Cloud cover forecasting device, power generation forecasting device, cloud cover forecasting method, and power generation forecasting method

Abstract

A cloud cover prediction device (10) is provided with: a learning unit (11) that takes a correlation between first time series data of a cloud cover rate derived on the basis of a sky image, which is an image obtained by photographing the sky from an observation location, and second time series data of cloud cover rates at a plurality of locations derived on the basis of a satellite image, and stores, as a time difference map, a correspondence relationship between each of the plurality of locations and a time difference between the time series data with high correlation; and a prediction unit (12) that refers to the time difference map stored in the learning unit by using a cloud movement direction acquired using the satellite image and a preset time difference, to derive a cloud cover rate at a selected location and output a predicted value of the cloud cover rate.

Description

Technical Field

[0001] The disclosed technology relates to cloud amount prediction technology.

Background Art

[0002] Cloud amount prediction technology is a technology for predicting the cloud amount related to the solar radiation amount, and is used, for example, in a power generation prediction system for predicting the power generation amount of a solar power plant or the like. Conventional power generation prediction systems are configured by dispersing a solar radiation sensor composed of a pyrheliometer and an all-sky camera for acquiring an image of the sky at intervals, and are based on the difference in solar radiation intensity by each solar radiation sensor. From the state of solar radiation at two or more points and the moving direction of clouds analyzed from the change in the cloud image, the change in the area affected by the cloud shadow is predicted. (Patent Document 1)

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0007] According to this disclosure, the cloud cover forecast results can be derived with a smaller configuration compared to conventional methods. [Brief explanation of the drawing]

[0008] [Figure 1] Figure 1 shows a basic configuration example of a cloud cover forecasting device 10 according to Embodiment 1 of this disclosure. [Figure 2] Figure 2 shows an example of the configuration of a power generation forecasting system 1A including a cloud cover forecasting device 10A according to Embodiment 1 of this disclosure. [Figure 3] Figure 3 is a flowchart showing an example of the processing of the cloud cover forecasting device 10A according to Embodiment 1 of this disclosure. [Figure 4] Figure 4 is a flowchart showing an example of the processing of the power generation prediction device 40 according to Embodiment 1 of this disclosure. [Figure 5] Figure 5 shows an example of the configuration of a power generation forecasting system 1B including a cloud cover forecasting device 10B according to Embodiment 2 of this disclosure. [Figure 6] Figure 6 is a flowchart showing a detailed example of the learning process performed by the learning unit 11B of the cloud cover forecasting device 10B according to Embodiment 2 of this disclosure. [Figure 7]Figure 7 is a flowchart showing an example of the processing of the cloud cover ratio derivation unit in the cloud cover forecasting device 10B according to Embodiment 2 of this disclosure. [Figure 8] Figure 8 is a schematic diagram illustrating the correlation calculation with point i. [Figure 9] Figure 9 shows an example of a time lag map. [Figure 10] Figure 10 is a flowchart showing a detailed example of the prediction processing performed by the prediction unit 12B of the cloud cover prediction device 10B according to Embodiment 2 of this disclosure. [Figure 11] Figure 11 shows an example of the configuration of a power generation forecasting system 1C including a cloud cover forecasting device 10C according to Embodiment 3 of this disclosure. [Figure 12] Figure 12 is a flowchart showing a detailed example of the prediction processing performed by the prediction unit 12C of the cloud cover prediction device 10C according to Embodiment 3 of this disclosure. [Figure 13] Figure 13 is a diagram showing an example configuration (first configuration example) of a cloud cover forecasting device 10D, a power generation forecasting device 40D, and a power generation forecasting system 1D including the device, according to Embodiment 4 of the present disclosure. [Figure 14] Figure 14 is a diagram showing an example configuration (second configuration example) of a cloud cover forecasting device 10D, a power generation forecasting device 40D, and a power generation forecasting system 1D including the device, according to Embodiment 4 of the present disclosure. [Figure 15] Figure 15 is a flowchart showing a detailed example of the prediction process by the solar shading rate prediction device 70D or the solar shading rate derivation unit 404D in the power generation amount prediction system 1D according to Embodiment 4 of this disclosure. [Figure 16] Figure 16 shows an example configuration (first configuration example) of a cloud cover forecasting device 10E, a power generation forecasting device 40E, and a power generation forecasting system 1E including the device, according to Embodiment 5 of this disclosure. [Figure 17] Figure 17 shows an example configuration (second configuration example) of a cloud cover forecasting device 10E, a power generation forecasting device 40E, and a power generation forecasting system 1E including the device, according to Embodiment 5 of the present disclosure. [Figure 18]FIG. 18 is a flowchart showing a detailed example of the prediction process by the prediction unit 12E of the cloud amount prediction apparatus 10E according to Embodiment 5 of the present disclosure. [Figure 19] FIG. 19 is a diagram showing a first example of a hardware configuration for realizing the functions according to the configuration of the present disclosure. [Figure 20] FIG. 20 is a diagram showing a second example of a hardware configuration for realizing the functions according to the configuration of the present disclosure. MODE FOR CARRYING OUT THE INVENTION

[0009] Hereinafter, in order to explain the present disclosure in more detail, embodiments of the present disclosure will be described with reference to the accompanying drawings.

[0010] Embodiment 1. In Embodiment 1, the basic form of the present disclosure will be described.

[0011] A configuration example of the cloud amount prediction apparatus according to Embodiment 1 of the present disclosure will be described. FIG. 1 is a diagram showing a basic configuration example of the cloud amount prediction apparatus 10 according to Embodiment 1 of the present disclosure. The cloud amount prediction apparatus 10 is an apparatus that predicts the cloud amount. The cloud amount is represented by the ratio of clouds occupying the entire sky at the observation point. In the embodiments of the present disclosure, the cloud amount is represented by the "cloud cover ratio". The cloud amount prediction apparatus 10 accumulates the correspondence relationship between the cloud cover ratio based on the sky image taken at the observation point and the cloud cover ratios of a plurality of points based on the satellite image, and the time difference based on the time difference between the point in the satellite image and the observation time at the observation point. Using the accumulated correspondence relationship and the cloud movement direction, the cloud cover ratio when a preset time has elapsed at the observation point is predicted. The "time difference setting value", which is a preset time difference, only needs to be set before executing the process of predicting the cloud cover ratio, and is received by the cloud amount prediction apparatus 10 and appropriately set before executing the prediction process. The cloud amount prediction apparatus 10 predicts the cloud cover ratio at the observation point when it is assumed that the time based on the set "time difference setting value" has elapsed. The cloud cover forecasting device 10 shown in Figure 1 is composed of a learning unit 11 and a forecasting unit 12.

[0012] The learning unit 11 has the function of executing learning processes. In the learning process, the learning unit 11 correlates a first time series of cloud occlusion rate data derived from sky images, which are images of the sky taken from an observation point, with a second time series of cloud occlusion rate data at multiple locations derived from satellite images. For each of the multiple locations, it stores the correspondence between the location and the time difference between the time series data with the highest correlation as a time difference map. Specifically, the learning unit 11 acquires sky images taken at the observation point and uses the acquired sky images to derive time-series data of cloud occlusion rate. This time-series data is also referred to as the "first time-series data" in this explanation. Furthermore, the learning unit 11 acquires satellite images taken by artificial satellites and uses the acquired satellite images to derive time-series data of cloud occlusion rates for each location at multiple points included in the satellite images. This time-series data is also referred to as the "second time-series data" in this explanation. The learning unit 11 correlates the first time series data with the second time series data and stores it as a time difference map that shows the correspondence between the time difference between time series data with high correlation to a given location for each of multiple locations.

[0013] The prediction unit 12 has the function of performing prediction processing. In the prediction process, the prediction unit 12 uses the cloud movement direction obtained using satellite images and a pre-set time difference to refer to a time difference map accumulated by the learning unit, derives the cloud occlusion rate at the selected location, and outputs a predicted cloud occlusion rate value. Specifically, the prediction unit 12 acquires satellite images and uses the acquired satellite images to derive the cloud movement direction. The cloud movement direction indicates the direction in which the clouds are moving and can be obtained by determining the movement of the clouds in images across multiple frames.

[0014] In addition to the above configuration, the cloud cover forecasting device 10 includes a control unit (not shown), a storage unit (not shown), and a communication unit (not shown). A control unit (not shown) controls the entire cloud cover forecasting device 10 and its individual components. For example, the control unit activates the cloud cover forecasting device 10 according to an external command. It also controls the state of the cloud cover forecasting device 10 (operating state = state such as startup, shutdown, or sleep). Furthermore, the control unit issues data input and data output commands to each component of the cloud cover forecasting device 10. Finally, the control unit issues commands to a storage unit (not shown) for data storage and data retrieval in response to requests from each component of the cloud cover forecasting device 10. A memory unit (not shown) stores the data used in the cloud cover forecasting device 10. For example, the memory unit stores the output (output data) from each component of the cloud cover forecasting device 10 and outputs the requested data to the requesting component. The communication unit (not shown) communicates with external devices. For example, it communicates between the cloud cover forecasting device 10 (100A) and peripheral devices (for example, a camera, satellite data distribution system, or power generation forecasting device, as described later). For example, if the cloud cover forecasting device 10 and the peripheral devices are not connected by wire, the communication unit (not shown) has the function of communicating between the cloud cover forecasting device 10 and the peripheral devices. The communication unit (not shown) also has the function of communicating with an external device, such as a server device. The control unit (not shown), storage unit (not shown), and communication unit (not shown) are the same in the embodiments described later.

[0015] Next, we will describe an example of the configuration of a power generation forecasting system that includes a cloud cover forecasting device, when the cloud cover forecasting device according to this embodiment is applied to the power generation forecasting system. Figure 2 shows an example of the configuration of a power generation forecasting system 1A including a cloud cover forecasting device 10A according to Embodiment 1 of this disclosure. The power generation forecasting system 1A(1) is a system that includes multiple devices for forecasting the amount of electricity generated by solar power generation. The power generation forecasting system 1A is, for example, a ground-based system installed on the ground. The power generation forecasting system 1A shown in Figure 2 comprises a cloud cover forecasting device 10A(10), a camera 20, a satellite data distribution system 30, and a power generation forecasting device 40.

[0016] Camera 20 is an imaging device installed at the observation site. Camera 20 acquires a sky image, which is a hemispherical image of the sky centered on the zenith at the observation site. If visible light is used for the observation wavelength of Camera 20, it measures reflected sunlight, and clouds will appear white. Camera 20 outputs the sky image to the cloud cover forecasting device 10A.

[0017] The satellite data distribution system 30 is a system that distributes satellite images acquired from weather satellites, etc. Weather satellites include, for example, Himawari 8 / 9. When the weather satellite is Himawari 8 / 9, the satellite data distribution system 30 can distribute images with a visible band and a ground spatial resolution of 0.5 to 1.0 km. The satellite data distribution system 30 outputs satellite images to the cloud cover forecasting device 10A.

[0018] The cloud cover forecasting device 10A predicts the cloud occlusion rate, similar to the cloud cover forecasting device 10 described earlier. The cloud cover forecasting device 10A shown in Figure 2 is communicated with the camera 20 and the satellite data distribution system 30. The cloud cover forecasting device 10A acquires sky images from the camera 20. The cloud cover forecasting device 10A acquires satellite images from the satellite data distribution system 30. The cloud cover forecasting device 10A shown in Figure 2 is composed of a learning unit 11A and a forecasting unit 12A.

[0019] The learning unit 11A acquires images of the sky taken at the observation site from the camera 20. The learning unit 11A acquires satellite images from the satellite data distribution system 30. The learning unit 11A is configured to have the function of executing learning processes, similar to the learning unit 11 described earlier.

