Wind condition observation period determination support device and wind condition observation period determination support method

Abstract

To determine a wind state observation period in which annual wind state can be highly accurately predicted.SOLUTION: A wind state observation period determination support apparatus 100 comprises: a similarity calculation unit 113 which calculates the similarity between annual wind state prediction data (prediction data 130) in a target place where a wind state is predicted and wind state prediction data (prediction data 130) in a wind state observation period shorter than a year in the target place; and a wind state observation period determination unit 114 which outputs the wind state observation period in which the similarity becomes maximum. The wind state prediction data is, for example, the occurrence frequency distribution of a wind speed, the occurrence frequency distributions of both the wind speed and wind direction or the average wind speed.SELECTED DRAWING: Figure 1

Description

Technical Field

[0003] , , , , , , , , , ,

[0004] , , , , ,

[0001] The present invention relates to a wind condition observation period determination support device and a wind condition observation period determination support method for assisting in determining a period for observing wind conditions at a target location for predicting wind conditions.

Background Art

[0002] In order to determine the construction site of a wind power plant, it is necessary to predict the annual wind conditions (distribution of wind direction and wind speed at the height of the wind turbine) at the construction site and calculate the power generation amount. To predict wind conditions, it is necessary to observe the wind direction and wind speed. For observation, it is desirable to install an observation tower (mast, wind condition observation tower) for observation, but it is costly. For this reason, observations using a Doppler lidar for meteorological observations that are easy to install have been attempted. On the other hand, in order to suppress the observation cost, it is desirable that the wind condition observation period be short. The wind condition prediction system described in Patent Document 1 predicts the wind conditions at a specific location (construction candidate site) based on observation data for, for example, one month at the specific location.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] According to the wind condition prediction system described in Patent Document 1, it becomes possible to predict wind conditions by short-term observation instead of annual observation. Although it is considered that there is a difference in the prediction accuracy depending on the time (season) of the wind condition observation period, there is no description of a method for determining an appropriate time. In particular, in the vicinity of Japan, due to the influence of seasonal winds, northwestern winds are likely to blow in winter, and southeastern or southwestern winds are likely to blow in summer. When considering seasonal variations, not only the method of correction by a seasonal correction unit but also the method of observing wind conditions at an appropriate time when various wind directions can be observed is useful. This invention has been made in view of the above background, and aims to provide a wind condition observation period determination support device and a wind condition observation period determination support method that support the determination of a wind condition observation period that enables highly accurate wind condition forecasting. [Means for solving the problem]

[0005] To solve the above-mentioned problems, the wind condition observation period determination support device according to the present invention is used in the area where wind conditions are predicted. Among the wind condition forecast data Wind condition forecast data for the first period and wind condition forecast data for a second period shorter than the first period at the target site. Based on this, a distribution calculation unit calculates the frequency distribution of wind speed, the probability distribution of wind speed, the frequency distribution of both wind speed and wind direction, or the probability distribution of both wind speed and wind direction. Based on the difference between the frequency distribution of wind speed, the probability distribution of wind speed, the frequency distribution of both wind speed and wind direction, or the probability distribution of both wind speed and wind direction in the wind condition forecast data for the first period and the wind condition forecast data for the second period, A similarity calculation unit calculates the similarity, and a wind condition observation period determination unit outputs the second period in which the similarity is maximized. and, It is equipped with. [Effects of the Invention]

[0006] According to the present invention, it is possible to provide a wind condition observation period determination support device and a wind condition observation period determination support method that assist in determining the wind condition observation period, which enables highly accurate wind condition forecasting. Problems, configurations, and effects other than those described above will be clarified by the following description of embodiments. [Brief explanation of the drawing]

[0007] [Figure 1] This is a functional block diagram of the wind condition observation period determination support device according to the first embodiment. [Figure 2] This graph illustrates the difference in wind speed distribution according to the first embodiment. [Figure 3] This is a flowchart of the wind condition observation period determination support process according to the first embodiment. [Figure 4] This figure illustrates the distribution of wind direction and wind speed according to a modified example of the first embodiment. [Figure 5] This is a functional block diagram of the wind condition observation period determination support device according to the second embodiment. [Figure 6] This is a flowchart of the wind condition observation period determination support process according to the second embodiment. [Figure 7]This is a functional block diagram of the wind condition observation period determination support device according to the third embodiment. [Figure 8] This is a diagram illustrating the predictive model according to the third embodiment. [Figure 9] This is a flowchart of the wind condition observation period determination support process according to the third embodiment. [Modes for carrying out the invention]

