes, more preferably within 2 minutes, even more preferably within 1 minute, and even more preferably within 10 seconds.

[0020] Furthermore, the correlation data generation unit 16 may generate correlation data not only from a single pair of measured values ​​and pixel values, but also from multiple combinations of measured values ​​and pixel values ​​based on images captured during the same time period as the measurement time of the measured values, by comparing multiple pairs of measured values ​​and pixel values ​​within the same period. For example, in the case of Figures 6-8, if there are multiple pairs of measured values ​​and pixel values ​​as data from 0:00 to 23:00 on December 20, 2021, correlation data is generated by comparing multiple pairs of measured values ​​and pixel values ​​within the same period from 9:00 to 18:00 on the same day. This means that the comparison of measured values ​​and pixel values ​​is narrowed down to data within the same period, thereby improving the accuracy of the correlation data. In the first embodiment, a transformation formula is generated using a fitting method such as the least squares method, but the invention is not limited to this example.

[0021] As shown in Figure 1, the weather observation system 1 has a weather condition acquisition unit 17 that acquires weather conditions. Weather conditions include, for example, the percentage of clouds in the sky area S1, the season (date), and the temperature. The correlation data generation unit 16 may generate multiple correlation data based on the weather conditions. For example, the correlation data generation unit 16 may generate correlation data for cloudy skies and correlation data for clear skies. Cloudy skies mean when the percentage of clouds in the sky area S1 is above a predetermined threshold, and clear skies mean when the percentage of clouds in the sky area S1 is below a predetermined threshold. The correlation data generation unit 16 may also generate correlation data for spring, summer, autumn, and winter. The season can be classified by date; for example, December to February can be considered winter, March to May can be considered spring, June to August can be considered summer, and September to November can be considered autumn. The estimation unit 10 (solar radiation estimation unit 13, power generation estimation unit 14) can switch the correlation data used according to the weather conditions acquired by the weather condition acquisition unit 17.

[0022] In the first embodiment, the correlation data generation unit 16 makes it possible to refer to the estimated values ​​(solar radiation estimated value G3, power generation estimated value G4) estimated by the estimation unit 10 (solar radiation estimation unit 13, power generation estimation unit 14). The correlation data generation unit 16 also makes it possible to refer to the measured values ​​(solar radiation measured value G5, power generation measured value G6) acquired by the measurement value acquisition unit 15. The correlation data generation unit 16 may regenerate correlation data if there is a discrepancy of a predetermined amount or more between the estimated value calculated using correlation data and the acquired measured value. This ensures that correlation data is regenerated when an unacceptable discrepancy (difference) occurs between the estimated value and the measured value. This ensures that estimation accuracy can be maintained because correlation data is regenerated when the estimation accuracy using correlation data deteriorates due to the passage of time since the last generation of correlation data.

[0023] In the first embodiment, the correlation data generation unit 16 may generate correlation data again after a predetermined time has elapsed since the time the correlation data was last generated. This can suppress the deterioration of estimation accuracy over time and maintain estimation accuracy.

[0024] As shown in Figure 1, the weather observation system 1 may have a prediction unit 18 and an output unit 19 that outputs the prediction results of the prediction unit 18. The prediction unit 18 refers to the image G1 acquired by the image acquisition unit 11 and estimated values ​​(estimated solar radiation G3, estimated power generation G4), and predicts the amount of solar radiation or power generation for the time period after the image was taken, based on the weather information in the image G1 captured by the camera 20. The weather information in image G1 includes the position of clouds, the advection of clouds, the position of the sun, etc. The prediction unit 18 predicts the amount of solar radiation and power generation 10 minutes later, for example. For example, it determines whether the cloud flow will block the sun, and how long it will take for the clouds to move to a position where they do not block the sun after blocking it, and predicts values ​​based on the estimated values ​​(estimated solar radiation G3, estimated power generation G4). The results predicted by the prediction unit 18 may be displayed on a web screen by the output unit 19, or transmitted to a destination terminal via a distribution server. The output unit 19 may also be called a display unit or a distribution unit. The correlation data generation unit 16 may regenerate correlation data if there is a discrepancy of a predetermined amount or more between the predicted value predicted by the prediction unit 18 and the acquired measured value. This ensures that correlation data is regenerated when an unacceptable discrepancy (difference) occurs between the predicted value and the measured value. This allows for the maintenance of prediction accuracy by regenerating correlation data if the prediction accuracy of the prediction unit 18 deteriorates due to the passage of time since the last generation of correlation data and the reference of values ​​estimated using the correlation data.