[0020] The forecasting unit 12A acquires satellite images from the satellite data distribution system 30. The prediction unit 12A is configured to have the function of performing prediction processing, similar to the prediction unit 12 described earlier. The forecasting unit 12A outputs the cloud shading rate forecast value to the power generation forecasting device 40.

[0021] The power generation forecasting device 40 forecasts the amount of power generated using the cloud occlusion rate. The power generation forecasting device 40 outputs the forecasted power generation value, which is the forecast result. The power generation forecasting device 40 shown in Figure 2 forecasts power generation by converting the cloud cover rate forecast value output by the cloud cover forecasting device 10A into a power generation amount, and outputs the forecasted power generation amount.

[0022] In this explanation, the cloud cover forecasting device 10A and the power generation forecasting device 40 are shown as separate devices, but they may be integrated into a single device.

[0023] Next, we will explain an example of processing by the cloud cover forecasting system. The cloud cover forecasting device 10 shown in Figure 1 and the cloud cover forecasting device 10A shown in Figure 2 differ in that the source of the information used in the processing can be clearly indicated in the cloud cover forecasting device 10A shown in Figure 2, and the output destination can also be clearly indicated. However, since the internal processing is the same for both, a representative example of the processing of the cloud cover forecasting device 10A will be explained here. Figure 3 is a flowchart showing an example of the processing of the cloud cover forecasting device 10A according to Embodiment 1 of this disclosure. The process shown in Figure 3 is a cloud cover prediction method using a cloud cover prediction device. This cloud cover prediction method may be included in the power generation prediction method using a power generation prediction device. For example, by having a computer execute this cloud cover prediction method using a program, the computer can function as a cloud cover prediction device. The cloud cover forecasting device 10A shown in Figure 2 starts the process shown in Figure 3 when it receives information from an external or internal control unit (not shown). Specifically, for example, when the cloud cover forecasting device 10A receives a cloud cover forecasting command or a power generation forecasting command from an external source, it starts the process shown in Figure 3. (Start)

[0024] The cloud cover forecasting device 10A then performs a learning process. (Step ST1110) During the learning process, the learning unit 11A of the cloud cover forecasting device 10A performs the learning process. In the learning process, the learning unit 11 correlates a first time series of cloud occlusion rate data derived from sky images, which are images of the sky taken from an observation point, with a second time series of cloud occlusion rate data at multiple locations derived from satellite images. For each of the multiple locations, it stores the correspondence between the location and the time difference between the time series data with the highest correlation as a time difference map. Specifically, the learning unit 11 acquires sky images taken at the observation point and uses the acquired sky images to derive time-series data of cloud occlusion rate. This time-series data is also referred to as the "first time-series data" in this explanation. Furthermore, the learning unit 11 acquires satellite images taken by artificial satellites and uses the acquired satellite images to derive time-series data of cloud occlusion rates for each location at multiple points included in the satellite images. This time-series data is also referred to as the "second time-series data" in this explanation. The learning unit 11 correlates the first time series data with the second time series data and stores the correspondence between locations at multiple points and the time differences with high correlation as a time difference map.

[0025] The cloud cover forecasting device 10A then performs forecasting processing. (Step ST1120) In the forecasting process, the forecasting unit 12A of the cloud cover forecasting device 10A uses the cloud movement direction acquired using satellite images and a preset time difference to refer to a time difference map accumulated by the learning unit, derive the cloud occlusion rate at the selected location, and output a cloud occlusion rate forecast value.

[0026] The cloud cover forecasting device 10A then performs forecast result output processing. (Step ST1130) In the forecast result output processing, the forecast unit 12A of the cloud cover forecasting device 10A outputs the cloud occlusion rate forecast value to the power generation forecasting device 40.

[0027] In the cloud cover forecasting device 10A, when the forecasting unit 12A outputs the cloud occlusion rate, which is the forecast result, in the forecast result output processing of step ST1130, it then executes the termination determination process (step ST1140 "Termination?"). In the termination determination process, a control unit (not shown) of the cloud cover forecasting device 10A determines whether to terminate the processing of the cloud cover forecasting device 10A. The control unit (not shown) determines whether to terminate the processing of the cloud cover forecasting device 10A according to an external termination command or execution program. If the control unit (not shown) determines that the cloud cover forecasting device 10A has not finished processing (step ST1140 "Finish?" "NO"), the process returns to step ST1110 and continues. If the control unit (not shown) determines that the cloud cover forecasting device 10A has finished processing (step ST1140 "Finish?" "YES"), the cloud cover forecasting device 10A finishes the processing shown in Figure 3.

[0028] Next, we will explain an example of the processing performed by the power generation forecasting device. Figure 4 is a flowchart showing an example of the processing of the power generation prediction device 40 according to Embodiment 1 of this disclosure. The process shown in Figure 4 is a method for predicting power generation using a power generation prediction device. For example, by having a computer execute this power generation prediction method using a program, the computer can function as a power generation prediction device. The power generation forecasting device 40 shown in Figure 2 starts the process shown in Figure 4 when it receives information from an external or internal control unit (not shown). Specifically, for example, when the power generation forecasting device 40 receives a power generation forecasting command from an external source, it starts the process shown in Figure 4. (Start)

[0029] The power generation forecasting device 40 then performs the cloud cover forecasting result acquisition process. (Step ST1210) In the cloud cover forecast result acquisition process, the power generation forecast device 40 acquires the cloud cover rate forecast value, which is the forecast result output by the cloud cover forecast device 10A.

[0030] The power generation forecasting device 40 then performs cloud cover forecast result input processing. (Step ST1220) In the cloud cover forecast result input processing, the power generation forecasting device 40 inputs, for example, the acquired cloud occlusion rate forecast value into the learning model.

[0031] The power generation forecasting device 40 then performs power generation forecasting processing. (Step ST1230) In the power generation forecasting process, the power generation forecasting device 40 acquires, for example, the power generation forecast value output from the learning model.

[0032] The power generation forecasting device 40 then performs forecast result output processing. (Step ST1240) In the prediction result output processing, the power generation prediction device 40 outputs the acquired power generation prediction value.

[0033] After the power generation forecasting device 40 executes the forecast result output processing in step ST1240, it then executes the termination determination processing (step ST1250 "Termination?"). In the termination determination process, a control unit (not shown) of the power generation forecasting device 40 determines whether to terminate the processing of the power generation forecasting device 40. The control unit (not shown) determines whether to terminate the processing of the power generation forecasting device 40 according to an external termination command or execution program. If the control unit (not shown) determines that it will not terminate the processing of the power generation prediction device 40 (step ST1250 "NO"), the process returns to step ST1210 and continues. If the control unit (not shown) determines that the power generation prediction device 40 has finished processing (step ST1250 "YES"), the power generation prediction device 40 finishes the processing shown in Figure 4. (end)

[0034] This embodiment shows an example of the following configuration. A learning unit correlates a first time-series data of cloud occlusion rate derived from sky images, which are images of the sky taken from an observation point, with a second time-series data of cloud occlusion rate at multiple locations derived from satellite images, and stores the correspondence between each location and the time difference between the time-series data with the highest correlation as a time-difference map for each of the multiple locations. The prediction unit uses the cloud movement direction obtained from satellite imagery and a pre-set time difference to derive the cloud occlusion rate at a selected location by referring to a time difference map accumulated by the learning unit, and outputs a predicted cloud occlusion rate value. A cloud cover forecasting device characterized by being equipped with the following features. This disclosure has the effect of providing a cloud cover forecasting device that enables the deriving of cloud cover forecast results with a smaller configuration than conventional devices. Furthermore, a power generation forecasting device that includes the cloud cover forecasting device configuration will produce the same effect as described above.

[0035] This embodiment shows an example of the following configuration. The cloud cover prediction value output by the cloud cover prediction device is input to a learning model that predicts power generation using the cloud cover rate as input, and the power generation prediction value output by the learning model is obtained. A power generation forecasting device characterized by the following features. This disclosure provides a power generation forecasting device that enables obtaining power generation forecasts using cloud cover forecasts derived with a smaller configuration than conventional methods.

[0036] This embodiment shows an example of the following configuration. A learning unit that correlates a first time series of cloud occlusion rate data derived from sky images (images of the sky taken from an observation point) with a second time series of cloud occlusion rate data from multiple locations derived from satellite images, and accumulates a time difference map showing the correspondence between each location and the time difference between the time series data with the highest correlation for each of the multiple locations. The prediction unit uses the cloud movement direction obtained using satellite imagery and the pre-accepted time difference to derive the cloud occlusion rate at a selected location by referring to a time difference map accumulated by the learning unit, and outputs a predicted cloud occlusion rate value. A power generation prediction value acquisition unit inputs the cloud occlusion rate prediction value into a learning model that predicts power generation using the cloud occlusion rate as input, and acquires the power generation prediction value output by the learning model. Equipped with, A power generation forecasting device characterized by the following features. This disclosure has the effect of providing a power generation forecasting device that enables the deriving of cloud cover forecast results with a smaller configuration than conventional devices.

[0037] This embodiment discloses an example configuration as follows: A cloud cover forecasting method using a cloud cover forecasting device, The learning unit of the cloud cover prediction device, The first time-series data of cloud occlusion derived from sky images, which are images of the sky taken from an observation point, and the second time-series data of cloud occlusion at multiple locations, derived from satellite images, are correlated, and for each of the multiple locations, the correspondence between the location and the time difference between the time-series data with high correlation is accumulated as a time difference map. The prediction unit of the cloud cover prediction device, Using the cloud movement direction obtained from satellite images and a preset time difference, the cloud occlusion rate at a selected location is derived by referring to the time difference map accumulated by the learning unit, and a predicted cloud occlusion rate value is output. A cloud cover forecasting method characterized by the following features. This disclosure has the effect of providing a cloud cover forecasting method that enables the deriving of cloud cover forecast results with a smaller configuration than conventional methods. Furthermore, a power generation forecasting method that includes a cloud cover forecasting method will have the same effect as described above.

[0038] This embodiment discloses an example configuration as follows: Computers, A learning unit correlates a first time-series data of cloud occlusion rate derived from sky images, which are images of the sky taken from an observation point, with a second time-series data of cloud occlusion rate at multiple locations derived from satellite images, and stores the correspondence between each location and the time difference between the time-series data with the highest correlation as a time-difference map for each of the multiple locations. The prediction unit uses the cloud movement direction obtained from satellite imagery and a pre-set time difference to derive the cloud occlusion rate at a selected location by referring to a time difference map accumulated by the learning unit, and outputs a predicted cloud occlusion rate value. A program characterized by operating as a cloud cover forecasting device equipped with the necessary features. This disclosure has the effect of providing a program that enables the deriving of cloud cover forecast results with a smaller configuration than conventional methods.

[0039] This embodiment shows an example of the following configuration. A method for predicting power generation using a power generation prediction device, The learning unit of the power generation prediction device, We correlate time-series data of cloud occlusion rates derived from sky images (images of the sky taken from observation points) with time-series data of cloud occlusion rates at multiple locations derived from satellite images, and accumulate the correspondence between satellite image locations and time differences where the correlation is high as a time-difference map. The prediction unit of the power generation prediction device, Using the cloud movement direction obtained from satellite images and a pre-set time difference, the learning unit references a time difference map to derive the cloud occlusion rate at a selected location and outputs a predicted cloud occlusion rate value. The power generation forecasting device's power generation forecasting value acquisition unit, The predicted cloud occlusion rate output by the prediction unit is input to a learning model that predicts power generation using the cloud occlusion rate as input, and the predicted power generation rate output by the learning model is obtained. A method for predicting power generation, characterized by the following features. This disclosure provides a method for predicting power generation that enables obtaining predicted power generation results using cloud cover prediction results derived with a smaller configuration than conventional methods.

[0040] Embodiment 2. Embodiment 2 describes a more detailed configuration example of the configuration according to Embodiment 1. In Embodiment 2, components of Embodiment 2 that are the same as those of Embodiment 1 already described are given the same names and the same reference numerals (with some modifications), and redundant explanations are omitted as appropriate.

