JP2024064060A - Photovoltaic power generation management method, photovoltaic power generation management program, and photovoltaic power generation management system - Google Patents

Abstract

[Problem] To predict the snow accumulation condition on a photovoltaic power generation panel with high accuracy. [Solution] A photovoltaic power generation management method according to an embodiment includes a snow accumulation prediction step of predicting the snow accumulation condition of a specific area at a target prediction time based on photographed image data of the specific area including the installation location of a photovoltaic power generation panel for which the snow accumulation condition is to be predicted, and weather forecast data, and outputting the snow accumulation prediction result, and a snow accumulation prediction step of predicting the snow accumulation condition on the photovoltaic power generation panel at the target prediction time based on the snow accumulation prediction result, and outputting the snow accumulation prediction result. [Selected Figure] Figure 5

Description

Embodiments of the present invention relate to a photovoltaic power generation management method, a photovoltaic power generation management program, and a photovoltaic power generation management system.

In recent years, the correlation between infrastructure issues and events and meteorological information has increased due to emergency measures in response to the normalization of disasters caused by abnormal weather, and the introduction of renewable energy (renewable energy generators) as a management strategy for power companies and industrial users. With regard to renewable energy, for example, there is a need for advanced technology to predict the amount of power generated by photovoltaic (PV) power generation facilities. Furthermore, by implementing highly accurate PV power generation predictions, it will ultimately be possible to contribute to reducing imbalance costs in renewable energy balancing.

PV power generation predictions are carried out using, for example, satellite image data and solar radiation data. However, if snow accumulates on the PV panels, the amount of PV power generation decreases. Therefore, if snow accumulation is not taken into account, there is a possibility of a large discrepancy occurring between the predicted and actual power generation values.

As a countermeasure, for example, there is technology that predicts the snow accumulation on PV panels using information mainly on snow depth and the number of days since snowfall ended.

Patent Publication No. 6852621

However, the snow accumulation on PV panels is affected by factors other than snow depth and the number of days since snowfall ended, so there is room for improvement in the prediction accuracy of the above-mentioned conventional technology.

The problem that the present invention aims to solve is to provide a photovoltaic power generation management method, a photovoltaic power generation management program, and a photovoltaic power generation management system that can predict the snow accumulation condition on photovoltaic power generation panels with high accuracy.

The photovoltaic power generation management method of the embodiment includes a snowfall prediction step of predicting the snowfall condition of the specified area at the prediction target time based on photographed image data of the specified area including the installation location of the photovoltaic power generation panel for which the snowfall condition is to be predicted and weather forecast data, and outputting the snowfall prediction result, and a snowfall prediction step of predicting the snowfall condition of the photovoltaic power generation panel at the prediction target time based on the snowfall prediction result, and outputting the snowfall prediction result.

FIG. 1 is a graph showing an example of the time progression of the PV power generation amount and the PV power generation amount prediction result in the conventional technology. FIG. 2 is a block diagram illustrating a functional configuration of the photovoltaic power generation management device according to the first embodiment. FIG. 3 is a diagram illustrating a first snow accretion prediction algorithm in the first embodiment. FIG. 4 is a diagram illustrating a second snow accretion prediction algorithm in the first embodiment. FIG. 5 is a flowchart showing the process performed by the photovoltaic power generation management device of the first embodiment. FIG. 6 is a block diagram illustrating a functional configuration of a photovoltaic power generation management device according to the second embodiment. FIG. 7 is a flowchart showing a process performed by the photovoltaic power generation management device according to the second embodiment.

Below, embodiments (first embodiment and second embodiment) of the solar power generation management method, solar power generation management program, and solar power generation management system of the present invention will be described with reference to the drawings.

(Premise)
1 is a graph showing an example of a time transition of a PV power generation amount and a PV power generation amount prediction result in the conventional technology. Here, the solar power generation amount prediction result P and an actual output value R1 on a sunny day are shown.

Normally, the predicted solar power generation amount on a sunny day will look like symbol P in Figure 1. The actual output value is thought to roughly follow this. However, if it snows the day before and snow accumulates on the PV panels (photovoltaic power generation panels), the actual output value will no longer follow the predicted solar power generation amount.