[0025] [method] The information processing method performed by the above-described weather observation system 1 will be explained with reference to Figure 9. Figure 9 is a flowchart showing the weather observation processing performed by the weather observation system 1 of the first embodiment. In step ST100 shown in Figure 9, the camera 20 captures an image G1 that includes at least the sky in the imaging area Ar1. In the next step ST101, the image acquisition unit 11 acquires the image captured by the camera 20. The image captured by the camera 20 is recorded and transmitted to the server or cloud where the image acquisition unit 11 is implemented via the camera 20's communication unit. The image acquisition unit 11 receives the image. In the next step ST102, the pixel value acquisition unit 12 acquires the pixel value G2 (luminance value) within a specific area S6 of the image G1 and stores it in the storage 1a. In the next step ST103, the estimation unit 10 (solar radiation estimation unit 13, power generation estimation unit 14) acquires correlation data (solar radiation conversion formula D1, power generation conversion formula D2) that represents the relationship between the measured value, which is either solar radiation or power generation, and the pixel value G2. In the next step, ST104, the estimation unit 10 (solar radiation estimation unit 13, power generation estimation unit 14) uses correlation data to calculate estimated values ​​(solar radiation estimated value G3, power generation estimated value G4) that correlate with the pixel value G2.

[0026] <Variation> (1) In the first embodiment, the all-sky camera is installed facing upward in the vertical direction, but is not limited to this. For example, the all-sky camera may be installed horizontally along the horizontal direction to image one or more targets in a specific direction. Alternatively, a first all-sky camera facing a first direction and a second all-sky camera facing the opposite direction may be installed back to back. Furthermore, camera 20 does not have to be an all-sky camera using a fisheye lens; it may be a non-all-sky camera capable of photographing only specific directions.

[0027] (2) In the first embodiment, the luminance value is used as the pixel value G2, but it is not limited to the luminance value as long as it is a pixel value. For example, brightness may be used as the pixel value G2. Brightness is a value based on the maximum and minimum values ​​of each RGB component.

[0028] (3) The specific region S6 having multiple pixels that are the subject of the calculation of the pixel value G2 may include natural objects or objects fixed to the ground within the image G1. In the first embodiment, it is the housing 21 of the camera 20 fixed to the ground, but it is not limited to this. For example, the specific region may be the side of a building such as a building, a concrete block, or the pole part of a streetlamp. Preferably, the specific region is a region other than the sky region S1 which may include the sun S2, clouds S3, and blue sky S4.

[0029] (4) The specific region S6 in the first embodiment is the housing 21 of the camera 20, but is not limited to this, and may be an object that is directly exposed to sunlight. Furthermore, while it is preferable that it is an object that is directly exposed to sunlight, it may be an object that is not directly exposed to sunlight, for example, the north side of a building, if it is outdoors.

[0030] (5) In the first embodiment, the specific region is only the lower end P1 shown in Figure 5, but it is not limited to this, and multiple locations may be set and pixel values ​​from multiple locations may be used. In particular, in Figure 5, the first part P2 in a certain direction of the sun may have abnormally high pixel values ​​(luminance values, etc.) due to reflection from the dome cover 23 and the camera lens. Therefore, the specific region may include the lower end P1, the first part P2, the second part P3, and the third part P4, and the specific region may be arranged in multiple directions as viewed from the installation position of the camera 20. By using specific regions arranged in multiple directions, it is possible to avoid abnormal pixel values ​​caused by the sun becoming dominant, and estimation accuracy can be improved.