[0041] This document describes an example of the configuration of a cloud cover forecasting device and a system including the device according to Embodiment 2 of this disclosure. Figure 5 shows an example of the configuration of a power generation forecasting system 1B including a cloud cover forecasting device 10B according to Embodiment 2 of this disclosure. The power generation forecasting system 1B shown in Figure 5 comprises a cloud cover forecasting device 10B(10), a camera 20, a satellite data distribution system 30, and a power generation forecasting device 40. Camera 20 and satellite data distribution system 30 are the same as those already described, so a detailed explanation will be omitted.

[0042] The cloud cover forecasting device 10B predicts the cloud occlusion rate, similar to the cloud cover forecasting device 10A described earlier. The cloud cover forecasting device 10B shown in Figure 5 is composed of a learning unit 11B(11) and a forecasting unit 12B(12).

[0043] The learning unit 11B is configured to have the function of executing a learning process, similar to the learning unit 11A which has already been described. The learning unit 11B shown in Figure 5 includes a camera image storage unit 101B (101), a first cloud occlusion rate derivation unit 102B (102), a first time series data generation unit 103B (103), a second cloud occlusion rate derivation unit 202B (202), a second time series data generation unit 203B (203), a correlation calculation unit 301B (301), a time difference map generation unit 302B (302), and a time difference map storage unit 303B (303).

[0044] The camera image storage unit 101B stores an empty image. The camera image storage unit 101B stores the image signals output by the camera 20 in chronological order.

[0045] The first cloud occlusion rate derivation unit 102B derives the cloud occlusion rate based on a sky image, which is an image of the sky taken from the observation point. Specifically, for example, the first cloud occlusion rate derivation unit 102B derives the cloud occlusion rate for the area of ​​interest in the sky image stored in the camera image storage unit 101B.

[0046] The first time-series data generation unit 103B generates time-series data of cloud occlusion rates derived based on sky images. Specifically, for example, the first time-series data generation unit 103B converts the cloud occlusion rate derived from the sky image into time-series data for a certain time range. This time range is, for example, from sunrise to sunset, and is a pre-set time range. This time-series data can be referred to in this explanation as the "first time-series data" or the "camera cloud occlusion rate time-series data." The first time-series data generation unit 103B outputs the first time-series data (camera cloud occlusion rate time-series data) to the correlation calculation unit 301B.

[0047] The second cloud occlusion rate derivation unit 202B derives the cloud occlusion rate at multiple locations based on satellite images. Specifically, for example, the second cloud occlusion rate derivation unit 202B derives cloud occlusion rates for a range of area in the satellite image that is roughly the same size as the area of ​​interest, for multiple points in a range that is correlated with the area of ​​interest from which the cloud occlusion rate of the sky image was derived. This ensures that the range of cloud occlusion rates derived from the satellite image and the range of the area of ​​interest in the sky image are not unrelated to each other.

[0048] The second time-series data generation unit 203B generates time-series data of cloud occlusion rates at multiple locations derived from satellite images. Specifically, for example, the second time-series data generation unit 203B converts the cloud occlusion rate derived from satellite images into time-series data for a certain time range. This time range is the same as the time range of the first time-series data (camera cloud occlusion rate time-series data) that has already been described. This time-series data can be referred to in this explanation as the "second time-series data" or the "satellite cloud occlusion rate time-series data." The second time-series data generation unit 203B outputs the second time-series data (satellite cloud occlusion rate time-series data) to the correlation calculation unit 301B.

[0049] The correlation calculation unit 301B performs correlation calculations between the first time series data (camera cloud occlusion rate time series data) and the second time series data (satellite cloud occlusion rate time series data). Specifically, for example, the correlation calculation unit 301B performs a correlation calculation between a first time series of cloud occlusion rate data derived based on sky images, which are images of the sky taken from an observation point, and a second time series of cloud occlusion rate data derived for multiple locations using satellite images. The correlation calculation unit 301B calculates, for example, the cross-correlation coefficient R(τ) between the first time series data and the second time series data.

[0050] The time difference map generation unit 302B determines the time difference between time series data with high correlation for each location at multiple points included in the satellite image, based on the results of the correlation calculation performed by the correlation calculation unit 301B. Specifically, for example, the time difference map generation unit 302B finds the time difference τ=τ0 at which the cross-correlation coefficient R(τ) calculated by the correlation calculation unit 301B is maximized, and generates a time difference map that shows the correspondence between the time range used in the calculation (e.g., date), the location in the satellite image, and the time difference τ0 (maximum time difference). The time difference map generation unit 302B stores the generated time difference map in the time difference map storage unit 303B.

[0051] The time difference map storage unit 303B stores the time difference maps generated by the time difference map generation unit 302B. Specifically, the time difference map storage unit 303B stores a time difference map that shows the correspondence between time series data highly correlated with a given location for each of multiple locations. As a result, the time difference map storage unit 303B stores the correspondence between time ranges (e.g., dates), satellite image locations, and time differences τ0 (maximum time difference).

[0052] The prediction unit 12B is configured to have the function of performing prediction processing, similar to the prediction unit 12A described earlier. The forecasting unit 12B shown in Figure 5 is composed of a cloud movement direction derivation unit 401B (401), a location selection unit 402B (402), and a third cloud occlusion rate derivation unit 403B (403).

[0053] The cloud movement direction derivation unit 401B uses the received image to derive the cloud movement direction, which is the direction in which the clouds moved. Specifically, for example, the cloud movement direction deriving unit 401B derives the cloud movement direction from the current multiple frames of satellite images received from the satellite data distribution system 30.

[0054] The location selection unit 402B selects a location by referring to a time difference map and using the cloud movement direction and a preset time difference. The pre-set time difference is a time difference setting value that indicates the time difference between the current time and the time at which the user wants to predict the cloud cover rate. The time difference setting value indicates the time difference between the current time and a time in the near future, for example, 15 minutes from the current time. Specifically, for example, the location selection unit 402B refers to a time-difference map, determines the direction of the location using the direction of cloud movement, and further determines the distance from the observation point using the time difference (pre-set time difference) that the user wants to predict the amount of power generated, thereby selecting a location from among multiple locations on the time-difference map.

[0055] The third cloud occlusion rate derivation unit 403B derives the cloud occlusion rate at the location selected by the location selection unit 402B based on the current cloud image. Specifically, for example, the third cloud occlusion rate derivation unit 403B acquires the current cloud image and uses the current cloud image to derive the cloud occlusion rate at the selected location. The current cloud image in this embodiment is a satellite image output by a satellite data distribution system 30 that distributes data from artificial satellites. In this case, the satellite data distribution system 30 can be described as a means for acquiring cloud images. The third cloud occlusion rate derivation unit 403B outputs the derived cloud occlusion rate to the power generation forecasting device 40. The cloud occlusion rate output by the third cloud occlusion rate derivation unit 403B of the cloud cover forecasting device 10B can be expressed as the cloud occlusion rate forecast value.

[0056] The power generation forecasting device 40 forecasts the amount of power generated using the cloud occlusion rate. The power generation forecasting device 40 outputs the forecasted power generation value, which is the forecast result. The power generation forecasting device 40 shown in Figure 5 forecasts power generation by converting the cloud cover rate forecast value output by the cloud cover forecasting device 10B into a power generation amount, and outputs the forecasted power generation amount. The power generation prediction device 40 shown in Figure 5 is composed of a power generation prediction value acquisition unit 41 and a learning model unit 42.

[0057] The power generation forecast acquisition unit 41 acquires the cloud occlusion rate (cloud occlusion rate forecast value) output by the cloud cover forecasting device 10B and inputs the acquired cloud occlusion rate to the learning model unit 42. The power generation forecast acquisition unit 41 acquires the power generation output by the learning model unit 42 and outputs the acquired power generation. In other words, the power generation forecast acquisition unit 41 acquires a power generation forecast value using the cloud occlusion rate forecast value.

[0058] The learning model unit 42 includes a learning model that has been pre-trained and constructed to take the cloud occlusion rate as input and output the amount of power generated. In other words, the learning model of the learning model unit 42 has been pre-trained and constructed to associate the cloud occlusion rate with the amount of power generated. The amount of power generated output by the learning model is an estimated value of the amount of power generated. In other words, when the learning model unit 42 receives a predicted cloud occlusion rate value from the power generation predicted value acquisition unit 41, it converts the predicted cloud occlusion rate value into a predicted power generation rate value and outputs it.

[0059] Next, an example of processing by the learning unit 11B in the cloud cover forecasting device 10B according to Embodiment 2 of this disclosure will be described. Figure 6 is a flowchart showing a detailed example of the learning process performed by the learning unit 11B of the cloud cover forecasting device 10B according to Embodiment 2 of this disclosure. The process shown in Figure 6 is included in the cloud cover prediction method using a cloud cover prediction device. The cloud cover prediction method may also be included in the power generation prediction method using a power generation prediction device. For example, by having a computer execute this cloud cover prediction method using a program, the computer can function as a cloud cover prediction device. The learning unit 11B of the cloud cover forecasting device 10B shown in Figure 5 starts the process shown in Figure 6 when it receives information from an external or internal control unit (not shown). Specifically, for example, the learning unit 11B of the cloud cover forecasting device 10A starts the process shown in Figure 6 when it receives a cloud cover forecasting command or a power generation forecasting command from an external source. (Start)

[0060] The learning unit 11B then executes the process of retrieving an image from the camera image storage unit 101B. (Step ST2301) In this process, the first cloud occlusion rate derivation unit 102B of the learning unit 11B retrieves multiple frames of camera images (sky images) within a certain time range from the camera image storage unit 101B.

[0061] The learning unit 11B then performs a process to derive the cloud occlusion rate. (Step ST2302) In this process, the first cloud occlusion rate derivation unit 102B of the learning unit 11B derives the cloud occlusion rate based on a sky image, which is an image of the sky taken from the observation point. The first cloud occlusion rate derivation unit 102B derives the cloud occlusion rate for each sky image received. The first cloud occlusion rate derivation unit 102B outputs the cloud occlusion rate to the first time-series data generation unit 103B.

[0062] Here, we will explain the derivation process used to derive the cloud occlusion rate. Figure 7 is a flowchart showing an example of the processing of the cloud cover ratio derivation unit in the cloud cover forecasting device 10B according to Embodiment 2 of this disclosure. The process shown in Figure 7 is a cloud cover prediction method using a cloud cover prediction device. The cloud cover prediction method may be included in the power generation prediction method using a power generation prediction device. For example, by having a computer execute this cloud cover prediction method using a program, the computer can function as a cloud cover prediction device. The cloud occlusion rate derivation units (102B, 202B) shown in Figure 5 start the processing shown in Figure 7 when they receive a command from a control unit (not shown) of the cloud cover prediction device 10B, learning unit 11B, or prediction unit 12B. Specifically, for example, when the cloud occlusion rate derivation units (102B, 202B) receive a cloud occlusion rate derivation command and a camera image or satellite image, they start the processing shown in Figure 7. (Start)

[0063] The cloud occlusion rate derivation unit then performs a process to divide the image into grid-like regions. (Step ST2401)

[0064] The cloud occlusion rate derivation unit then performs the process of taking the average value of the pixel values ​​in each region. (Step ST2402)

[0065] The cloud occlusion rate derivation unit then performs a process of binarizing the image at a certain threshold. (Step ST2403)

[0066] The cloud occlusion rate derivation unit then performs a process to calculate the cloud occlusion rate by dividing the sum of the binarized values ​​by the area. (Step ST2404) In this process, the cloud occlusion rate derivation unit takes the sum of the binarized values ​​within the region of interest and divides it by the number of regions of interest as the cloud occlusion rate.

