Information processing apparatus and information processing method

Abstract

To reduce an imbalance shortage amount when self-consignment of power generated by a solar panel is performed by a method different from a conventional method.SOLUTION: An information processor (for example, an adjustment device 2A) for determining a planned value of power to be applied for in advance when self-consignment of power generated by a solar panel is performed derives the planned value so as to reduce an imbalanced shortage amount in the self-consignment of the power on the basis of a prediction value of the power acquired from a prediction model (ML) generated in advance. The information processing apparatus derives an approximate curve approximately representing a correspondence relationship between the degree of clear weather and the adjustment coefficient based on the adjustment coefficient derived in at least a partial range of the degree of clear weather. In a range in which the approximate curve is present, the information processing device derives an approximate value of the adjustment coefficient corresponding to the fine weather according to the approximate curve, and derives a value obtained by multiplying the predicted value by the approximate value as the planned value.SELECTED DRAWING: Figure 12

Description

[Technical Field]

[0001] One aspect of the present invention relates to an information processing device that determines a planned value of power to be applied for in advance when power generated by a solar panel is self-consignment. [Background technology]

[0002] Various technologies related to the self-consignment of electric power have been proposed. For example, Patent Document 1 below discloses a method for controlling the operation of a power generation device and related components so that the electric power generated by the power generation device can be self-consignment according to a planned value. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2021-141769 Summary of the Invention [Problem to be solved by the invention]

[0004] An object of one aspect of the present invention is to reduce the imbalance shortage amount when power generated by a solar panel is self-consignment, by using a method different from the conventional method. [Means for solving the problem]

[0005] An information processing device according to one aspect of the present invention is an information processing device that determines a planned value of power to be applied for in advance when power generated by a solar panel is self-consignment, and an imbalance shortage amount in the self-consignment of power is determined as a value obtained by subtracting an actual value of the power from the planned value when the actual value of the power is lower than the planned value, and the information processing device derives the planned value based on a predicted value of the power obtained from a prediction model generated in advance so as to reduce the imbalance shortage amount, and a coefficient for converting the predicted value to the planned value is referred to as an adjustment coefficient, and a ratio of a forecast value of the amount of solar radiation to a theoretical maximum value of the amount of solar radiation in clear weather is referred to as a degree of weather clarity, and the information processing device calculates an adjustment coefficient based on the actual value and The system derives the degree of weather at the installation site of the solar panel for each time point belonging to time series data showing the actual value and the predicted value, classifies the pairs of actual value and predicted value at each time point belonging to the time series data into a plurality of groups according to the degree of weather, derives the adjustment coefficient for each of the plurality of groups classified according to the degree of weather, derives an approximation curve that approximately represents the correspondence between the degree of weather and the adjustment coefficient based on the adjustment coefficient derived for at least a part of the range of the degree of weather, derives an approximation of the adjustment coefficient corresponding to the degree of weather according to the approximation curve within the range in which the approximation curve exists, and derives the value obtained by multiplying the predicted value by the approximation of the adjustment coefficient as the planned value.

[0006] An information processing method according to one aspect of the present invention is an information processing method for determining a planned value of power to be applied for in advance when power generated by a solar panel is self-consignment, wherein an imbalance shortage amount in the self-consignment of power is determined as a value obtained by subtracting an actual value of the power from the planned value when the actual value of the power is lower than the planned value, the information processing method includes a step of deriving the planned value based on a predicted value of the power obtained from a prediction model generated in advance so as to reduce the imbalance shortage amount, a coefficient for converting the predicted value to the planned value is referred to as an adjustment coefficient, and a ratio of a forecast value of the amount of solar radiation to a theoretical maximum value of the amount of solar radiation in clear weather is referred to as a degree of weather clarity, and the information processing method includes a step of deriving the planned value based on a time series data indicating the actual value and the predicted value. the adjustment coefficient for each of the groups classified according to the degree of weather; deriving an approximation curve that approximately represents the correspondence between the degree of weather and the adjustment coefficient based on the adjustment coefficient derived for at least a part of the range of the degree of weather; and, within the range where the approximation curve exists, deriving an approximation of the adjustment coefficient corresponding to the degree of weather according to the approximation curve, and deriving the planned value by multiplying the predicted value by the approximation of the adjustment coefficient. [Effects of the Invention]