Here, we will assume that snow accumulates on the PV panel at time t0, shortly after sunrise, and that the snow begins to melt and slide off gradually from around time t1, and is completely gone by time t2. Note that the presence or absence of snow on the PV panel is thought to be related to the presence or absence of snow on the ground, but since PV panels in Japan are installed at an inclination of about 10° (this varies depending on the latitude), there is a possibility that snow on the PV panel will slide off. Also, if the weather is good the day after the snow accumulates, the temperature of the PV panel will rise, causing the snow to melt or slide off.

In the example of Figure 1, the actual output value, as indicated by the symbol R1, is zero between times t0 and t1, starts to increase from time t1, and catches up with the solar power generation forecast result P at time t2. In other words, in the period from time t0 to just before time t2, the actual output value R1 is not proportional to the amount of solar radiation due to snow accumulation on the PV panel.

Although there is a method of installing cameras or sensors to recognize the snow accumulation condition on PV panels, this is problematic in terms of time and cost. Therefore, there is a need to predict the snow accumulation condition on PV panels with high accuracy without installing cameras or sensors.

However, as mentioned above, conventional technology predicts the snow accumulation condition on PV panels mainly using information on snow depth and the number of days since snowfall ended, and there is room for improvement in terms of prediction accuracy. Therefore, below we will explain a technology that can predict the snow accumulation condition on PV panels with higher accuracy than conventional technology.

In this specification, the state in which snow adheres to the PV panels is referred to as "snow accumulation," and the state in which snow accumulates on the ground or other surfaces is referred to as "snow accumulation."

First Embodiment
2 is a block diagram showing a functional configuration of the photovoltaic power generation management device 1 according to the first embodiment. The photovoltaic power generation management device 1 is a computer device, and includes a processing unit 2, a storage unit 3, an input unit 4, an output unit 5, and a communication unit 6.

Note that some of the functions of the photovoltaic power generation management device 1 can be realized, for example, by cloud computing, but in this embodiment, for the sake of simplicity, the photovoltaic power generation management device 1 is described as being a single computer device.

The storage unit 3 is realized by, for example, a RAM (Random Access Memory), a ROM (Read Only Memory), a flash memory, a HDD (Hard Disk Drive), an SSD (Solid State Drive), etc. The storage unit 3 stores, for example, the operating program of the processing unit 2, various data, and various calculation results by the processing unit 2.

Specifically, the memory unit 3 stores, for example, as data to be used for predicting the snow accumulation conditions on the PV panels, photographed image data (such as satellite image data) of a specific area including the installation location of the PV panels (hereinafter also simply referred to as the "installation location"), weather forecast data for that specific area (temperature, humidity, solar radiation, etc.), data for threshold determination, data for machine learning, location information of the installation location, and time characteristic information relating to the time to be predicted (information on characteristics (temperature, humidity, solar radiation, etc.) such as season, time of day (morning, noon, night, etc.)).

The processing unit 2 is realized, for example, by a CPU (Central Processing Unit). The processing unit 2 operates based on the operating program and various data stored in the memory unit 3, thereby realizing an acquisition unit 21, a snow accumulation prediction unit 22, and a snow accumulation prediction unit 23 as functional units.

The acquisition unit 21 acquires various information from the memory unit 3, external devices, etc.

The snow accumulation prediction unit 22 predicts the snow accumulation conditions in a specified area (hereinafter also simply referred to as the "specified region") at the target time based on photographed image data (e.g., satellite image data) of the specified area including the installation location of the solar power generation panel for which the snow accumulation conditions are to be predicted, and weather forecast data, and outputs the snow accumulation prediction results.

For example, the snowfall prediction unit 22 estimates the snowfall conditions in a specified area at the time of prediction execution based on satellite image data, and predicts the snowfall conditions in a specified area at the target prediction time using threshold judgment or machine learning based on the snowfall estimation result and weather forecast data, and outputs information on snow depth or the presence or absence of snowfall as the snowfall prediction result.