[0067] The cloud occlusion rate derivation unit then completes the process shown in Figure 7. (End)

[0068] Returning to the explanation of Figure 6. The learning unit 11B then executes the process of generating time-series data (= first time-series data (camera cloud occlusion rate time-series data) "Ac_cam(t)"). (Step ST2303) In this process, the first time-series data generation unit 103B of the learning unit 11B organizes the cloud occlusion rate as time-series data. The first time-series data generation unit 103B outputs the first time-series data (camera cloud occlusion rate time-series data) to the correlation calculation unit 301B.

[0069] The learning unit 11B executes the processes from step ST2304 to step ST2306 in parallel with the processes from step ST2301 to step ST2303. The learning unit 11B performs the process of receiving satellite images. (Step ST2304) In this process, the second cloud occlusion rate derivation unit 202B of the learning unit 11B acquires multiple frames of satellite images from the satellite data distribution system 30, covering a time range similar to that extracted in the process of step ST2301.

[0070] The learning unit 11B then performs a process to derive the cloud occlusion rate. (Step ST2305) In this process, the second cloud occlusion rate derivation unit 202B of the learning unit 11B derives the cloud occlusion rate for multiple locations using satellite images. Specifically, the second cloud occlusion rate derivation unit 202B derives the cloud occlusion rate according to the flowchart shown in Figure 7. However, since the spatial resolution of satellite images is expected to be lower than that of camera images, in this case, the second cloud occlusion rate derivation unit 202B omits the process in step ST2402 if there is only one pixel in the grid-like region that divides the image in the process of step ST2401.

[0071] The learning unit 11B then executes the process of generating time series data (= second time series data (satellite cloud occlusion rate time series data) "Ac_sat_i(t)"). (Step ST2306) In this process, the second time-series data generation unit 203B of the learning unit 11B organizes the cloud occlusion rate as time-series data. The time-series data generated from the cloud occlusion rate at point i is denoted as Ac_sat_i(t). The second time-series data generation unit 203B outputs the second time-series data (satellite cloud occlusion rate time-series data) to the correlation calculation unit 301B.

[0072] The learning unit 11B executes the processes from step ST2301 to step ST2303, and the processes from step ST2304 to step ST2306, respectively, and then executes the correlation process. (Step ST2307) Figure 8 is a schematic diagram illustrating the correlation calculation with point i. Figure 8 shows an image of the processing in steps ST2307 and ST2308. In this process, the correlation calculation unit 301B of the learning unit 11B calculates the cross-correlation coefficient R(τ) of the camera cloud occlusion rate time series data Ac_cam(t) and the satellite cloud occlusion rate time series data Ac_sat_i(t) based on equation (1). TIFF0007884701000001.tif26166 Here, μ cam , μ sat_i These represent the average value of the camera cloud occlusion rate time series data and the average value of the satellite cloud occlusion rate time series data at point i, respectively, and are expressed by equations (2) and (3). TIFF0007884701000002.tif59166

[0073] The learning unit 11B then performs the process of finding the time difference τ0 that maximizes the cross-correlation coefficient. (Step ST2308) In this process, the time difference map generation unit 302B of the learning unit 11B finds the time difference τ0 that maximizes the cross-correlation coefficient R(τ) received from the correlation calculation unit 301B.

[0074] The learning unit 11B then performs the process of adding to the time lag map. (Step ST2309) In this process, the time difference map generation unit 302B of the learning unit 11B adds the time difference τ0 to the time difference map along with the location and the time range (e.g., date) used in the calculation.

[0075] The learning unit 11B then executes the process of storing the time difference map in the storage unit. (Step ST2310) In this process, the time difference map generation unit 302B of the learning unit 11B adds new data to the time difference map stored in the time difference map storage unit 303B and stores it there.

[0076] The learning unit 11B then performs a process to determine whether calculations have been performed at all locations. (Step ST2311) In this process, the control unit (not shown) of the learning unit 11B determines whether the time difference map generation unit 302B has generated time difference maps for all predetermined locations.

[0077] If the control unit (not shown) of the learning unit 11B determines that there are incomplete points (step ST2311 "Calculate at all points?" "NO"), it proceeds to the process in step S2312. The learning unit 11B then executes the location change process. (Step ST2312) In the location change process, the learning unit 11B changes the location and returns to the process in step S2305.

[0078] The control unit (not shown) of the learning unit 11B terminates the process shown in Figure 6 when it determines that calculations have been performed at all locations (step ST2311 "Calculation performed at all locations?" "YES"). (end)

[0079] Figure 9 shows an example of a time lag map. Time Difference Map 1000 maps time differences between time-series data for each time range, such as date, and for each location.

[0080] Figure 10 is a flowchart showing a detailed example of the prediction processing performed by the prediction unit 12B of the cloud cover prediction device 10B according to Embodiment 2 of this disclosure. The process shown in Figure 10 is a cloud cover prediction method using a cloud cover prediction device. The cloud cover prediction method may be included in the power generation prediction method using a power generation prediction device. For example, by having a computer execute this cloud cover prediction method using a program, the computer can function as a cloud cover prediction device. The cloud cover forecasting device 10B shown in Figure 5 starts the process shown in Figure 10 when it receives information from an external or internal control unit (not shown). Specifically, for example, when the cloud cover forecasting device 10B receives a cloud cover forecasting command or a power generation forecasting command from an external source, it starts the process shown in Figure 10. (Start)

[0081] The prediction unit 12B then performs the process of receiving satellite images. (Step ST2501) In this process, the cloud movement direction deriving unit 401B of the forecasting unit 12B receives multiple frames of images, including images taken at a time close to the current time, from the satellite data distribution system 30.

[0082] The forecasting unit 12B then performs a process to derive the direction of cloud movement. (Step ST2502) In this process, the cloud movement direction derivation unit 401B of the prediction unit 12B derives the cloud movement direction from the received multiple frames of images. The cloud movement direction derivation unit 401B derives the cloud movement direction using the current multiple frames of satellite images.

[0083] The prediction unit 12B then performs the process of selecting a location. (Step ST2503) In this process, the location selection unit 402B of the prediction unit 12B selects a location by referring to a time difference map and using the cloud movement direction and a preset time difference. Specifically, the location selection unit 402B compares the derived current cloud movement direction with the time difference map stored in the learning unit 11B and selects a location that has a high correlation with the cloud occlusion rate at the time difference to be predicted (for example, 15 minutes later).

[0084] The forecasting unit 12B then performs a process to derive the cloud occlusion rate for the selected location. (Step ST2504) In this process, the third cloud occlusion rate derivation unit 403B of the prediction unit 12B acquires the current cloud image and derives the cloud occlusion rate at the selected location using the current cloud image. Specifically, for example, the third cloud occlusion rate derivation unit 403B uses the satellite image taken at the time closest to the current time from among the satellite images received in the process of step S2501 to derive the cloud occlusion rate for the location selected by the location selection unit 402B and output it to the power generation forecasting device 40. The process for deriving the cloud occlusion rate is the same as the process described in Figure 7, so the explanation is omitted.

[0085] When the third cloud shading rate derivation unit 403B outputs the cloud shading rate to the power generation forecasting device 40, the forecasting unit 12B terminates the process shown in Figure 10. (end)

[0086] As described above, by performing correlation processing using a statistical quantity called the cloud occlusion rate without using individual cloud mass information from image data, the computational load is reduced, and the system can be built at low cost. Furthermore, by correlating the cloud occlusion rate, which is directly related to the location and the amount of power generated, it is possible to determine the time difference, which contributes to improving the accuracy of power generation prediction.

[0087] This embodiment further discloses the following configuration example. The aforementioned learning unit, A correlation calculation unit performs a correlation calculation between the first time series data, which is time series data of cloud occlusion rate derived based on sky images, which are images of the sky taken from an observation point, and the second time series data, which is time series data of cloud occlusion rate derived for multiple points using satellite images. A time difference map generation unit that uses the results of the correlation calculation to determine the time difference between time series data with high correlation for each location at multiple points included in the satellite image, A time difference map storage unit stores a time difference map that shows the correspondence between time series data highly correlated with a location at multiple locations, and Equipped with, The aforementioned prediction unit, A cloud movement direction derivation unit that derives the cloud movement direction using current satellite images from multiple frames, A location selection unit that selects a location using the cloud movement direction and a preset time difference by referring to the aforementioned time difference map, A cloud occlusion rate derivation unit that acquires a current cloud image and uses the current cloud image to derive the cloud occlusion rate at the selected location, Equipped with, A cloud cover forecasting device characterized by the following features. As a result, this disclosure also has the effect of providing a cloud cover forecasting device that enables the deriving of cloud cover forecast results with a smaller configuration than conventional devices. Furthermore, by applying the above configuration to the system, the cloud cover forecasting method, or the program described above, the same effects as described above can be achieved. Furthermore, this disclosure allows the power generation forecasting device to achieve the same effects as described above by including the cloud cover forecasting device configuration in the power generation forecasting device.

[0088] This embodiment further discloses the following configuration example. The current cloud image mentioned above is This is a satellite image output by a satellite data distribution system that distributes data from artificial satellites. A cloud cover forecasting device characterized by the following features. As a result, this disclosure can also provide a cloud cover prediction device that can use current cloud images in conjunction with the device, enabling the deriving of cloud cover prediction results with a smaller configuration compared to conventional methods. Furthermore, by applying the above configuration to the system, the cloud cover forecasting method, or the program described above, the same effects as described above can be achieved. Furthermore, this disclosure allows the power generation forecasting device to achieve the same effects as described above by including the cloud cover forecasting device configuration in the power generation forecasting device.

[0089] This embodiment further discloses the following configuration example. The cloud cover prediction value output by the cloud cover prediction device is input to a learning model that predicts power generation using the cloud cover rate as input, and the power generation prediction value output by the learning model is obtained. A power generation forecasting device characterized by the following features. As a result, this disclosure also has the effect of providing a power generation forecasting device that enables obtaining power generation forecasts using cloud cover forecasts derived with a smaller configuration than conventional devices. Furthermore, by applying the above configuration to the system, the power generation forecasting method, or the program described above, the same effects as described above can be achieved.

[0090] Embodiment 3. In the above second embodiment, we described a system in which satellite images received from a satellite data distribution system are used as input to the third cloud occlusion rate derivation unit of the prediction section of the cloud cover prediction device. In this embodiment, we show an embodiment in which the cloud occlusion rate is determined by taking as input the image from camera i installed at location i selected by the location selection unit from among cameras distributed in various locations. In Embodiment 3, components related to Embodiment 3 that are the same as those related to Embodiment 1 or Embodiment 2 already described will be given the same names and the same reference numerals (with some modifications), and redundant explanations will be omitted as appropriate.

[0091] This document describes an example of the configuration of a cloud cover forecasting device and a system including the device according to Embodiment 3 of this disclosure. Figure 11 shows an example of the configuration of a power generation forecasting system 1C including a cloud cover forecasting device 10C according to Embodiment 3 of this disclosure. The power generation forecasting system 1C(1) shown in Figure 11 comprises a cloud cover forecasting device 10C(10), a camera 20, a satellite data distribution system 30, a power generation forecasting device 40, and a camera i50 (camera i50 will also be referred to as "camera(i)50" in this description). Camera 20 is configured to have the same functions as the camera 20 already described. The satellite data distribution system 30 is configured to have the functions of the satellite data distribution system 30 as described above.

[0092] Camera (i) 50 is installed at each location different from the observation site. At the observation site, it acquires an image of the sky (sky image) of a hemisphere centered on the zenith and outputs it to the cloud cover forecasting device 10C.

[0093] The cloud cover forecasting device 10C shown in Figure 11 is composed of a learning unit 11C(11) and a forecasting unit 12C(12).