[0007] According to one aspect of the present invention, it is possible to reduce the imbalance shortage amount when power generated by a solar panel is self-consignment by using a method different from the conventional method. [Brief explanation of the drawings]

[0008] [Figure 1] 1 shows an example of the configuration of an information processing system according to a reference embodiment. [Figure 2] 1 is a flowchart illustrating a process flow in a reference embodiment. [Figure 3] 1 shows an example of the configuration of an information processing system according to a first embodiment. [Figure 4] FIG. 10 is a diagram for explaining the imbalance shortage amount, showing an example in which the actual value is lower than the predicted value. [Figure 5] FIG. 10 is a diagram for explaining the imbalance shortage amount, showing an example in which the actual value exceeds the predicted value. [Figure 6] An example of data to be adjusted is shown below. [Figure 7] 10 shows an example of data generated based on data to be adjusted. [Figure 8] 10 shows an example of data further generated based on the data to be adjusted. [Figure 9] 10 shows an example of an adjustment coefficient table. [Figure 10] 10 is a graph showing an example of the relationship between the degree of sunny weather and the adjustment coefficient obtained for a certain solar panel. [Figure 11] 10 is a flowchart illustrating a processing flow up to generation of an adjustment coefficient table in the first embodiment. [Figure 12] 10 shows an example of the configuration of an information processing system according to a second embodiment. [Figure 13] 10 shows an example of a graph including an approximate curve that approximately represents the correspondence relationship between the degree of clearness and the adjustment coefficient. [Figure 14] Examples of experimental results performed by the inventors of the present application are shown below. DETAILED DESCRIPTION OF THE INVENTION

[0009] [Reference form] Prior to the description of the first embodiment, a reference embodiment will be described. For the sake of convenience, components having the same functions as those described in the reference embodiment will be denoted by the same reference numerals in the following embodiments, and their descriptions will not be repeated. For the sake of brevity, descriptions of matters similar to those in the publicly known art will also be omitted as appropriate.

[0010] Unless otherwise specified, the components and values ​​described in this specification are merely examples. Therefore, for example, unless otherwise specified, the positional and connection relationships of the components are not limited to the examples in the drawings. The processing flow of the flowcharts is also not limited to the examples in the drawings unless otherwise specified.

[0011] (Configuration example of information processing system 100) 1 shows an example configuration of an information processing system 100 according to a reference embodiment. The information processing system 100 includes a learning device 1, a solar radiation measuring device 12, a temperature measuring device 13, an RTC (Real Time Clock) 14, and a power generation measuring device 15. The learning device 1 may also be referred to as a model generation device.

[0012] The solar radiation measuring device 12 is installed around a solar panel included in a PV (Photovoltaic Power Generation) system (not shown). The solar radiation measuring device 12 measures the amount of solar radiation (hereinafter referred to as "solar radiation") around the solar panel and supplies the measured value of the solar radiation to the learning device 1. In this specification, the solar radiation is exemplified as the total solar radiation on the horizontal surface of the solar panel.

[0013] The temperature measuring device 13 is installed around the solar panel. It measures the temperature around the solar panel (hereinafter referred to as “temperature”) and supplies the measured temperature value to the learning device 1.

[0014] The RTC 14 has a timekeeping function. The RTC 14 supplies time information representing the current time obtained by the timekeeping function to the learning device 1. As an example, the RTC 14 supplies time information expressed in "year / month / day / hour / minute" format ("yyyy / mm / dd / hh / mm" format) to the learning device 1. The time may be used as a key in a data structure. The time resolution of the RTC 14 may be selected arbitrarily.

[0015] The power generation amount measuring device 15 is, for example, built into a solar panel, and in this specification, the value of the power generated by the solar panel is referred to as the "power generation amount." The power generation amount measuring device 15 measures the power generation amount of the solar panel and supplies the measured power generation amount to the learning device 1. In other words, the power generation amount measuring device 15 supplies the performance value (actual value) of the power generation amount to the learning device 1. In this specification, the performance value of the power generation amount is represented as P_true.