[0008] <<First Embodiment: Overview of the Wind Condition Observation Period Determination Support Device>> Next, a wind condition observation period determination support device in an embodiment for carrying out the present invention will be described. The wind condition observation period determination support device determines a short period (second period) shorter than one year, for example, one month, for observing wind conditions in order to predict the annual (first period) wind conditions (frequency distribution or probability distribution of wind speed occurrence) at a prediction target site, which is a candidate site for the construction of a wind power plant and is the target site for predicting wind conditions. The wind condition observation period determination support device determines the annual wind conditions (wind condition prediction data) and a short period (wind condition observation period) in which the similarity between the wind conditions of that period and the annual wind conditions (wind condition prediction data) is high. Users of the wind condition observation period determination support device will be able to predict annual wind conditions with high accuracy while keeping observation costs down by conducting short-term observations.

[0009] <<First Embodiment: Configuration of the Wind Condition Observation Period Determination Support Device>> Figure 1 is a functional block diagram of the wind condition observation period determination support device 100 according to the first embodiment. The wind condition observation period determination support device 100 is a computer and comprises a control unit 110, a storage unit 120, and an input / output unit 180. User interface devices such as a display, keyboard, and mouse are connected to the input / output unit 180. The input / output unit 180 may also be equipped with a communication device, enabling the transmission and reception of data (for example, weather observation data and forecast data) with other devices. Furthermore, a media drive may be connected to the input / output unit 180, enabling data exchange using a recording medium.

[0010] ≪First Embodiment: Wind Condition Observation Period Determination Support Device: Memory Unit≫ The storage unit 120 is configured to include storage devices such as a ROM (Read Only Memory), a RAM (Random Access Memory), and an SSD (Solid State Drive). The storage unit 120 stores setting data 121, prediction data 130, and a program 128. The setting data 121 stores data set by the user of the wind condition observation period determination support device 100. The setting contents include the prediction target area (target area) and the length of the wind condition observation period (wind condition observation period length).

[0011] The prediction data 130 stores the annual wind condition prediction data for the prediction target area. Examples of the prediction data include data calculated using a numerical weather prediction model (e.g., the Mesoscale model). The prediction data includes, for example, wind direction and wind speed data at 10-minute intervals. The program 128 includes a description of the procedure of the wind condition observation period determination support process (see FIG. 3 described later).

[0012] ≪First Embodiment: Wind Condition Observation Period Determination Support Device: Control Unit≫ The control unit 110 is configured to include a CPU (Central Processing Unit) and includes a setting unit 111, a distribution calculation unit 112, a similarity calculation unit 113, and a wind condition observation period determination unit 114. The setting unit 111 receives instructions such as the prediction target area and the length of the wind condition observation period from the user and stores them in the setting data 121. The distribution calculation unit 112 acquires wind condition data from the prediction data 130 and calculates the distribution of wind speed (wind speed distribution, frequency distribution of wind speed, probability distribution of wind speed). For example, the distribution calculation unit 112 calculates the annual frequency distribution of wind speed.

[0013] The similarity calculation unit 113 calculates the similarity by calculating the difference between the frequency distributions of the wind speeds. FIG. 2 is graphs 411 and 412 for explaining the difference in the frequency distribution of the wind speed according to the first embodiment. The horizontal axis of the graphs 411 and 412 is the wind speed, and the vertical axis is the frequency of occurrence. The solid-line graph 411 shows the annual wind conditions (frequency distribution of the wind speed), and the dashed-line graph 412 shows the wind conditions in January (Mutsuki as the second period). The similarity calculation unit 113 calculates the area of the region 413 sandwiched between the graphs 411 and 412, which is the difference (the difference) between the annual wind conditions (wind condition prediction data for the first period) and the wind conditions in January (wind condition prediction data for the second period). The similarity calculation unit 113 determines that the smaller the difference, the higher the similarity between the two wind conditions. Note that the similarity calculation unit 113 may calculate the similarity by calculating the difference between the probability distributions of the wind speed appearances. The wind condition observation period determination unit 114 determines and outputs, as the wind condition observation period, the wind condition observation period candidate having the highest similarity (the smallest difference) with the annual wind conditions among a plurality of wind condition observation period candidates.