[0094] The learning unit 11C is configured to have the same functions as the learning unit 11B, which has already been described. The learning unit 11C shown in Figure 11 includes a camera image storage unit 101C (101), a first cloud occlusion rate derivation unit 102C (102), a first time-series data generation unit 103C (103), a second cloud occlusion rate derivation unit 202C (202), a second time-series data generation unit 203C (203), a correlation calculation unit 301C (301), a time-difference map generation unit 302C (302), and a time-difference map storage unit 303C (303). Since this configuration is the same as that described earlier with the same names and similar codes (with some code changes), a detailed explanation is omitted.

[0095] The forecasting unit 12C shown in Figure 11 is composed of a cloud movement direction derivation unit 401C (401), a location selection unit 402C (402), and a third cloud occlusion rate derivation unit 403C (403).

[0096] The cloud movement direction derivation unit 401C, like the cloud movement direction derivation unit 401B described earlier, is configured to derive the cloud movement direction, which is the direction in which the clouds moved, using the received image.

[0097] The location selection unit 402C selects a location by referring to a time difference map and using the cloud movement direction and a preset time difference. The pre-set time difference is a time difference setting value that indicates the time difference between the current time and the time at which the user wants to predict the cloud cover rate. The time difference setting value indicates the time difference between the current time and a time in the near future, for example, 15 minutes from the current time. Specifically, for example, the location selection unit 402C refers to a time-difference map, determines the direction of a location using the direction of cloud movement, and further determines the distance from the observation point using the time difference (a pre-set time difference) that the user wants to predict the amount of power generated, thereby selecting a location from among multiple locations on the time-difference map. The location selection unit 402C instructs the camera (i) 50 installed at the selected location (i) to output the captured cloud image to the cloud cover forecasting device 10C.

[0098] The third cloud occlusion rate derivation unit 403C of the cloud cover forecasting device 10C derives the cloud occlusion rate at a selected location from the acquired current cloud image. The current cloud images were taken by camera (i)50 installed at multiple locations different from the observation site. In this case, camera (i) 50 can be described as a means for acquiring cloud images.

[0099] The power generation forecasting device 40 is configured to function in the same way as the power generation forecasting device 40 described earlier.

[0100] Next, an example of the processing of the cloud cover forecasting device according to Embodiment 3 of this disclosure will be described. Figure 12 is a flowchart showing a detailed example of the prediction processing performed by the prediction unit 12C of the cloud cover prediction device 10C according to Embodiment 3 of this disclosure. The process shown in Figure 12 is a cloud cover prediction method using a cloud cover prediction device. The cloud cover prediction method may be included in the power generation prediction method using a power generation prediction device. For example, by having a computer execute this cloud cover prediction method using a program, the computer can function as a cloud cover prediction device. The cloud cover forecasting device 10A shown in Figure 11 starts the process shown in Figure 12 when it receives a command from an external or internal control unit (not shown). Specifically, for example, when the cloud cover forecasting device 10A receives a cloud cover forecasting command or a power generation forecasting command from an external source, it starts the process shown in Figure 12. (Start) Of the processes shown in Figure 12, steps ST3501, ST3502, and ST3503 are the same as steps ST2501, ST2502, and ST2503 in the flowchart of Figure 10, which have already been explained, so their explanation will be omitted.

[0101] The prediction unit 12C then performs the process of receiving the camera image of the selected location. (Step ST3504) In this process, the location selection unit 402C of the forecasting unit 12C instructs the camera (i) 50 installed at the selected location (i) to output the captured cloud image to the cloud cover forecasting device 10C. Camera (i) 50 outputs the captured cloud image to the cloud cover forecasting device 10C according to the command from the location selection unit 402C. In the cloud cover forecasting device 10C, the third cloud occlusion rate derivation unit 403C of the forecasting unit 12C acquires the current cloud image (camera image) using the camera (i) 50.

[0102] The prediction unit 12C then performs a process to derive the cloud occlusion rate of the camera image. (Step ST3505) In this process, the third cloud occlusion rate derivation unit 403C of the prediction unit 12C derives the cloud occlusion rate and outputs a predicted cloud occlusion rate value, similar to the process shown in the flowchart of Figure 7. Derive the cloud occlusion rate of the current cloud image (camera image). In this process, the third cloud occlusion rate derivation unit 403C of the forecasting unit 12B derives the cloud occlusion rate for the location selected by the location selection unit 402C and outputs it to the power generation forecasting device 40. The process for deriving the cloud occlusion rate is the same as the process described in Figure 7, so the explanation is omitted.

[0103] When the third cloud shading rate derivation unit 403C outputs the cloud shading rate to the power generation forecasting device 40, the forecasting unit 12C terminates the process shown in Figure 12. (End)

[0104] With the configuration according to this embodiment, the selected location and the cloud occlusion rate correspond more closely, thus improving the accuracy of the cloud cover forecast.

[0105] This embodiment further discloses the following configuration example. The current cloud image mentioned above is These are images captured by cameras installed at multiple locations different from the aforementioned observation point. A cloud cover forecasting device characterized by the following features. As a result, this disclosure provides a cloud cover forecasting device that enables the deriving of cloud cover forecast results with a smaller configuration than conventional devices, and also improves the accuracy of the cloud cover forecast results. Furthermore, by applying the above configuration to the system, the cloud cover forecasting method, or the program described above, the same effects as described above can be achieved. Furthermore, this disclosure allows the power generation forecasting device to achieve the same effects as described above by including the cloud cover forecasting device configuration in the power generation forecasting device.

[0106] Embodiment 4. In embodiments 2 and 3 described above, the power generation prediction device was described in which only the third cloud occlusion rate derivation unit of the prediction section of the cloud cover prediction device was used as input. In this embodiment, an embodiment is shown in which the amount of power generated is corrected by inputting the solar occlusion rate at the observation position at a time difference after the predicted time difference, which is predicted from the cloud thickness at the location selected by the location selection unit, into the power generation prediction device. In Embodiment 4, components related to Embodiment 4 that are the same as those related to Embodiment 1, Embodiment 2, or Embodiment 3 already described are given the same names and the same reference numerals (with some modifications), and redundant explanations are omitted as appropriate. This document describes an example configuration of a cloud cover forecasting device, a power generation forecasting device, and a system including such devices according to Embodiment 4 of this disclosure. Figure 13 is a diagram showing an example configuration (first configuration example) of a cloud cover forecasting device 10D, a power generation forecasting device 40D, and a power generation forecasting system 1D including the device, according to Embodiment 4 of the present disclosure. Figure 14 is a diagram showing an example configuration (second configuration example) of a cloud cover forecasting device 10D, a power generation forecasting device 40D, and a power generation forecasting system 1D including the device, according to Embodiment 4 of the present disclosure. The power generation forecasting system 1D(1) shown in Figure 13 comprises a cloud cover forecasting device 10D(10), a camera 20, a satellite data distribution system 30, a power generation forecasting device 40D(40), a cloud altitude data distribution system 60, and a solar shading rate forecasting device 70D(70). Camera 20 and satellite data distribution system 30 are the same as those already described, so a detailed explanation will be omitted.

[0107] The cloud cover forecasting device 10D shown in Figure 13 is composed of a learning unit 11D(11) and a forecasting unit 12D(12).

[0108] The learning unit 11D is configured to have the same functions as the learning unit 11B or learning unit 11C described earlier. Since this is the same configuration as those already described with the same name and similar codes (with some modifications), a detailed explanation will be omitted.

[0109] The forecasting unit 12D shown in Figure 13 includes a cloud movement direction derivation unit 401D (401), a location selection unit 402D (402), and a third cloud occlusion rate derivation unit 403D (403).

[0110] The cloud movement direction derivation unit 401D of the forecast unit 12D is configured to derive the cloud movement direction, which is the direction in which the clouds moved, using the received image, similar to the cloud movement direction derivation units 401B or 401C already described.

[0111] The location selection unit 402D of the forecasting unit 12D selects a location using the cloud movement direction and a preset time difference by referring to a time difference map, and outputs the result to the third cloud shading rate derivation unit 403D and the solar shading rate forecasting device 70D. When applied to the configuration according to Embodiment 3, the location selection unit 402D instructs the camera (i) installed at the selected location (i) to output the captured cloud image to the cloud cover forecasting device 10C.

[0112] The third cloud occlusion rate derivation unit 403D of the forecasting unit 12D derives the cloud occlusion rate at a selected location from the acquired current cloud image. The third cloud shading rate derivation unit 403D outputs the derived cloud shading rate to the power generation forecasting device 40D.

[0113] The power generation forecasting system 1D shown in Figures 13 and 14 is configured to obtain the degree of sunlight shading from the thickness of the clouds and to predict the amount of power generation using the degree of sunlight shading. The thickness of clouds can be derived using known techniques, such as the following:

[0114] For example, there is a technique that calculates cloud base altitude by subtracting the cloud base altitude, obtained from a cloud base altimeter installed on the ground, from the cloud top altitude, which is obtained using infrared images from a satellite and ground temperature. (For example, see Reference (Patent Document 2)) References (Patent Document 2): Japanese Unexamined Patent Publication No. 2022-042380

[0115] Furthermore, there are techniques for obtaining the three-dimensional or vertical distribution of clouds using weather radar mounted on flying objects such as artificial satellites. (For example, see reference (Patent Document 3)) References (Patent Document 3): Japanese Unexamined Patent Publication No. 2002-350537

[0116] The cloud altitude data distribution system 60 acquires data (cloud altitude data) that shows the altitude at which clouds exist at locations different from the observation point, and outputs it to the solar shading rate prediction device 70D. Cloud altitude data specifically refers to: For example, as disclosed in Patent Document 2, cloud top altitude data obtained from infrared images acquired by an artificial satellite and ground temperature, and cloud base altitude data obtained by a cloud base altimeter, Alternatively, for example, data on the three-dimensional or vertical distribution of clouds obtained using weather radar from an aerial object, as disclosed in Patent Document 3, That is the case.

[0117] The solar shading rate prediction device 70D derives a predicted solar shading rate value that indicates the degree to which sunlight will be blocked at a given location. The solar shading rate prediction device 70D uses cloud height data for the location selected by the prediction unit 12D to calculate the cloud thickness, converts it into a solar shading rate, and outputs it to the power generation prediction device 40D. Specifically, the solar shading rate prediction device 70D derives the cloud thickness at a location based on the cloud altitude at that location selected by the prediction unit 12D (location selection unit 402D of the prediction unit 12D), and uses the cloud thickness to derive the predicted solar shading rate. The solar shading rate prediction device 70D (or solar shading rate derivation unit 404D) outputs the solar shading rate derived based on the cloud altitude received from the cloud altitude data distribution system 60 for the location selected by the prediction unit 12D. The functions of the solar shading rate forecasting device 70D may be configured to be included in the cloud cover forecasting device 10D or the power generation forecasting device 40D. If the functions of the solar shading rate prediction device 70D are included in the cloud cover prediction device 10D, then, for example, as shown in Figure 14, the cloud cover prediction device 10D further includes a solar shading rate derivation unit 404D. If the functions of the solar shading rate prediction device 70D are included in the power generation prediction device 40D, then the power generation prediction device 40D further includes a solar shading rate derivation unit 404D.

[0118] In the power generation forecasting device 40D, the power generation forecast value acquisition unit 41 acquires the power generation forecast value output by the learning model by inputting the input data for the learning model derived based on the cloud shading rate forecast value and the solar shading rate forecast value into the learning model which predicts power generation using at least the input data for the learning model based on the cloud shading rate as input. Specifically, the power generation forecasting device 40D obtains the power generation forecast output by the learning model, for example, by inputting the corrected cloud shading rate forecast value, which has been corrected using the solar shading rate forecast value, into the learning model that predicts power generation using the cloud shading rate as input. More specifically, the power generation forecasting device 40D takes the cloud shading rate output from the forecasting unit 12D of the cloud cover forecasting device 10D and the solar shading rate output by the solar shading rate forecasting device 70D as input and corrects the cloud shading rate by multiplying the cloud shading rate by the solar shading rate. Otherwise, it is configured to function in the same way as the power generation forecasting device 40 already described.