[0016] The learning device 1 generates performance data indicating the amount of solar radiation, temperature, and power generation amount at each time point by associating the time point information acquired from the RTC 14 with the amount of solar radiation acquired from the solar radiation measuring device 12, the temperature acquired from the temperature measuring device 13, and the amount of power generation acquired from the power generation measuring device 15. The performance data is an example of time-series data indicating the amount of solar radiation, temperature, and power generation amount at each time point.

[0017] The learning device 1 generates a prediction model ML that outputs a predicted value of the amount of power generated by the solar panel. In this specification, the predicted value of the amount of power generated is referred to as P_predict. The prediction model ML may be any model that can output P_predict as a response variable based on any explanatory variables.

[0018] In the reference embodiment, a case where the amount of solar radiation and the temperature are the explanatory variables is exemplified. The learning device 1 generates a prediction model ML by performing machine learning using P_true at each time point belonging to the actual data as a correct answer. The prediction model ML generated in this manner can output P_true at a certain time point, for example, based on the amount of solar radiation and the temperature at that time point.

[0019] (Example of processing flow in information processing system 100) 2 is a flowchart illustrating the flow of processing in the information processing system 100. In step S1, the solar radiation measuring instrument 12 measures the amount of solar radiation and supplies the measured value of the solar radiation to the learning device 1. In step S2, the temperature measuring instrument 13 measures the temperature and supplies the measured value of the temperature to the learning device 1. In step S3, the RTC 14 acquires time point information and supplies the time point information to the learning device 1. As described above, the solar radiation measuring instrument 12 and the temperature measuring instrument 13 acquire the explanatory variables described above.

[0020] In step S4, the power generation amount measuring device 15 measures P_true, which is the actual value of the power generation amount, and supplies P_true to the learning device 1. The power generation amount measuring device 15 obtains the correct value of the above-mentioned objective variable.

[0021] The above-mentioned performance data can be obtained by the above steps S1 to S4. Then, in step S5, the learning device 1 generates a prediction model ML based on the performance data. The learning device 1 may generate the prediction model ML based on the performance data by executing any type of machine learning algorithm.

[0022] The machine learning algorithm according to one embodiment of the present invention is not particularly limited as long as it can solve a regression task, and examples of the machine learning algorithm include neural networks (NN), support vector regression (SVR), lasso regression, and ridge regression.

[0023] In the example of the reference embodiment, the learning device 1 divides the performance data into learning data and verification data. In step S5, the learning device 1 applies a machine learning algorithm to the learning data to generate a prediction model ML.

[0024] Next, in step S6, the learning device 1 determines whether or not to terminate the series of processes shown in Fig. 2. Specifically, the learning device 1 determines whether or not a termination condition for the processes is satisfied. As an example, the learning device 1 supplies verification data to the prediction model ML generated in step S5, thereby deriving an index value indicating the quality of the prediction model ML.

[0025] The index value is not particularly limited as long as it is used in the field of machine learning to evaluate the prediction accuracy in regression tasks. The index value may be any statistical quantity derived based on the mean absolute error (MAE) or the root mean square error (RMSE), etc. In the examples herein, the higher the index value, the better the prediction accuracy of the prediction model ML.

[0026] As an example, a case will be described in which the "processing termination condition" is set as "the index value is equal to or greater than a predetermined threshold value." In this case, the learning device 1 compares the index value with the threshold value. If the index value is equal to or greater than the threshold value, the learning device 1 determines that the processing termination condition is satisfied (YES in step S6). In this case, the learning device 1 terminates the series of processes shown in FIG. 2.

[0027] On the other hand, if the index value is less than the threshold value, the learning device 1 determines that the end condition of the process is not satisfied (No in step S6). In this case, the process returns to step S1. Therefore, the above-described processes are repeated until the result in step S6 becomes YES.

[0028] Note that, during the repetition of the series of processes in the example of Figure 2, hyperparameter tuning of the machine learning algorithm may be performed. In addition, a machine learning algorithm that is considered appropriate may be selected from among multiple machine learning algorithms using a predetermined optimization method. This ensures the quality of the predictive model ML obtained upon completion of the process of Figure 2.

[0029] [Embodiment 1] In the first embodiment, an example will be described in which a prediction model ML generated in advance by the learning device 1 of the reference embodiment is used. That is, in the first embodiment, an example will be described in which a prediction model ML is used in the operation phase.