[0014] As described above, the wind condition observation period determination support device 100 includes a similarity calculation unit 113 that calculates the similarity between the wind condition prediction data for the first period (annual wind conditions) at the target location where the wind conditions are predicted and the wind condition prediction data for the second period shorter than the first period (wind conditions of the wind condition observation period candidates) at the target location. The wind condition observation period determination support device 100 also includes a wind condition observation period determination unit 114 that outputs the second period (as the wind condition observation period) with the maximum similarity.

[0015] The similarity calculation unit 113 calculates the difference between the frequency distribution of the wind speed calculated by the distribution calculation unit 112 according to the first embodiment from the wind condition prediction data for the first period and the wind condition prediction data for the second period, or the difference between the probability distributions of the wind speed appearances, and determines that the case where the difference is the smallest is the case where the similarity is the maximum. Regarding the wind condition prediction data for the first period, the first period is one year. Regarding the wind condition prediction data for the second period, the second period is less than one year (for example, one month).

[0016] ≪First Embodiment: Wind Condition Observation Period Determination Support Process≫ Figure 3 is a flowchart of the wind condition observation period determination support process according to the first embodiment. The wind condition observation period determination support process will be explained with reference to Figure 3. In step S11, the distribution calculation unit 112 calculates the annual wind speed distribution in the area to be predicted.

[0017] In step S12, the wind condition observation period determination unit 114 starts the process of repeating steps S13 to S15 for each candidate wind condition observation period. If the wind condition observation period length (length of the second period) is set to 1 month, the candidate periods are, for example, January, February, ..., December. Depending on the user's settings, the start dates may be at 15-day intervals, such as January 1st to January 31st, January 16th to February 15th, February 1st to February 28th, etc. Also, there may be multiple wind condition observation period lengths, such as 1 month and 1.5 months.

[0018] In step S13, the distribution calculation unit 112 calculates the wind speed distribution for the candidate period in the area to be predicted. In step S14, the similarity calculation unit 113 calculates the similarity between the annual wind speed distribution calculated in step S11 and the wind speed distribution for the candidate period calculated in step S13. In step S15, the wind condition observation period determination unit 114 proceeds to step S16 if it has performed steps S13 to S14 for all candidate periods. If there are any unprocessed candidate periods, it returns to step S13 and performs steps S13 to S14 for the next candidate period. In step S16, the wind condition observation period determination unit 114 determines the candidate period that has the maximum similarity (minimum difference) calculated in step S14, and outputs this candidate period as the wind condition observation period.

[0019] ≪First Embodiment: Features of the Wind Condition Observation Period Determination Support Device≫ The wind condition observation period determination support device 100 compares the annual wind conditions (frequency distribution of wind speed occurrences) with the wind conditions of candidate periods and selects the candidate period with the smallest difference (highest similarity) as the wind condition observation period. By observing wind conditions during an observation period similar to the annual wind conditions, it becomes possible to predict the annual wind conditions with high accuracy.

[0020] <<Modification of the First Embodiment: Distribution of Wind Direction and Conditions>> In the first embodiment described above, the wind condition observation period determination support device 100 compares the distribution of wind speed (see steps S13 and S14 in Figure 3). Alternatively, the distributions of both wind direction and wind speed (frequency distribution and probability distribution) may be compared.

[0021] Figure 4 is a diagram illustrating a wind direction and speed distribution 420 according to a modified example of the first embodiment. The horizontal axis of the distribution 420 represents wind speed, and the vertical axis represents wind direction in a clockwise direction with north being 0 degrees. The wind speed is divided into intervals of 2 m / s, and the wind direction is divided into intervals of 45 degrees. In the distribution 420, for example, the probability of occurrence of winds with a wind speed of 2 m / s or less and a wind direction of north to northeast (0 degrees to 45 degrees) is 0.1.