[0031] (6) The pixel value G2 is the average of the pixel values ​​of multiple pixels in a specific region S6, but is not limited to this, as long as it is a value based on the statistical value of the pixel values. For example, the median, mode, etc. can be used. Since it is a value based on statistical values, the adverse effects of outliers can be reduced.

[0032] (7) In the first embodiment, the correlation data is a transformation formula, but is not limited thereto. For example, it may be a table relating pixel values ​​and estimated values, or it may be a learning model generated by machine learning as shown in Figure 11 (solar radiation model D11, power generation model D12). Examples of learning models include DNN (Deep Neural Network), neural networks other than DNN, RNN (Recurrent Neural Network), LSTM (Long-short term model), CNN, SVM (Support Vector Machine), Bayesian network, linear regression, regression tree, multiple regression, random forest, ensemble, and other learning algorithms. For example, when using multiple regression with multiple types of variables, the correlation data may be represented by a transformation formula using polynomial regression.

[0033] (8) As described in (7) above, as shown in Figure 11, when the correlation data is a learning model (solar radiation model D11, power generation model D12), the correlation data generation unit 16 becomes a so-called learning unit. The learning unit may generate a learning model that outputs an estimated value which is either solar radiation or power generation when pixel value data is input, based on the training data. The training data may consist, for example, of pixel value data (pixel value time series data D3) obtained from an image G1 including the sky acquired by the camera 20, and measurement data (solar radiation time series data D4, power generation time series data D5).

[0034] (9) [Method for generating a learning model] In the case of (8) above, it can also be said that the weather observation system 1 executes the learning model generation method shown in Figure 10. Figure 10 is a flowchart of the learning model generation process executed by the weather observation system 1 of the modified embodiment of the first embodiment. In step ST200 shown in Figure 10, the camera 20 captures an image G1 that includes at least the sky in the imaging area Ar1. In the next step ST201, the image acquisition unit 11 acquires the image captured by the camera 20. The image captured by the camera 20 is recorded and transmitted to a server or cloud where the image acquisition unit 11 is implemented via the communication unit of the camera 20. The image acquisition unit 11 receives the image. In the next step ST202, the pixel value acquisition unit 12 acquires the pixel value G2 (luminance value) in a specific area S6 of the image G1 and stores it in the storage 1a. In the next step, ST203, the measurement value acquisition unit 15 acquires measurement values ​​(solar radiation measurement value G5, power generation measurement value G6) measured by a device capable of measuring either solar radiation or power generation, and stores them in storage 1a. In the next step, ST204, the correlation data generation unit 16 acquires measurement values ​​(solar radiation measurement value G5, power generation measurement value G6) measured by a device capable of measuring either solar radiation or power generation. In the next step, ST205, the correlation data generation unit 16 generates a learning model (solar radiation model D11, power generation model D12) that outputs an estimated value that is either solar radiation or power generation when pixel value data is input, based on the acquired training data.

[0035] (10) The correlation data generation unit 16 and the measurement value acquisition unit 15 in the first embodiment can be omitted if correlation data is not to be generated.

[0036] (11) The weather observation system 1 of the first embodiment is a system that estimates both solar radiation and power generation, but it may also be a system that estimates only one of either solar radiation or power generation.