Here, we will explain how to estimate the snow cover in a given area at the time of making a prediction based on satellite image data. By subjecting satellite image data to image synthesis processing (e.g., RGB (Red, Green, Blue) synthesis processing), it is possible to determine the presence or absence of snow on the ground, as well as vegetation and clouds in the sky. For example, snow has a high reflectance to visible light and a high absorption rate to near-infrared light, while clouds have a high reflectance to both visible light and near-infrared light, and these characteristics make it possible to distinguish between snow cover and clouds.

The image synthesis processing techniques (recipe) used may be, for example, those that result in the synthesis of visible light or those that allow for the distinction between snow and fog, but it is also possible to combine multiple techniques.

Next, we will explain the prediction of snow conditions in a specific area at the target time (e.g. the next day) based on the snowfall estimation results for the specific area at the time the prediction is made (e.g. the present) and weather forecast data. For this prediction, for example, two types of methods can be used: threshold judgment and machine learning.

In the case of threshold judgment, a parameter is used to determine the threshold. In the case of machine learning, for example, random forests, regression analysis, logistic regression, etc. are used. In addition, snowfall prediction results can be, for example, of two types: snow depth and the presence or absence of snow, but it is also possible to finally set a threshold value for snow depth and make a binary classification of the presence or absence of snow.

When comparing threshold judgment and machine learning, it is believed that machine learning is more accurate when the amount of training data for the snowfall prediction model in machine learning is relatively large, and that threshold judgment is more accurate when the amount of training data is relatively small. This is because if the amount of training data is insufficient, the snowfall prediction model may become overtrained. Therefore, the amount of training data at that boundary is set to the first predetermined value.

The snowfall prediction unit 22 uses threshold judgment, for example, if the amount of learning data for the machine learning snowfall prediction model is less than a first predetermined value, and uses machine learning if the amount of learning data for the machine learning snowfall prediction model is equal to or greater than the first predetermined value. Specifically, for example, for a location where data is difficult to obtain, threshold judgment is used if the amount of learning data is less than the first predetermined value. Also, it is possible to perform snowfall prediction using threshold judgment when the amount of learning data is less than the first predetermined value immediately after the system is introduced, and switch the prediction method to machine learning after a predetermined period of time has passed and the amount of learning data is equal to or greater than the first predetermined value. In this way, threshold judgment and machine learning can be used appropriately depending on the amount of learning data.

The scope of the snowfall prediction model can be determined arbitrarily. That is, for example, it may be a uniform model for the entire country, or it may be a model for different areas. Here, the area may be an administrative division such as a city, ward, town, village, prefecture, or region, or it may be set arbitrarily by the implementer.

For example, a business operator who has installed multiple solar power plants in a prefecture can create a snowfall forecast model for all the power plants in the prefecture at once. The business operator can also designate area divisions within the prefecture and create snowfall forecast models for each area. However, when arbitrarily setting areas, weather forecast data for at least one location in the area must be available.

The snow accumulation prediction unit 23 predicts the snow accumulation condition on the solar power generation panel at the predicted time based on the snow accumulation prediction result, and outputs the snow accumulation prediction result.

For example, the snow accumulation prediction unit 23 predicts the snow accumulation condition of the solar power generation panel at the predicted target time using threshold judgment or machine learning based on at least one of the following in addition to the snow accumulation prediction result: weather forecast data, location information of the installation location, and time characteristic information related to the predicted target time, and outputs information on the presence or absence of snow accumulation or the probability of snow accumulation as the snow accumulation prediction result.

Below, the case where information on the presence or absence of snow accretion is output will be described using FIG. 3, and the case where information on the probability of snow accretion is output will be described using FIG. 4. FIG. 3 is a diagram that shows a schematic diagram of a first snow accretion prediction algorithm in the first embodiment. FIG. 4 is a diagram that shows a schematic diagram of a second snow accretion prediction algorithm in the first embodiment.

In FIG. 3, first, in step S11, it is determined whether there is snow on the ground at the target prediction time. If the answer is Yes, the process proceeds to step S12. If the answer is No, the process proceeds to step S13, where it is determined that there is no snow and the process ends.