[0022] The modified distribution calculation unit 112 calculates a wind direction and wind speed distribution 420 instead of a wind speed distribution (see Figure 2). The modified similarity calculation unit 113 calculates the similarity by summing the differences between the annual occurrence probability and the occurrence probability for the candidate period for each set of wind direction and wind speed probabilities. The wind condition observation period determination support device 100 compares the annual wind conditions (wind direction and wind speed distribution) with the wind conditions for the candidate period and selects the candidate period with the smallest difference as the wind condition observation period. By comparing from the perspectives of both wind direction and wind speed, it is possible to determine a wind condition observation period with higher similarity than in the first embodiment, and more accurate wind condition forecasts can be expected.

[0023] For example, in the vicinity of Japan, due to the influence of seasonal winds, northwesterly winds tend to blow in winter, while southeasterly or southwesterly winds tend to blow in summer. When considering seasonal variations, in addition to methods that correct using a seasonal correction unit, as in Patent Document 1, methods that allow observation of various wind directions and observe wind conditions at a time similar to the annual wind direction distribution are also useful.

[0024] As described above, the wind condition observation period determination support device 100 includes a distribution calculation unit 112 that calculates the frequency distribution of wind speed occurrences, the probability distribution of wind speed occurrences, the frequency distribution of both wind speed and wind direction occurrences, or the probability distribution of both wind speed and wind direction occurrences, based on the wind condition forecast data for the first period (annual wind conditions) and the wind condition forecast data for the second period (wind conditions for candidate wind condition observation periods). The similarity calculation unit 113 calculates the difference between the frequency distribution of wind speed, the probability distribution of wind speed, the frequency distribution of both wind speed and wind direction, or the probability distribution of both wind speed and wind direction in the wind condition forecast data for the first period and the wind condition forecast data for the second period, and determines that the similarity is maximized when the difference is smallest.

[0025] ≪Second Embodiment≫ The wind condition observation period determination support device 100 of the first embodiment and its modifications compare the distribution of wind speed or the distribution of wind direction and wind speed as wind conditions with an annual value and a candidate period. The wind condition observation period determination support device may compare the average wind speed instead of the distribution of wind speed.

[0026] Figure 5 is a functional block diagram of the wind condition observation period determination support device 100A according to the second embodiment. Compared with the wind condition observation period determination support device 100 of the first embodiment, the distribution calculation unit 112 and the similarity calculation unit 113 provided in the control unit 110 are replaced by the average wind speed calculation unit 112A and the similarity calculation unit 113A.

[0027] The average wind speed calculation unit 112A obtains wind condition data from the forecast data 130 and calculates the average wind speed for the year and the candidate period. The similarity calculation unit 113A calculates the difference using the following formula (1) and calculates the similarity such that the smaller the difference, the greater the similarity. Difference = |Average wind speed for the year - Average wind speed for the candidate period| ÷ Average wind speed for the year (1)

[0028] ≪Second Embodiment: Support Processing for Determining Wind Condition Observation Period≫ Figure 6 is a flowchart of the wind condition observation period determination support process according to the second embodiment. The wind condition observation period determination support process will be explained with reference to Figure 6. In step S21, the average wind speed calculation unit 112A calculates the annual average wind speed at the prediction site.

[0029] In step S22, the wind condition observation period determination unit 114 starts the process of repeating steps S23 to S25 for each candidate wind condition observation period. In step S23, the average wind speed calculation unit 112A calculates the average wind speed for the candidate period at the prediction target site. In step S24, the similarity calculation unit 113A calculates the similarity by calculating the difference between the annual average wind speed calculated in step S21 and the average wind speed for the candidate period calculated in step S23 (see formula (1)).

[0030] In step S25, the wind condition observation period determination unit 114 proceeds to step S26 if it has performed steps S23 to S24 for all candidate periods. If there are any unprocessed candidate periods, it returns to step S23 and performs steps S23 to S24 for the next candidate period. In step S26, the wind condition observation period determination unit 114 determines the candidate period with the highest similarity calculated in step S24 and outputs this candidate period as the wind condition observation period.