[0037] As described above, although not particularly limited, the weather observation system 1 may include, as in the first embodiment and its modified versions, an image acquisition unit 11 that acquires an image G1 obtained by a camera 20 that includes at least the sky in the imaging area Ar1; a pixel value acquisition unit 12 that acquires pixel values ​​G2 of a specific area S6 within the imaging area Ar1; and an estimation unit 10 (solar radiation estimation unit 13, power generation estimation unit 14) that calculates estimated values ​​(solar radiation estimated value G3, power generation estimated value G4) that are either solar radiation or power generation correlated with the pixel value G2, using correlation data (solar radiation conversion formula D1, power generation conversion formula D2) that represent the relationship between a measured value, which is either solar radiation or power generation, and the pixel value G2. Although not particularly limited, as in the first embodiment and its modified versions, the weather observation method may involve acquiring an image G1 obtained by the camera 20, which includes at least the sky in the imaging area Ar1; acquiring pixel values ​​G2 of a specific area S6 within the imaging area Ar1; and calculating an estimated value, which is either solar radiation or power generation, that correlates with the pixel value, using correlation data (solar radiation conversion formula D1, power generation conversion formula D2) that represents the relationship between a measured value, which is either solar radiation or power generation, and the pixel value G2. By using correlation data (solar radiation conversion formula D1, power generation conversion formula D2) that represents the relationship between the pixel value G2 of a specific region S6 in image G1 including the sky and the measured value (either solar radiation measurement value G5 or power generation measurement value G6), estimated values ​​(solar radiation estimate G3, power generation estimate G4) correlated with the pixel value G2 are calculated, making it possible to estimate solar radiation or power generation in real time.

[0038] While not particularly limited, the specific region S6 may include natural objects or objects fixed to the ground within the image G1, as in the first embodiment. This is a preferred embodiment.

[0039] Although not particularly limited, the specific region S6 may include an object that is directly exposed to sunlight, as in the first embodiment. Objects that receive direct sunlight are more susceptible to changes in pixel values ​​depending on the amount of solar radiation, which can improve estimation accuracy.

[0040] Although not particularly limited, as in the modified version of the first embodiment, the specific region S6 may be arranged in multiple directions when viewed from the installation position of the camera 20. This makes it possible to reduce the effects of abnormal pixel values ​​that may occur depending on the movement of the sun.

[0041] Although not particularly limited, as in the first embodiment and its modifications, the pixel value G2 may be a brightness value or a lightness value. Brightness or luminance values ​​change in response to solar radiation, making them useful for estimating solar radiation or power generation.

[0042] Although not particularly limited, as in the first embodiment and its modifications, the pixel value G2 may be a value based on statistical values ​​in a specific region S6. This makes it possible to reduce the adverse effects caused by outliers.

[0043] Although not particularly limited, the system may further include, as in the first embodiment of the weather observation system 1, a measurement value acquisition unit 15 that acquires measurement values ​​(solar radiation measurement value G5, power generation measurement value G6) measured by a device capable of measuring either solar radiation or power generation, and a correlation data generation unit 16 that generates correlation data based on the measurement values ​​G5 and G6 acquired by the measurement value acquisition unit 15 and the pixel value G2 acquired by the pixel value acquisition unit 12. Correlation data is generated based on the pixel values ​​used for estimation and the measured values, making system operation easier.

[0044] Although not particularly limited, the system may further include a weather condition acquisition unit 17 for acquiring weather conditions, as in the first embodiment of the weather observation system 1, and the correlation data generation unit 16 may generate multiple correlation data based on the weather conditions. Since multiple correlation data sets are generated based on weather conditions, it becomes possible to use the appropriate correlation data depending on the weather conditions, thereby improving estimation accuracy.

[0045] While not particularly limited, as in the first embodiment, correlation data (solar radiation conversion formula D1, power generation conversion formula D2) may be generated by comparing measured values ​​G5 and G6 with pixel values ​​G2 based on image G1 captured during the same time period as the measurement time of the measured values. This is a preferred embodiment.

[0046] While not particularly limited, as in the first embodiment, correlation data (solar radiation conversion formula D1, power generation conversion formula D2) may be generated by comparing multiple sets of measured values ​​and pixel values ​​within the same period, from among multiple combinations of measured values ​​G5 and G6 and pixel values ​​G2 based on images G1 captured during the same time period as the measurement time of the measured values. This makes it possible to improve the accuracy of correlation data.

[0047] Although not particularly limited, as in the first embodiment, the correlation data generation unit 16 may regenerate correlation data if there is a discrepancy of a predetermined amount or more between the estimated value calculated using the correlation data and the acquired measured value. If the estimation accuracy based on correlation data deteriorates, the correlation data is regenerated, thus maintaining the estimation accuracy.