In step S12, it is determined whether there has been snowfall in the last h hours. If the answer is Yes, the process proceeds to step S14. If the answer is No, the process proceeds to step S15, where it is determined that there is no snowfall and the process ends.

In step S14, it is determined whether the average temperature over the last h hours is less than the temperature threshold T. If the answer is Yes, the process proceeds to step S16. If the answer is No, the process proceeds to step S17, where it is determined that there is no snow accumulation and the process ends.

In step S16, it is determined whether the average solar radiation amount for the last h hours is less than the solar radiation amount threshold R. If the answer is Yes, the process proceeds to step S18, where it is determined that snow has accumulated and the process ends. If the answer is No, the process proceeds to step S19, where it is determined that no snow has accumulated and the process ends.

Next, in FIG. 4, first, in step S21, it is determined whether there is snow on the ground at the predicted target time, and if the answer is Yes, the process proceeds to step S22, and if the answer is No, the process proceeds to step S23, where it is determined that the probability of snow accumulation is 0%, and the process ends.

In step S22, it is determined whether there has been snowfall in the last h hours. If the answer is Yes, the process proceeds to step S24. If the answer is No, the process proceeds to step S25, where it is determined that the probability of snowfall is 10%, and the process ends.

In step S24, it is determined whether the average temperature over the last h hours is less than the temperature threshold T. If the answer is Yes, the process proceeds to step S26. If the answer is No, the process proceeds to step S27, where it is determined that the probability of snow accumulation is 50%, and the process ends.

In step S26, it is determined whether the average solar radiation amount for the last h hours is less than the solar radiation amount threshold R. If the answer is Yes, the process proceeds to step S28, where it is determined that the probability of snow accumulation is 100% and the process ends. If the answer is No, the process proceeds to step S29, where it is determined that the probability of snow accumulation is 80%, and the process ends.

In this way, the snowfall prediction result can be judged as a simple binary value of whether or not snow will accumulate, or it can also be judged as a probability.

Note that the judgment conditions used in Figures 3 and 4 are examples and are not limited to these. Other information that may be used include location information such as the latitude and longitude of the equipment to be predicted, equipment information such as the inclination angle and material of the PV panels, and time characteristic information such as the time and season. This information is related to the melting and sliding of snow on the PV panels, and is therefore effective in improving the accuracy of snow accumulation predictions. The probability value of the judgment result shown in Figure 4 can also be set arbitrarily, and it is also possible to vary the probability value for each equipment to be predicted using location information, equipment information, time characteristic information, etc., in the same way as the judgment conditions.

Returning to the explanation of FIG. 2, the snowfall prediction unit 23 uses threshold judgment if the amount of learning data for the snowfall learning model in machine learning is less than a second predetermined value, and uses machine learning if the amount of learning data for the snowfall learning model in machine learning is equal to or greater than the second predetermined value. As with snowfall prediction, if the amount of learning data is insufficient, the model may become over-learned, so the amount of learning data at the boundary is set to the second predetermined value, and threshold judgment and machine learning are used appropriately.

The input unit 4 is a means for the user to input information, such as a mouse, keyboard, or touch panel.

The output unit 5 is a display device such as an LCD (Liquid Crystal Display) or an audio output device such as a speaker.

The communication unit 6 is a communication interface for communicating (transmitting and receiving) various information with external devices and various sensors within the power system.

Figure 5 is a flowchart showing the processing by the photovoltaic power generation management device 1 of the first embodiment. First, in step S1, the snowfall prediction unit 22 estimates the current (time of prediction execution) snowfall condition in a given area based on satellite image data.

Next, in step S2, the snowfall prediction unit 22 predicts the snowfall conditions in a specified area at the prediction target time using threshold judgment or machine learning based on the snowfall estimation result from step S1 and the weather forecast data, and outputs the snowfall prediction result. For example, using the estimation result from step S1, it predicts the presence or absence of snowfall at the location, and obtains the snowfall prediction result for a specified number of days (e.g., about one week) from the current day.