[0031] ≪Second Embodiment: Features of the Wind Condition Observation Period Determination Support Device≫ The wind condition observation period determination support device 100A compares the annual wind conditions (average wind speed) with the wind conditions for candidate periods and selects the candidate period with the smallest difference as the wind condition observation period. By observing wind conditions during observation periods similar to the annual wind conditions, it becomes possible to predict the annual wind conditions with high accuracy. Furthermore, compared to the wind condition observation period determination support device 100 in the first embodiment, the wind condition observation period can be determined at a higher speed.

[0032] As described above, the wind condition observation period determination support device 100A includes an average wind speed calculation unit 112A that calculates the average wind speed based on wind condition forecast data for the first period (annual wind conditions) and wind condition forecast data for the second period (wind conditions for candidate wind condition observation periods). The similarity calculation unit 113A determines that the similarity is maximized when the difference between the average wind speed in the wind condition forecast data for the first period and the average wind speed in the wind condition forecast data for the second period is smallest.

[0033] ≪Third Embodiment≫ The wind condition observation period determination support device 100 of the first embodiment calculates the similarity of wind conditions at the prediction site between the year and the candidate period based on the prediction data 130. Alternatively, machine learning technology may be used to predict wind conditions based on observation data near the ground, such as AMeDAS, and the wind condition observation period may be determined by calculating the similarity of wind conditions at the prediction site between the year and the candidate period based on the prediction results.

[0034] Figure 7 is a functional block diagram of the wind condition observation period determination support device 100B according to the third embodiment. Compared to the wind condition observation period determination support device 100 of the first embodiment, the control unit 110 does not have a distribution calculation unit 112, and the prediction data 130B stored in the storage unit 120 is different. The control unit 110 also includes a learning unit 115 and a prediction unit 116, and the storage unit 120 includes observation data 140 and a prediction model database 150 (labeled as prediction model DB (database) in Figure 7).

[0035] Prediction data 130B stores predicted wind conditions for the target area and its surrounding areas (also simply referred to as surrounding areas). Observation data 140 is observation data for the surrounding areas of the target area, such as data observed at AMeDAS, existing observation towers, Doppler lidar, and wind power plants. Prediction model database 150 stores prediction models 158 for each candidate period (see Figure 8 below). To increase the number of training data points, it is desirable that the surrounding areas are close to the target area, and that the number of surrounding areas and the number of data points (years) in prediction data 130B and observation data 140 are large.

[0036] Figure 8 is a diagram illustrating the prediction model 158 according to the third embodiment. The prediction model 158 is a machine learning model in which the explanatory variable is the observed data 421 and the dependent variable is the wind condition 422 (see prediction data 130B). Assume that the candidate periods are January, February, ..., December. Then, the prediction model 158 for January is generated (trained) using the observed data 421 for January in the surrounding area and the wind condition 422 for January in the surrounding area as training data. The prediction model 158 is, for example, a Gaussian process regression model using an exponential kernel.

[0037] Returning to Figure 7, let's explain the control unit 110. The learning unit 115 generates a prediction model 158 for each candidate period and stores it in the prediction model database 150. The prediction unit 116 uses the prediction model 158 to predict wind conditions at the target site based on observational data of the target site.

[0038] As described above, the wind condition observation period determination support device 100B includes a prediction unit 116 that calculates wind condition forecast data for the second period at the target site based on the observation data for the second period at the surrounding area, using a prediction model 158, which is a machine learning model generated using training data in which the observation data for the second period at the target site is used as explanatory variables and the wind condition forecast data for the second period at the target site is used as the objective variable.

[0039] ≪Third Embodiment: Support Processing for Determining Wind Condition Observation Period≫ Figure 9 is a flowchart of the wind condition observation period determination support process according to the third embodiment. The wind condition observation period determination support process will be explained with reference to Figure 9. In step S31, the wind condition observation period determination unit 114 starts the process of repeating steps S32 to S41 for each candidate period. Hereinafter, the candidate period in the repeated process will be referred to as the candidate period to be processed.