[0048] Although not particularly limited, as in the first embodiment, the correlation data generation unit 16 may generate correlation data again after a predetermined time has elapsed since the time the correlation data was previously generated. This method can suppress the deterioration of estimation accuracy over time and maintain estimation accuracy.

[0049] Although not particularly limited, the system may further include a prediction unit 18 that predicts the amount of solar radiation or power generation for the time period from the present (from the time the image was captured) based on weather information in the image G1 captured by the camera 20, as in the weather observation system 1 of the first embodiment. This will enable the provision of a system capable of predicting solar radiation or power generation for the time period from now on.

[0050] Although not particularly limited, as in the modified version of the first embodiment, the correlation data generation unit 16 may acquire training data consisting of pixel value data obtained from an image including the sky acquired by the camera 20, and measurement data measured by a device capable of measuring either solar radiation or power generation, and generate a learning model (correlation data) based on the acquired training data that outputs an estimated value which is either solar radiation or power generation when pixel value data is input.

[0051] Although not particularly limited, the weather observation system 1 may further include a camera 20, as in the first embodiment.

[0052] The program according to the first embodiment is a program that causes one or more processors to execute the above method. Although not particularly limited, the program according to the first embodiment may involve the camera 20 acquiring an image G1 in the imaging area Ar1 that includes at least the sky, acquiring the pixel values ​​of a specific area S6 within the imaging area Ar1, and having one or more processors perform a process to calculate an estimated value, which is either solar radiation or power generation, correlated with the pixel value G2, using correlation data (solar radiation conversion formula D1, power generation conversion formula D2) that represents the relationship between the measured value, which is either solar radiation or power generation, and the pixel value. While not particularly limited, a modified version of the first embodiment may involve a program that acquires training data consisting of pixel value data obtained from an image including the sky acquired by the camera 20, and measurement data measured by a device capable of measuring either solar radiation or power generation, and then causes one or more processors to execute a process to generate a learning model (correlation data) that outputs an estimated value, which is either solar radiation or power generation, when pixel value data is input, based on the acquired training data. Furthermore, the computer-readable temporary storage medium stores the above program.

[0053] Although embodiments of this disclosure have been described above with reference to the drawings, it should be understood that the specific configurations are not limited to these embodiments. The scope of this disclosure is indicated not only by the description of the embodiments above but also by the claims, and further includes all modifications within the meaning and scope equivalent to the claims.

[0054] For example, the execution order of operations, procedures, steps, and stages in the devices, systems, programs, and methods shown in the claims, specifications, and drawings can be implemented in any order, as long as the output of a previous process is not used in a later process. Even if the flow in the claims, specifications, and drawings is described using terms such as "first," "next," etc., for convenience, it does not mean that the execution must be in that order.

[0055] Each component shown in Figure 1 is implemented by executing a predetermined program on a single unit or processor, but each component may also be configured using dedicated memory or dedicated circuitry.

[0056] Although the system in the above embodiment has its components implemented on a single computer processor, the components may be distributed and implemented on multiple computers or in the cloud. In other words, the above method may be executed on one or more processors.

[0057] The structures used in each of the above embodiments can be adopted in any other embodiment. In Figure 1, each part is implemented for the sake of explanation, but some of these can be omitted as needed. [Explanation of Symbols]

[0058] 1. Weather observation system 20 cameras 10 Estimation part 11 Image acquisition unit 12 Pixel value acquisition unit 13 Solar radiation estimation section (estimation section) 14. Power generation estimation unit (estimation unit) 15 Measurement Value Acquisition Unit 16. Correlation Data Generation Unit 17. Weather Condition Acquisition Department 18 Prediction Section Ar1 imaging area D1 Solar radiation conversion formula (correlation data) D2 Power generation conversion formula (correlation data) S6 Specific area

Claims

1. An image acquisition unit that acquires an image that includes the sky in the imaging area, captured by a camera, A pixel value acquisition unit that acquires pixel values ​​for a specific region within the imaging region, An estimation unit that obtains an estimated value, which is either solar radiation or power generation, correlated with the pixel value, using correlation data that represents the relationship between a measured value, which is either solar radiation or power generation, and the pixel value. Equipped with, The aforementioned image is a photograph taken with an all-sky camera, and the specified region is the peripheral part of the image, according to the weather observation system.