Next, in step S3, the snow accumulation prediction unit 23 predicts the snow accumulation condition on the photovoltaic power generation panel at the target time using threshold judgment or machine learning based on at least one of the weather forecast data, the location information of the installation location, and the time characteristic information related to the target time of prediction in addition to the snow accumulation prediction result in step S2, and outputs the snow accumulation prediction result.

In this way, according to the photovoltaic power generation management device 1 of the first embodiment, as described above, first, the current (time of prediction) snow accumulation condition in a specified area is estimated based on satellite image data, etc., then the snow accumulation condition in the specified area at the target prediction time is predicted based on the snow accumulation estimation result and weather forecast data, etc., and then the snow accumulation condition on the photovoltaic power generation panel at the target prediction time is predicted based on the snow accumulation prediction result. Through such step-by-step processing, it is possible to predict the snow accumulation condition on the PV panel with higher accuracy than before, under the condition that no dedicated camera or sensor is installed.

In addition, the snowfall prediction results can be used for various subsequent information processing. As an example, the following second embodiment will explain the case where the solar power generation prediction results are corrected based on the snowfall prediction results.

Second Embodiment
A second embodiment will be described. Descriptions of matters similar to those of the first embodiment will be omitted as appropriate. Fig. 6 is a block diagram showing the functional configuration of the photovoltaic power generation management device 1 of the second embodiment. Compared to Fig. 2, a correction unit 24 is added to the processing unit 2.

The acquisition unit 21 acquires the result of a prediction of the amount of solar power generation by the PV panel. This result of a prediction of the amount of solar power generation may be created by the solar power generation management device 1, or may be acquired from another computer device.

The correction unit 24 corrects the solar power generation forecast result based on the snow accretion forecast result. The correction unit 24 corrects the solar power generation forecast result based on, for example, a correction coefficient according to the snow accretion forecast result.

The following describes in detail how to correct the solar power generation forecast results. For example, a method can be considered in which the power generation output forecast results are multiplied by a correction coefficient c_r (0≦c_r≦1) for the time when it is determined that snow has accumulated.

The method for calculating this correction coefficient can be a constant value or a variable value. When using a constant value, a correction coefficient value can be determined in advance for any area unit and any period of time and given to the system.

When using a variable value, possible methods for variation include using time information (by season, month, week, day, time of day, etc.), using the probability of snow accumulation based on snow accumulation forecast results, or using location information such as the latitude and longitude of the solar power generation facility to be corrected.

When using the snow accumulation probability from the snow accumulation forecast result, a calculation method can be considered in which the correction coefficient is set to 0.9 when the snow accumulation probability is 10%, and 0.8 when the snow accumulation probability is 20%. In other words, when the snow accumulation probability is SP and the correction coefficient is CF, the correction coefficient can be calculated by the following formula (1).
CF = (100 - SP) × 0.01 ... (1)

Figure 7 is a flowchart showing the processing by the photovoltaic power generation management device 1 of the second embodiment. Steps S1 to S3 are the same as steps S1 to S3 in Figure 5.

After step S3, in step S4, the correction unit 24 determines whether or not correction of the PV power generation prediction result is necessary. If Yes, the process proceeds to step S5, and if No, the process ends.

In step S5, the correction unit 24 corrects the solar power generation forecast result based on a correction coefficient according to the snow accumulation forecast result, etc.

In this way, according to the second embodiment, highly accurate solar power generation forecast results can be obtained by correcting the solar power generation forecast results based on the snow accumulation forecast results.

Although the embodiments of the present invention have been described above, these embodiments are presented as examples and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, substitutions, and modifications can be made without departing from the gist of the invention. These embodiments and their modifications are included in the scope of the invention and its equivalents as described in the claims, as well as in the scope and gist of the invention.

The program executed by the photovoltaic power generation management device 1 of this embodiment can be provided by recording it in an installable or executable file format on a recording medium that can be read by a computer device, such as a CD (Compact Disc)-ROM (Read Only Memory), a flexible disk (FD), a CD-R (Recordable), or a DVD (Digital Versatile Disk). The program may also be provided or distributed via a network such as the Internet.