[0040] In step S32, the learning unit 115 starts the process of repeating steps S33 to S37 for each (past) year stored in the prediction data 130 and observation data 140. Hereinafter, the year in the repeated processing will be referred to as the processing target year. In step S33, the learning unit 115 starts the process of repeating steps S34 to S36 for each surrounding area. Hereafter, the surrounding areas in the repeated process will be referred to as the processing target surrounding areas.

[0041] In step S34, the learning unit 115 obtains observation data for the candidate processing period of the processing year in the area surrounding the processing target from observation data 140, and obtains wind conditions (distribution of wind speed or distribution of wind direction and wind speed) from prediction data 130B. In step S35, the learning unit 115 generates training data in which the observation data acquired in step S35 is used as explanatory variables and wind conditions as the target variable (ground truth data).

[0042] In step S36, if the learning unit 115 has executed steps S34 to S35 for all processing target peripherals, it proceeds to step S37. If there are any processing target peripherals that have not yet been processed, it returns to step S34 and executes steps S34 to S35 for the next processing target peripheral. In step S37, if the learning unit 115 has executed steps S33 to S36 for all target years, it proceeds to step S38. If there are any unprocessed target years, it returns to step S33 and executes steps S33 to S36 for the next target year.

[0043] In step S38, the learning unit 115 generates a predictive model 158 for the candidate processing period using the training data generated in step S35. In step S39, the prediction unit 116 obtains observation data of the area surrounding the prediction target from the observation data 140. This observation data may be instantaneous observation data or the average value of observation data for the candidate period processed by the observation data 140, or it may be instantaneous observation data or the average value of observation data for the candidate period to be processed over the past few years. In step S40, the prediction unit 116 uses the prediction model 158 generated in step S38 to predict (calculate) the wind conditions based on the observation data acquired in step S39.

[0044] In step S41, the wind condition observation period determination unit 114 proceeds to step S42 if it has executed steps S32 to S40 for all candidate processing periods. If there are any unprocessed candidate processing periods, it returns to step S32 and executes steps S32 to S40 for the next candidate processing period. In step S42, the wind condition observation period determination unit 114 obtains the annual wind conditions from the forecast data 130B. In step S43, the wind condition observation period determination unit 114 determines the wind condition predicted in step S40 that has the greatest similarity (minimum difference) to the annual wind condition obtained in step S42, and outputs the candidate period for that wind condition as the wind condition observation period (second period). The similarity is calculated by the similarity calculation unit 113.

[0045] ≪Third Embodiment: Features of the Wind Condition Observation Period Determination Support Device≫ The wind condition observation period determination support device 100B uses a prediction model 158 generated using observation data from surrounding areas and prediction data from the target area to predict (calculate) the wind conditions of the target area based on the observation data from surrounding areas. Based on these wind conditions, the wind condition observation period determination support device 100B determines the wind condition observation period that has the greatest similarity to the annual wind conditions. Since it uses wind conditions based on observation data, it is expected that the wind condition observation period with a high similarity to the annual wind conditions will be determined.

[0046] ≪Variations≫ It should be noted that the present invention is not limited to the embodiments described above and can be modified without departing from its spirit. Furthermore, although several embodiments of the present invention have been described, these embodiments are merely illustrative and do not limit the technical scope of the present invention. For example, in the third embodiment, the wind condition observation period determination support device 100B includes a learning unit 115 and generates a prediction model 158 to predict wind conditions, but predictions may also be made using a prediction model 158 generated by another device.

[0047] In the above embodiment, the length of the first period is set to one year, but the lengths of the first and second periods may be set to 10 years and 1 year, respectively. Furthermore, the wind condition observation period may be divided according to the user's settings. For example, if the wind condition observation period is 30 days long, it may be divided into two 15-day wind condition observation periods, or into three 10-day wind condition observation periods.

[0048] In the embodiments described above, similarity is calculated based on the difference between the frequency distribution and the probability distribution. The L2 norm may be used instead of the difference (L1 norm). KL (Kullback-Leibler) divergence or JS (Jensen-Shannon) divergence may be used to calculate the similarity of the probability distributions.