2. The weather observation system according to claim 1, wherein the specified region includes natural objects or objects fixed to the ground within the image.

3. The weather observation system according to claim 1, wherein the specified area includes the housing of the camera fixed to the ground.

4. The weather observation system according to any one of claims 1 to 3, wherein the specified area is arranged in multiple directions as viewed from the camera's installation position.

5. The weather observation system according to any one of claims 1 to 4, wherein the pixel value is a brightness value or a lightness value.

6. The weather observation system according to any one of claims 1 to 5, wherein the pixel value is a value based on statistical values ​​in the specific region.

7. The correlation data is a learning model that outputs an estimated value, which is either solar radiation or power generation, when pixel value data obtained from an image captured by the camera is input. The weather observation system according to any one of claims 1 to 6, wherein the estimation unit inputs the pixel value data into the learning model and obtains an estimated value which is either solar radiation or power generation.

8. A measurement value acquisition unit that acquires the measurement value measured by a device capable of measuring either the amount of solar radiation or the amount of power generated, A correlation data generation unit generates correlation data based on the measured values ​​acquired by the measured value acquisition unit and the pixel values ​​acquired by the pixel value acquisition unit. A weather observation system according to any one of claims 1 to 7, further comprising:

9. It is further equipped with a weather condition acquisition unit that acquires weather conditions, The weather observation system according to claim 8, wherein the correlation data generation unit generates a plurality of correlation data based on the weather conditions.

10. The weather observation system according to claim 8 or 9, wherein the correlation data is generated by comparing the measured value with the pixel value based on an image taken during the same time period as the measurement time of the measured value.

11. The weather observation system according to any one of claims 8 to 10, wherein the correlation data generation unit generates the correlation data again if there is a discrepancy of a predetermined amount or more between the estimated value calculated using the correlation data and the acquired measured value.

12. The weather observation system according to any one of claims 8 to 10, wherein the correlation data generation unit generates the correlation data again after a predetermined time has elapsed since the time the correlation data was previously generated.

13. The weather observation system according to any one of claims 1 to 12, further comprising a prediction unit that predicts the amount of solar radiation or power generation for a period of time after the time the image was captured, based on weather information in the image captured by the camera.

14. The camera acquires an image that includes at least the sky in the imaging area. The pixel values ​​of a specific region within the imaging region are acquired. This includes calculating an estimated value, which is either solar radiation or power generation, that correlates with the pixel value, using correlation data that represents the relationship between a measured value, which is either solar radiation or power generation, and the pixel value. The aforementioned image is an image taken with an all-sky camera, and the aforementioned specific region is the peripheral part of the aforementioned image, in a meteorological observation method.

15. Training data is acquired consisting of pixel value data obtained from images including the sky acquired by a camera, and measurement data measured by a device capable of measuring either solar radiation or power generation. The process includes generating a learning model that outputs an estimated value, which is either solar radiation or power generation, when the pixel value data is input, based on the acquired training data. A method for generating a learning model, wherein the aforementioned image is an image taken with a panoramic camera, and the pixel value data includes the pixel values ​​of a specific region which is the peripheral part of the image.

16. The camera acquires an image that includes at least the sky in the imaging area. The pixel values ​​of a specific region within the imaging region are acquired. Using correlation data representing the relationship between measured values, which are either solar radiation or power generation, and the aforementioned pixel values, an estimated value, which is either solar radiation or power generation, that correlates with the pixel value is calculated. The process is executed by one or more processors. The program states that the aforementioned image is an image taken with a panoramic camera, and the aforementioned specific region is the peripheral portion within the image.

Images