[0119] Next, an example of processing in a power generation forecasting system 1D including a cloud cover forecasting device 10D, a power generation forecasting device 40D, or a power generation forecasting system 1D including a power generation forecasting device 40D according to Embodiment 4 of this disclosure will be described. In the power generation forecasting system 1D, the power generation forecasting device 40D further executes processes including a process to acquire the solar shading rate (solar shading rate acquisition process) and a process to correct the cloud cover forecast result (cloud cover forecast result correction process). In the solar shading rate acquisition process, the power generation forecast value acquisition unit 41 of the power generation forecasting device 40D acquires the solar shading rate from the solar shading rate forecasting device 70D (or solar shading rate derivation unit 404D). In the cloud cover forecast result correction process, the power generation forecast value acquisition unit 41 of the power generation forecasting device 40D derives input data for the learning model based on the cloud cover forecast value and the solar shading rate forecast value. For example, it corrects the cloud cover forecast value using the solar cover forecast value and outputs the corrected cloud cover forecast value as input data for the learning model. The power generation forecast value acquisition unit 41 inputs the learning model by outputting the input data for the learning model to the learning model unit 42. The above process is executed, for example, between the cloud cover forecast result acquisition process (step ST1210 shown in Figure 4) and the cloud cover forecast result input process (step ST1220 shown in Figure 4), as previously explained.

[0120] Here, an example of the process for deriving the solar shielding rate in the configuration according to Embodiment 4 will be described. Figure 15 is a flowchart showing a detailed example of the prediction process by the solar shading rate prediction device 70D or the solar shading rate derivation unit 404D in the power generation amount prediction system 1D according to Embodiment 4 of this disclosure. The process shown in Figure 15 is a method for predicting the solar shading rate using a solar shading rate prediction device. The solar shading rate prediction method (or the cloud thickness prediction method included in the solar shading rate prediction method) may also be included in a cloud cover prediction method using a cloud cover prediction device or a power generation prediction method using a power generation prediction device. For example, by having a computer execute a cloud cover prediction method or power generation prediction method, including this solar shading rate prediction method, through a program, the computer can function as a cloud cover prediction device or power generation prediction device. The solar shading rate prediction device 70D (or solar shading rate derivation unit 404D) shown in Figure 15 starts the process shown in Figure 15 upon receiving a command from the location selection unit 402D. (Start)

[0121] The solar shading rate prediction device 70D (or solar shading rate derivation unit 404D) then receives the selected location from the location selection unit 402D. (Step ST4601)

[0122] The solar shading rate prediction device 70D (or solar shading rate derivation unit 404D) then receives cloud altitude data near the selected location from the cloud altitude data distribution system 60. (Step ST4602) In this process, the solar shading rate prediction device 70D (or solar shading rate derivation unit 404D) instructs the cloud altitude data distribution system 60 to output the current cloud altitude data near the selected location to the solar shading rate prediction device 70D (or solar shading rate derivation unit 404D). The cloud altitude data distribution system 60 outputs current cloud altitude data near the selected location to the solar shading rate prediction device 70D (or solar shading rate derivation unit 404D) in accordance with a command from the solar shading rate prediction device 70D (or solar shading rate derivation unit 404D). The solar shading rate prediction device 70D (or solar shading rate derivation unit 404D) acquires current cloud altitude data in the vicinity of the selected location from the cloud altitude data distribution system 60.

[0123] The solar shading rate prediction device 70D (or solar shading rate derivation unit 404D) then performs a process to calculate the cloud thickness (cloud thickness calculation process). (Step ST4603) In this process, the cloud thickness (cloud thickness value) is calculated from the cloud altitude data. The calculation method involves subtracting the cloud base altitude from the cloud top altitude, which is obtained from cloud altitude data, to determine the cloud thickness. If the vertical distribution of the clouds is available, any gaps in the clouds at altitudes between the cloud top and cloud base altitudes are taken into account when calculating the thickness.

[0124] The solar shading rate prediction device 70D (or solar shading rate derivation unit 404D) then derives the solar shading rate. (Step ST4604) In this process, the solar shading rate prediction device 70D (or solar shading rate derivation unit 404D) obtains the solar shading rate by multiplying the cloud thickness by a predetermined coefficient. The predetermined coefficient is set in advance and registered, for example, based on the relationship between the cloud thickness and the degree to which the cloud blocks sunlight. The predetermined coefficient is stored, for example, in a memory unit (not shown).

[0125] The solar shading rate prediction device 70D (or solar shading rate derivation unit 404D) outputs the solar shading rate to the power generation prediction device 40, and then terminates the process shown in Figure 15. (End)

[0126] As described above, by adding cloud thickness information, it becomes possible to provide a power generation forecasting device that can improve the accuracy of power generation forecasts by multiplying the cloud cover forecast by a correction coefficient based on the solar shading rate.

[0127] This embodiment further discloses the following configuration example. The system further includes a solar shielding rate derivation unit that derives a predicted solar shielding rate value indicating the degree to which sunlight will be shielded based on the cloud altitude at the location selected by the prediction unit, The aforementioned power generation forecast value acquisition unit is: The input data for the learning model derived based on the predicted cloud shading rate and the predicted solar shading rate is input to the learning model that predicts power generation, using at least the input data for the learning model based on the cloud shading rate as input, thereby obtaining the predicted power generation output by the learning model. A power generation forecasting device characterized by the following features. Alternatively, the power generation forecast value acquisition unit may be: The corrected cloud shielding rate prediction value, which has been corrected using the aforementioned solar shielding rate prediction value, is input to the learning model that predicts power generation using the cloud shielding rate as input, thereby obtaining the power generation prediction value output by the learning model. As a result, this disclosure can provide a power generation forecasting device that enables improved accuracy in predicting power generation. Furthermore, by applying the above configuration to a system, a power generation forecasting method, or the above program, the same effects as described above can be achieved.

[0128] Embodiment 5. In the embodiments described above, the cloud movement direction was assumed to be limited to one direction. In this embodiment, an embodiment is shown in which the cloud occlusion rate and the sunlight occlusion rate are corrected by deriving the cloud movement direction at different altitudes using the cloud altitude data and wind direction data at different altitudes acquired in Embodiment 4. In Embodiment 5, components related to Embodiment 5 that are the same as those related to Embodiment 1, Embodiment 2, Embodiment 3, or Embodiment 4 already described are given the same names and the same reference numerals (with some modifications), and redundant explanations are omitted as appropriate.

[0129] This document describes an example configuration of a cloud cover forecasting device, a power generation forecasting device, and a system including such devices according to Embodiment 5 of this disclosure. Figure 16 shows an example configuration (first configuration example) of a cloud cover forecasting device 10E, a power generation forecasting device 40E, and a power generation forecasting system 1E including the device, according to Embodiment 5 of this disclosure. Figure 17 shows an example configuration (second configuration example) of a cloud cover forecasting device 10E, a power generation forecasting device 40E, and a power generation forecasting system 1E including the device, according to Embodiment 5 of the present disclosure. Figures 16 and 17 show an example configuration of power generation forecasting system 1E, which further includes a configuration for predicting the solar shading rate. The power generation forecasting system according to this embodiment is configured to derive one or more cloud movement directions based on the wind direction at each altitude obtained by the altitude-specific wind direction data distribution system and the cloud altitude received from the cloud altitude data distribution system, and to forecast the cloud occlusion rate and solar radiation occlusion rate at each point. The power generation forecasting system 1E(1) shown in Figure 17 comprises a cloud cover forecasting device 10E(10), a camera 20, a satellite data distribution system 30, a power generation forecasting device 40E(40), a cloud altitude data distribution system 60, a solar shading rate forecasting device 70E(70), and an altitude-specific wind direction distribution system 80. Camera 20, satellite data distribution system 30, and cloud altitude data distribution system 60 are the same as those already described, so a detailed explanation will be omitted.

[0130] The altitude-specific wind direction distribution system 80 is a system that distributes altitude-specific wind direction measurement data acquired from a Doppler lidar or the like. The altitude-specific wind direction distribution system 80 outputs the altitude-specific wind direction measurement data to the cloud cover forecasting device 10E.

[0131] The cloud cover forecasting device 10E shown in Figure 17 is composed of a learning unit 11E(11) and a forecasting unit 12E(12).

[0132] The learning unit 11E is configured to have the same functions as the learning units 11B, 11C, or 11D that have already been described. The configuration example of the learning unit 11E is the same as the configuration with the same name and similar codes (with some modifications) that has already been described, so a detailed explanation will be omitted.

[0133] The forecasting unit 12E further uses the cloud altitude at one or more selected locations, and the wind direction at each altitude, to derive a predicted cloud occlusion rate. The forecasting unit 12E shown in Figure 17 includes a cloud movement direction derivation unit 401E (401), a location selection unit 402E (402), and a third cloud occlusion rate derivation unit 403E (403).

[0134] The cloud movement direction derivation unit 401E of the forecasting unit 12E derives the cloud movement direction for each altitude using cloud altitude data acquired from the cloud altitude data distribution system 60 and altitude-specific wind direction data acquired from the altitude-specific wind direction distribution system 80. The cloud movement direction derivation unit 401E outputs cloud altitude data and cloud movement directions for each altitude to the location selection unit 402E.

[0135] The location selection unit 402E of the forecasting unit 12E selects a location by referring to a time difference map and using the cloud movement direction for each altitude and a preset time difference, and outputs the selected location to the third cloud occlusion rate derivation unit 403E. The selected location, cloud altitude data, and cloud movement direction for each altitude are output to the solar occlusion rate forecasting device 70E. When applied to the configuration according to Embodiment 3, the location selection unit 402E instructs the camera (i) installed at the selected location (i) to output the captured cloud image to the cloud cover forecasting device 10E.

[0136] The third cloud occlusion rate derivation unit 403E of the forecasting unit 12E derives the cloud occlusion rate at one or more selected locations from the acquired current cloud image. The third cloud shading rate derivation unit 403E outputs the derived cloud shading rate to the power generation forecasting device 40E.

[0137] The solar shading rate prediction device 70E derives a predicted solar shading rate value by using the wind direction at each altitude, in addition to the cloud altitude at the location selected by the prediction unit 12E. The solar shading rate forecasting device 70E uses altitude-specific wind direction data and cloud height data from one or more selected locations to calculate the cloud thickness at each altitude, convert it into a predicted solar shading rate, and output it to the power generation forecasting device 40E. The functions of the solar shading rate forecasting device 70E may be configured to be included in the cloud cover forecasting device 10E or the power generation forecasting device 40E. If the functions of the solar shading rate prediction device 70E are included in the cloud cover prediction device 10E, then, for example, as shown in Figure 17, the cloud cover prediction device 10E further includes a solar shading rate derivation unit 404E. If the functions of the solar shading rate prediction device 70E are included in the power generation prediction device 40E, then the power generation prediction device 40E further includes a solar shading rate derivation unit 404E.

[0138] The power generation forecasting device 40E (power generation forecasting value acquisition unit 41 of the power generation forecasting device 40E) acquires the power generation forecasting value output by the learning model by inputting input data for the learning model derived based on the cloud shading rate forecasting value and the solar shading rate forecasting value at one or more locations into the learning model that predicts power generation using at least the input data for the learning model based on the cloud shading rate as input. The power generation forecasting device 40E (power generation forecasting value acquisition unit 41 of the power generation forecasting device 40E) acquires the power generation forecasting value output by the learning model by inputting the corrected cloud shading rate forecasting value, which has been corrected using the solar shading rate forecasting value, to the learning model that predicts power generation using the cloud shading rate as input, for each of one or more locations. In this configuration, the power generation forecasting device 40E corrects the power generation by multiplying the solar shading rate output by the solar shading rate forecasting device 70E for each of one or more selected locations by the output of the third cloud shading rate derivation unit 403E and summing the values ​​to use as a correction coefficient. Otherwise, it is configured to function in the same way as the power generation forecasting device 40 already described.