[0049] In the embodiments described above, the wind condition observation period is determined based on the similarity of any one of the following: the frequency distribution of wind speed occurrences, the frequency distribution of wind direction and wind speed occurrences, or the average wind speed. The wind condition observation period may also be determined by combining these. For example, the wind condition observation period may be determined using a similarity obtained by assigning predetermined weights to the similarity of the frequency distribution of occurrences, the frequency distribution of wind direction and wind speed occurrences, and the average wind speed.

[0050] The present invention can take on various other embodiments, and furthermore, various modifications such as omissions and substitutions can be made without departing from the spirit of the invention. These embodiments and their variations are included in the scope and spirit of the invention as described herein, and are included in the scope of the invention and its equivalents as described in the claims. [Explanation of Symbols]

[0051] 100 Wind Condition Observation Period Determination Support Device 111 Settings Section 112 Distribution calculation section 112A Average wind speed calculation section 113,113A Similarity calculation unit 114 Wind Condition Observation Period Determination Section 115 Learning Department 116 Prediction Section 121 Configuration Data 128 Programs 130,130B forecast data 140 observation data 158 Predictive Models

Claims

1. A distribution calculation unit calculates the frequency distribution of wind speed, the probability distribution of wind speed, the frequency distribution of both wind speed and wind direction, or the probability distribution of both wind speed and wind direction, based on wind condition forecast data for a first period within the wind condition forecast data for a target area for which wind conditions are to be predicted, and wind condition forecast data for a second period shorter than the first period at the same target area. A similarity calculation unit calculates similarity based on the difference between the frequency distribution of wind speed, the probability distribution of wind speed, the frequency distribution of both wind speed and wind direction, or the probability distribution of both wind speed and wind direction in the wind condition forecast data for the first period and the wind condition forecast data for the second period. The system includes a wind condition observation period determination unit that outputs the second period in which the similarity is maximized. A device to support the determination of the wind condition observation period.

2. An average wind speed calculation unit calculates the average wind speed based on wind condition forecast data for a first period within the wind condition forecast data for the target area for which wind conditions are to be predicted, and wind condition forecast data for a second period shorter than the first period at the same target area. A similarity calculation unit calculates similarity based on the difference between the average wind speed in the wind condition forecast data for the first period and the average wind speed in the wind condition forecast data for the second period. The system includes a wind condition observation period determination unit that outputs the second period in which the similarity is maximized. A device to support the determination of the wind condition observation period.

3. The system includes a prediction unit that calculates wind condition forecast data for the second period at the target site based on the observation data for the second period at the surrounding area of ​​the target site, using a prediction model which is a machine learning model generated using training data in which the observation data for the second period at the target site is used as explanatory variables and the wind condition forecast data for the second period at the target site is used as the dependent variable. A wind condition observation period determination support device according to claim 1 or 2.

4. With regard to the wind condition forecast data for the first period, the first period is defined as one year. A wind condition observation period determination support device according to claim 1 or 2.

5. Regarding the wind condition forecast data for the second period, the second period is defined as less than one year. A wind condition observation period determination support device according to claim 1 or 2.

6. The wind condition observation period determination support device, A step of calculating the frequency distribution of wind speed, the probability distribution of wind speed, the frequency distribution of both wind speed and wind direction, or the probability distribution of both wind speed and wind direction, based on the wind condition forecast data for a first period in the wind condition forecast data for the target area for which wind conditions are to be predicted, and the wind condition forecast data for a second period shorter than the first period in the target area. A step of calculating similarity based on the difference between the frequency distribution of wind speed, the probability distribution of wind speed, the frequency distribution of both wind speed and wind direction, or the probability distribution of both wind speed and wind direction in the wind condition forecast data for the first period and the wind condition forecast data for the second period. The steps include: outputting the second period for which the similarity is maximized; and executing the following: A method for supporting the determination of wind condition observation periods.

7. The wind condition observation period determination support device, A step of calculating the average wind speed based on wind condition forecast data for a first period in the wind condition forecast data for the target area for which wind conditions are to be predicted, and wind condition forecast data for a second period shorter than the first period in the target area. A step of calculating similarity based on the difference between the average wind speed in the wind condition forecast data for the first period and the average wind speed in the wind condition forecast data for the second period, The steps include: outputting the second period for which the similarity is maximized; and executing the following: A method for supporting the determination of wind condition observation periods.

Images