The prediction of snow accretion on the PV panels and the correction of the solar power generation forecast results may be performed by the same device (program) or by separate devices (programs). In the latter case, for example, as in the first embodiment, the solar power generation management device 1 predicts snow accretion on the PV panels and transmits the snow accretion forecast results to another computer device, which then predicts the solar power generation amount and corrects the solar power generation forecast results using the snow accretion forecast results.

In addition, the photographed image data of a specified area including the installation location of the PV panels is not limited to satellite image data, but may be other data such as aerial photograph data.

1...Photovoltaic power generation management device, 2...Processing unit, 3...Memory unit, 4...Input unit, 5...Output unit, 6...Communication unit, 21...Acquisition unit, 22...Snowfall prediction unit, 23...Snow accumulation prediction unit, 24...Correction unit

Claims (10)

a snowfall prediction step of predicting the snowfall condition in a specific area at a prediction target time based on photographed image data of the specific area including the installation location of the solar power generation panel to be predicted for snowfall conditions and weather forecast data, and outputting a snowfall prediction result;
A photovoltaic power generation management method including a snow accumulation prediction step of predicting the snow accumulation condition on the photovoltaic power generation panel at the prediction target time based on the snow accumulation prediction result and outputting the snow accumulation prediction result. The photovoltaic power generation management method according to claim 1, wherein the snowfall prediction step estimates the snowfall condition of the specified area at the time of prediction execution based on the captured image data, predicts the snowfall condition of the specified area at the prediction target time based on the snowfall estimation result and the weather forecast data using threshold judgment or machine learning, and outputs information on snowfall depth or the presence or absence of snowfall as the snowfall prediction result. The photovoltaic power generation management method according to claim 1, wherein the snowfall prediction step uses satellite image data as the captured image data. The photovoltaic power generation management method according to claim 1, wherein the snowfall prediction step predicts the snowfall status of the photovoltaic power generation panel at the target time of prediction using threshold judgment or machine learning based on at least one of the weather forecast data, the location information of the installation location, and the time characteristic information related to the target time of prediction in addition to the snowfall prediction result, and outputs information on the presence or absence of snowfall or the probability of snowfall as the snowfall prediction result. The snowfall prediction step includes:
If the amount of learning data of the snowfall prediction model in the machine learning is less than a first predetermined value, the threshold determination is used,
The photovoltaic power generation management method according to claim 2 , wherein the machine learning is used if an amount of learning data of the snowfall prediction model in the machine learning is equal to or greater than a first predetermined value. The snow accumulation prediction step includes:
If the amount of learning data of the snow accretion learning model in the machine learning is less than a second predetermined value, the threshold determination is used,
The photovoltaic power generation management method according to claim 4 , wherein the machine learning is used if an amount of learning data of a snow accretion learning model in the machine learning is equal to or greater than a second predetermined value. An acquisition step of acquiring a predicted result of solar power generation by the solar power generation panel;
The photovoltaic power generation management method according to claim 1 , further comprising: a correction step of correcting the photovoltaic power generation forecast result based on the snow accretion forecast result. The photovoltaic power generation management method according to claim 7, wherein the correction step corrects the photovoltaic power generation forecast result based on a correction coefficient according to the snow accretion forecast result. Computer,
a snowfall prediction unit that predicts the snowfall condition of a specific area at a prediction target time based on photographed image data of the specific area including the installation location of the solar power generation panel that is a target for predicting the snowfall condition and weather forecast data, and outputs a snowfall prediction result;
A solar power generation management program for functioning as a snow accumulation prediction unit that predicts the snow accumulation condition on the solar power generation panel at the predicted time based on the snow accumulation prediction result and outputs the snow accumulation prediction result. a snowfall prediction unit that predicts the snowfall condition of a specific area at a prediction target time based on photographed image data of the specific area including the installation location of the solar power generation panel that is a target for predicting the snowfall condition and weather forecast data, and outputs a snowfall prediction result;
A solar power generation management system comprising: a snow accumulation prediction unit that predicts the snow accumulation condition on the solar power generation panel at the prediction target time based on the snow accumulation prediction result and outputs the snow accumulation prediction result.

Images