[0139] Next, an example of the processing of the cloud cover forecasting device according to Embodiment 5 of this disclosure will be described. Figure 18 is a flowchart showing a detailed example of the prediction processing performed by the prediction unit 12E of the cloud cover prediction device 10E according to Embodiment 5 of this disclosure. The process shown in Figure 18 is a cloud cover prediction method using a cloud cover prediction device. The cloud cover prediction method may be included in the power generation prediction method using a power generation prediction device. For example, by having a computer execute this cloud cover forecasting method or power generation forecasting method using a program, the computer can be made to function as a cloud cover forecasting device or a power generation forecasting device. The cloud cover forecasting device 10E shown in Figure 17 starts the process shown in Figure 18 when it receives a command from an external or internal control unit (not shown). Specifically, for example, when the cloud cover forecasting device 10E receives a cloud cover forecasting command or a power generation forecasting command from an external source, it starts the process shown in Figure 18. (Start)

[0140] The forecasting unit 12E then performs the process of receiving wind direction data by altitude. (Step ST5701) In this process, the cloud movement direction deriving unit 401E of the forecasting unit 12E receives wind direction data for each altitude from the altitude-specific wind direction distribution system 80.

[0141] The forecasting unit 12E then performs the process of receiving cloud altitude data. (Step ST5702) In this process, the cloud movement direction deriving unit 401E of the forecasting unit 12E receives cloud altitude data from the cloud altitude data distribution system 60.

[0142] The forecasting unit 12E then performs a process to derive the direction of cloud movement at each altitude. (Step ST5703) In this process, the cloud movement direction derivation unit 401E of the forecasting unit 12E derives the cloud movement direction at each altitude from the received altitude-specific wind direction data and cloud altitude data.

[0143] The prediction unit 12E then performs the process of selecting one or more locations. (Step ST5704) In this process, the location selection unit 402E of the prediction unit 12E selects one or more locations by referring to a time difference map and using the cloud movement direction at each altitude and a preset time difference. Specifically, the location selection unit 402E compares the derived current cloud movement direction at each altitude with the time difference map stored in the learning unit 11E and selects locations that have a high correlation with the cloud occlusion rate at the time difference to be predicted (for example, 15 minutes later). The location selection unit 402E outputs information indicating the selected location to the third cloud occlusion rate derivation unit 403E. The location selection unit 402E of the forecasting unit 12E outputs information indicating the selected location, altitude-specific wind direction data at the selected location, and cloud altitude data at the selected location to the solar shading rate forecasting device 70E (or solar shading rate derivation unit 404E). This output data is used in the process of deriving the solar shading rate, which will be described later.

[0144] The prediction unit 12E then performs the process of receiving satellite images. (Step ST5705) In this process, the third cloud occlusion rate derivation unit 403E of the prediction unit 12E receives multiple frames of images, including images taken at a time close to the current time, from the satellite data distribution system 30.

[0145] The forecasting unit 12E then performs a process to derive the cloud occlusion rate for the selected location. (Step ST5706) In this process, the third cloud occlusion rate derivation unit 403E of the forecasting unit 12E uses the satellite image received in the process of step ST5705 to derive the cloud occlusion rate for the location selected by the location selection unit 402E and output it to the power generation forecasting device 40E. The process for deriving the cloud occlusion rate is the same as the process described in Figure 7, so the explanation is omitted.

[0146] If the control unit (not shown) of the prediction unit 12E determines that there are incomplete points (step ST5707 "Calculate at all points?" "NO"), it proceeds to the process in step ST5708. The prediction unit 12E then performs a location change process. (Step ST5708) In the location change process, the prediction unit 12E changes the location and returns to the process of step ST5706.

[0147] The control unit (not shown) of the prediction unit 12E terminates the process shown in Figure 18 when it determines that calculations have been performed at all locations (step ST5707 "Calculation performed at all locations?" "YES"). (end)

[0148] When the third cloud shading rate derivation unit 403E outputs the cloud shading rate to the power generation forecasting device 40E, the forecasting unit 12E terminates the process shown in Figure 18. (end)

[0149] Next, an example of the process for deriving the solar shielding rate in the configuration according to Embodiment 5 will be described. In the configuration according to Embodiment 5, the process for deriving the solar shielding rate is the same as the process for deriving the solar shielding rate already described (see the process shown in Figure 15), but in addition to the cloud altitude, the solar shielding rate is further derived based on the wind direction at different altitudes.

[0150] When the solar shading rate prediction device 70E (or solar shading rate derivation unit 404E) starts the process of predicting the solar shading rate, it then receives the selected location from the location selection unit 402E. (This corresponds to step ST4601 shown in Figure 15.) In this process, the solar shading rate prediction device 70E (or solar shading rate derivation unit 404E) receives information indicating the selected location, wind direction data by altitude at the selected location, and cloud altitude data at the selected location from the location selection unit 402E. If multiple locations are selected by the location selection unit 402E, the device receives a set of information indicating the location, wind direction data by altitude, and cloud altitude data for each selected location from the location selection unit 402E. Furthermore, cloud altitude data may be received from the cloud altitude data distribution system 60, similar to the processing described in the previously explained embodiment. (See step ST4602 shown in Figure 15.)

[0151] The solar shading rate prediction device 70E (or solar shading rate derivation unit 404E) then performs a process to calculate the cloud thickness (cloud thickness calculation process). In this process, cloud thickness is calculated from cloud altitude data. (This corresponds to step ST4603 shown in Figure 15.) In the cloud thickness calculation process according to this embodiment, the solar shading rate prediction device 70E (or solar shading rate derivation unit 404E) calculates the cloud thickness (cloud thickness value) using information indicating the location selected by the location selection unit 402E, altitude-specific wind direction data at the selected location, and cloud altitude data at the selected location. This makes it possible to derive the cloud thickness by considering, for example, the movement of clouds according to the wind direction at each altitude.

[0152] The solar shading rate prediction device 70E (or solar shading rate derivation unit 404E) then performs a process to derive the solar shading rate. The process for deriving the solar shading rate is the same as the process for deriving the solar shading rate already explained, so the explanation will be omitted. (See the process in step ST4604 shown in Figure 15.) The solar shading rate prediction device 70E (or solar shading rate derivation unit 404E) outputs the derived solar shading rate to the power generation prediction device 40E.

[0153] The solar shading rate prediction device 70E (or solar shading rate derivation unit 404E) terminates processing after outputting the solar shading rate to the power generation prediction device 40E. (This corresponds to the “end” shown in Figure 15.)

[0154] The power generation forecasting device 40E in the power generation forecasting system 1E predicts power generation based on the cloud shading rate and solar shading rate derived considering the wind direction at different altitudes as described above. This process is the same as that performed by the power generation forecasting device already described, so the explanation is omitted.

[0155] As described above, by adding wind direction information for different altitudes, it becomes possible to predict the differences in cloud movement direction depending on altitude, and thus we can provide a power generation forecasting device that improves the accuracy of power generation forecasts.

[0156] This embodiment further discloses the following configuration example. The aforementioned prediction unit, The cloud altitude at one or more selected locations, and the wind direction at each altitude, are used to derive the predicted cloud occlusion rate. The solar light shielding rate derivation unit is, In addition to the cloud altitude at each location selected by the aforementioned forecasting unit, the wind direction at each altitude is further used to derive the predicted solar shielding rate. The aforementioned power generation forecast value acquisition unit is: The input data for the learning model derived based on the predicted cloud shading rate and the predicted solar shading rate is input to the learning model that predicts power generation, using at least the input data for the learning model based on the cloud shading rate as input, thereby obtaining the predicted power generation output by the learning model. A power generation forecasting device characterized by the following features. Alternatively, the power generation forecast value acquisition unit may be: The corrected cloud shading rate prediction values, which are adjusted using the predicted solar shading rate for each altitude, are input into the learning model that predicts power generation using the cloud shading rate as input, thereby obtaining the predicted power generation value output by the learning model. As a result, this disclosure can provide a power generation forecasting device that enables improved accuracy in predicting power generation. Furthermore, by applying the above configuration to a system, a power generation forecasting method, or the above program, the same effects as described above can be achieved.

[0157] Here, we will describe the hardware configuration required to realize the functions of this disclosure. Figure 19 shows a first example of a hardware configuration for realizing the functions of the configuration of this disclosure. Figure 20 shows a second example of a hardware configuration for realizing the functions of the configuration of this disclosure. The cloud cover forecasting devices 10, 10A, 10B, and 10C of this disclosure are each implemented by hardware as shown in Figure 19 or Figure 20.

[0158] The cloud cover forecasting devices 10, 10A, 10B, 10C and the power generation forecasting device 40 are each composed of, for example, a processor 10001, a memory 10002, an input / output interface 10003, and a communication circuit 10004, as shown in Figure 19. The processor 10001 and memory 10002 are, for example, components installed in a computer. Each function of this disclosure is implemented by software, firmware, or a combination of software and firmware. The software or firmware is written as a program and stored in memory 10002. In other words, the memory 10002 contains the computer as follows: learning units 11, 11A, 11B, 11C, first cloud occlusion rate derivation unit 102 (102B, 102C), first time series data generation unit 103 (103B, 103C), second cloud occlusion rate derivation unit 202 (202B, 202C), second time series data generation unit 203 (203B, 203C), correlation calculation unit 301 (301B, 301C), The system stores programs for the following functions: a time-difference map generation unit 302 (302B, 302C), prediction units 12, 12A, 12B, 12C, cloud movement direction derivation unit 401 (401B, 401C), location selection unit 402 (402B, 402C), third cloud occlusion rate derivation unit 403 (403B, 403C), power generation amount prediction value acquisition unit 41, learning model unit 42, and a control unit (not shown). The processor 10001 reads and executes the program stored in memory 10002, thereby enabling the following: learning units 11, 11A, 11B, 11C, first cloud occlusion rate derivation unit 102 (102B, 102C), first time series data generation unit 103 (103B, 103C), second cloud occlusion rate derivation unit 202 (202B, 202C), second time series data generation unit 203 (203B, 203C), correlation The functions of the calculation unit 301 (301B, 301C), time difference map generation unit 302 (302B, 302C), prediction units 12, 12A, 12B, 12C, cloud movement direction derivation unit 401 (401B, 401C), location selection unit 402 (402B, 402C), third cloud occlusion rate derivation unit 403 (403B, 403C), power generation amount prediction value acquisition unit 41, learning model unit 42, and a control unit (not shown) are realized. The program causes the computer to execute the procedures or methods of each of the above components. Furthermore, the camera image storage unit 101 (101B, 101C), the time difference map storage unit 303 (303B, 303C), and a storage unit (not shown) are realized by memory 10002 or other memory (not shown). Furthermore, a communication unit (not shown) is realized by the communication circuit 10004.

[0159] The processor 10001 uses, for example, a CPU (Central Processing Unit), GPU (Graphics Processing Unit), microprocessor, microcontroller, or DSP (Digital Signal Processor). The memory 10002 may be a non-volatile or volatile semiconductor memory such as RAM (Random Access Memory), ROM (Read Only Memory), EPROM (Erasable Programmable ROM), EEPROM (Electrically Erasable Programmable Read Only Memory), or flash memory, or a magnetic disk such as a hard disk or flexible disk, or an optical disk such as a CD (Compact Disc) or DVD (Digital Versatile Disc), or a magneto-optical disk. The processor 10001, the memory 100, and the communication circuit 10004 are connected in a state where they can transmit data to each other. Further, the processor 10001, the memory 10002, and the communication circuit 10004 are connected in a state where they can transmit data to other hardware via the input / output interface 10003.

[0160] Alternatively, the functions of the learning units 11, 11A, 11B, 11C, the first cloud cover ratio derivation units 102 (102B, 102C), the first time-series data generation units 103 (103B, 103C), the second cloud cover ratio derivation units 202 (202B, 202C), the second time-series data generation units 203 (203B, 203C), the correlation calculation units 301 (301B, 301C), the time difference map generation units 302 (302B, 302C), the prediction units 12, 12A, 12B, 12C, the cloud movement direction derivation units 401 (401B, 401C), the location selection units 402 (402B, 402C), the third cloud cover ratio derivation units 403 (403B, 403C), the power generation prediction value acquisition unit 41, the learning model unit 42, and the control unit (not shown) in the cloud amount prediction devices 10, 10A, 10B, 10C and the power generation amount prediction device 40 may be realized by a dedicated processing circuit 20001 as shown in FIG. 20.

[0161] The processing circuit 20001 uses, for example, a single circuit, a composite circuit, a programmed processor, a parallel-programmed processor, an ASIC (Application Specific Integrated Circuit), a PLD (Programmable Logic Device), an FPGA (Field-Programmable Gate Array), a SoC (System-on-a-Chip), or a system LSI (Large-Scale Integration), etc. Or, a combination of these may be used. Also, the camera image storage unit 101 (101B, 101C), the time difference map storage unit 303 (303B, 303C), and the storage unit (not shown) are realized by the memory 20002 or another memory (not shown). Memory 20002 may be a non-volatile or volatile semiconductor memory such as RAM (Random Access Memory), ROM (Read Only Memory), EPROM (Erasable Programmable ROM), EEPROM (Electrically Erasable Programmable Read Only Memory), or flash memory; it may be a magnetic disk such as a hard disk or flexible disk; it may be an optical disk such as a CD (Compact Disc) or DVD (Digital Versatile Disc); or it may be a magneto-optical disk. Furthermore, a communication unit (not shown) is realized by the communication circuit 20004. The processing circuit 20001 and the memory 20002 or the communication circuit 20004 are connected in a way that allows them to transmit data to each other. Furthermore, the processing circuit 20001, the memory 20002, and the communication circuit 20004 are connected in a way that allows them to transmit data to other hardware via the input / output interface 20003. Furthermore, the cloud cover forecasting devices 10, 10A, 10B, 10C and the power generation forecasting device 40 consist of: learning units 11, 11A, 11B, 11C, a first cloud occlusion rate derivation unit 102 (102B, 102C), a first time-series data generation unit 103 (103B, 103C), a second cloud occlusion rate derivation unit 202 (202B, 202C), a second time-series data generation unit 203 (203B, 203C), and a correlation calculation unit 301 (301B, 301C). The functions of the time-difference map generation unit 302 (302B, 302C), prediction units 12, 12A, 12B, 12C, cloud movement direction derivation unit 401 (401B, 401C), location selection unit 402 (402B, 402C), third cloud occlusion rate derivation unit 403 (403B, 403C), power generation amount prediction value acquisition unit 41, learning model unit 42, and control unit (not shown) may be implemented by separate processing circuits or by a single processing circuit.

[0162] Alternatively, in the cloud cover forecasting devices 10, 10A, 10B, 10C and the power generation forecasting device 40, the learning units 11, 11A, 11B, 11C, the first cloud occlusion rate derivation unit 102 (102B, 102C), the first time series data generation unit 103 (103B, 103C), the second cloud occlusion rate derivation unit 202 (202B, 202C), the second time series data generation unit 203 (203B, 203C), the correlation calculation unit 301 (301B, 301C), and the time difference map generation unit 302 (302 Some functions of the following components (B, 302C), forecasting units 12, 12A, 12B, 12C, cloud movement direction derivation unit 401 (401B, 401C), location selection unit 402 (402B, 402C), third cloud occlusion rate derivation unit 403 (403B, 403C), power generation forecast value acquisition unit 41, learning model unit 42, and control units (not shown) may be implemented by the processor 10001 and memory 10002, while the remaining functions are implemented by the processing circuit 20001. In this way, each of the functions of the above components can be realized by hardware, software, firmware, or a combination thereof.

[0163] Within the scope of this disclosure, it is possible to freely combine the embodiments, modify any component of each embodiment, or omit any component of each embodiment. [Industrial applicability]

[0164] This disclosure enables the derivation of cloud cover forecast results with a smaller configuration compared to conventional methods, making it suitable for use in systems that include a cloud cover forecasting device for forecasting cloud cover used to forecast power generation, and a power generation forecasting device for forecasting power generation from the cloud cover. [Explanation of Symbols]

[0165] 1(1A,1B,1C,1D,1E) Power generation forecasting system, 10,10A,10B,10C,10D,10E Cloud cover forecasting device, 11,11A,11B,11C,11D,11E Learning unit, 12,12A,12B,12C,12D,12E Forecasting unit, 20 Camera, 30 Satellite data distribution system, 40,40D,40E Power generation forecasting device, 50 Camera i (Camera (i)), 60 Cloud altitude data distribution system, 70(70D,70E) Solar shading rate forecasting device, 80 Altitude-specific wind direction distribution system, 101(101B,101C) Camera image storage unit, 102(102B,102C) First cloud shading rate derivation unit, 103(103B,103C) First time-series data generation unit, 202 (202B, 202C) Second cloud occlusion rate derivation unit, 203 (203B, 203C) Second time-series data generation unit, 301 (301B, 301C) Correlation calculation unit, 302 (302B, 302C) Time difference map generation unit, 303 (303B, 303C) Time difference map storage unit, 12, 12A, 12B, 12C Forecast unit, 401 (401B, 401C, 401D, 401E) Cloud movement direction derivation unit, 402 (402B, 402C, 402D, 402E) Location selection unit, 403 (403B, 403C, 403D, 403E) Third cloud occlusion rate derivation unit, 404 (404D, 404E) Solar shading rate derivation unit, 1000 time difference map, 10001 processor, 10002 memory, 10003 input / output interface, 10004 communication circuit, 20001 processing circuit, 20002 memory, 20003 input / output interface, 20004 communication circuit.

Claims

1. A learning unit correlates a first time-series data of cloud occlusion rate derived from sky images, which are images of the sky taken from an observation point, with a second time-series data of cloud occlusion rate at multiple locations derived from satellite images, and stores the correspondence between each location and the time difference between the time-series data with the highest correlation as a time-difference map for each of the multiple locations. A prediction unit that uses the cloud movement direction obtained using satellite imagery and a preset time difference to derive the cloud occlusion rate at a selected location by referring to a time difference map accumulated by the learning unit, and outputs a predicted cloud occlusion rate value. A cloud cover forecasting device characterized by being equipped with the following features.

2. The aforementioned learning unit, A correlation calculation unit performs a correlation calculation between the first time series data, which is time series data of cloud occlusion rate derived based on sky images, which are images of the sky taken from an observation point, and the second time series data, which is time series data of cloud occlusion rate derived for multiple points using satellite images. A time difference map generation unit that uses the results of the correlation calculation to determine the time difference between time series data with high correlation for each location at multiple points included in the satellite image, A time difference map storage unit stores a time difference map that shows the correspondence between time series data highly correlated with a location at multiple locations, and Equipped with, The aforementioned prediction unit, A cloud movement direction derivation unit that derives the cloud movement direction using current satellite images from multiple frames, A location selection unit that selects a location using the cloud movement direction and a preset time difference by referring to the aforementioned time difference map, A cloud occlusion rate derivation unit that acquires a current cloud image and uses the current cloud image to derive the cloud occlusion rate at the selected location, Equipped with, The cloud cover forecasting device according to feature 1.

3. The current cloud image mentioned above is This is a satellite image output by a satellite data distribution system that distributes data from artificial satellites. The cloud cover forecasting device according to claim 2, characterized in that

4. The current cloud image mentioned above is These are images captured by cameras installed at multiple locations different from the aforementioned observation point. Characterized by, The cloud cover forecasting device according to claim 2.

5. The cloud cover prediction value output by the cloud cover prediction device according to claim 1 or claim 2 is input to a learning model that predicts power generation using the cloud cover rate as input, and the power generation prediction value output by the learning model is obtained. A power generation forecasting device characterized by the following features.

6. A learning unit that correlates a first time series of cloud occlusion rate data derived from sky images (images of the sky taken from an observation point) with a second time series of cloud occlusion rate data from multiple locations derived from satellite images, and accumulates a time difference map showing the correspondence between each location and the time difference between the time series data with the highest correlation for each of the multiple locations. A prediction unit that uses the cloud movement direction obtained using satellite imagery and the pre-accepted time difference to derive the cloud occlusion rate at a selected point by referring to a time difference map accumulated by the learning unit and outputs a predicted cloud occlusion rate value. A power generation prediction value acquisition unit inputs the aforementioned cloud occlusion rate prediction value into a learning model that predicts power generation using at least input data for a learning model based on the cloud occlusion rate as input, and acquires the power generation prediction value output by the learning model. Equipped with, A power generation forecasting device characterized by the following features.

7. The system further includes a solar shielding rate derivation unit that derives a predicted solar shielding rate value indicating the degree to which sunlight will be shielded based on the cloud altitude at the location selected by the prediction unit, The aforementioned power generation forecast value acquisition unit is: The input data for the learning model derived based on the predicted cloud shading rate and the predicted solar shading rate is input to the learning model that predicts power generation, using at least the input data for the learning model based on the cloud shading rate as input, thereby obtaining the predicted power generation output by the learning model. The power generation amount prediction device according to claim 6, characterized in that

8. The aforementioned prediction unit, The cloud altitude at one or more selected locations, and the wind direction at each altitude, are used to derive the predicted cloud occlusion rate. The solar shading rate derivation unit is, In addition to the cloud altitude at each location selected by the aforementioned forecasting unit, the wind direction at each altitude is further used to derive the predicted solar shielding rate. The aforementioned power generation forecast value acquisition unit is: The input data for the learning model derived based on the predicted cloud shading rate and the predicted solar shading rate is input to the learning model that predicts power generation, using at least the input data for the learning model based on the cloud shading rate as input, thereby obtaining the predicted power generation output by the learning model. The power generation prediction device according to feature 7.

9. A cloud cover prediction method using a cloud cover prediction device, The learning unit of the cloud cover prediction device, The first time-series data of cloud occlusion derived from sky images, which are images of the sky taken from an observation point, and the second time-series data of cloud occlusion at multiple locations, derived from satellite images, are correlated, and for each of the multiple locations, the correspondence between the location and the time difference between the time-series data with high correlation is accumulated as a time difference map. The prediction unit of the cloud cover prediction device, Using the cloud movement direction obtained from satellite images and a preset time difference, the cloud occlusion rate at a selected location is derived by referring to the time difference map accumulated by the learning unit, and a predicted cloud occlusion rate value is output. A cloud cover forecasting method characterized by the following features.

10. A method for predicting power generation using a power generation prediction device, The learning unit of the aforementioned power generation prediction device We correlate time-series data of cloud occlusion rates derived from sky images (images of the sky taken from observation points) with time-series data of cloud occlusion rates at multiple locations derived from satellite images, and accumulate the correspondence between satellite image locations and time differences where the correlation is high as a time-difference map. The prediction unit of the power generation prediction device, Using the cloud movement direction obtained from satellite images and a pre-set time difference, the cloud occlusion rate at the selected location is derived by referring to the time difference map accumulated by the learning unit, and the predicted cloud occlusion rate is output. The power generation prediction value acquisition unit of the power generation prediction device, The cloud occlusion rate forecast value output by the forecasting unit is input to a learning model that predicts power generation using the cloud occlusion rate as input, and the power generation forecast value, which is the prediction result output by the learning model, is obtained. A method for predicting power generation, characterized by the following features.

Images