[0030] 3 shows an example of the configuration of an information processing system 200 according to the first embodiment. Like the information processing system 100 described above, the information processing system 200 includes a solar radiation measuring device 12, a temperature measuring device 13, an RTC 14, and a power generation measuring device 15. The information processing system 200 includes a prediction model ML generated in advance by the learning device 1. Therefore, the prediction model ML can output P_predict, as described in the reference embodiment.

[0031] The information processing system 200 further includes an adjustment device 2. The adjustment device 2 acquires P_predict from the prediction model ML. The adjustment device 2 is an information processing device that derives adjustment coefficients, which will be described later. The information processing system 200 further includes a weather information acquisition unit 16 and a location information acquisition unit 17. The adjustment device 2 can generate an adjustment coefficient table TB, which will be described later, based on data acquired from each component of the information processing system 200.

[0032] (Explanation of imbalance shortage) Before describing specific processing in the information processing system 200, the imbalance deficiency will be described with reference to Figures 4 and 5. Figures 4 and 5 show schematic curves that represent the time transition of power generated by solar panels. When a business has a PV system as a private power generation facility, it is conceivable that the power generated by the solar panels of the PV system will be self-consigned to another electrical facility of the business (e.g., a factory).

[0033] When self-delivery is to be carried out, it is necessary to apply in advance to a predetermined institution for the planned value of the power generation amount. In this specification, the planned value of the power generation amount is denoted as P_plan. In the self-delivery described in this specification, the power exceeding the planned value shall not be used for self-delivery.

[0034] The deviation between P_plan, which is the planned value in self-delivery, and P_true, which is the actual value, is called imbalance. Since imbalance has an adverse impact on the power company, it is desirable to determine P_plan in accordance with the actual power generation capacity of the solar panels. That is, it is desirable to determine P_plan as close as possible to P_true.

[0035] As the simplest example, it is conceivable to directly adopt P_predict output by the prediction model ML as P_plan. That is, P_plan = P_predict …(1) It is conceivable to determine P_plan as such.

[0036] However, the prediction model ML does not necessarily output highly accurate P_predict. For example, when the weather is not clear, there is a possibility that the accuracy of P_predict will decrease compared to when the weather is clear.

[0037] In FIG. 4, the case where the actual value is less than the predicted value, that is, the case where P_true < P_predict, is illustrated. Reference numeral 410 in FIG. 4 schematically shows the relationship among P_true, P_predict, and P_plan when P_plan is determined as in the above formula (1).

[0038] In the example of reference numeral 410, P_true < P_plan = P_predict The relationship holds. That is, in the example of reference numeral 410, the actual value is less than the planned value.

[0039] In this specification, the imbalance shortage amount in self-delivery is denoted as IL. IL is defined as the value obtained by subtracting P_true from P_plan when P_true < P_plan. That is, IL is IL = P_plan - P_true (where P_true < P_plan) …(2) defined as such.

[0040] When IL occurs, the power company needs to separately generate electricity to compensate for IL. Therefore, when IL occurs, a fine is imposed on the operator performing self-delivery. Since the burden on the power company increases as IL increases, the amount of the fine is determined to increase as IL increases. From this, it is desirable to make IL as small as possible.

[0041] Therefore, it is conceivable to determine P_plan by performing some arithmetic processing on P_predict. As an example, it is conceivable to correct (adjust) P_predict and adopt the corrected value as P_plan.

[0042] In this specification, the correction coefficient for P_predict is denoted as AC. AC may also be referred to as an adjustment coefficient. In the following description, P_plan = AC × P_predict …(3) It is assumed that P_plan is determined as such. That is, it is assumed that the value obtained by multiplying P_predict by AC is defined as P_plan.

[0043] It can be said that AC is a coefficient for converting P_predict to P_plan. Since AC is a coefficient introduced for self-delivery, it is also referred to as a delivery coefficient. AC in this specification is assumed to satisfy 0 ≤ AC ≤ 1.

[0044] Reference numeral 420 in FIG. 4 schematically shows the relationship between P_true, P_predict, and P_plan when P_plan is determined according to the above-mentioned formula (3). The example of reference numeral 420 shows a case where a relatively large AC is adopted. Therefore, in the example of reference numeral 420, P_true <P_plan=AC×P_predict<P_predict The relationship is established.

[0045] In the example of reference numeral 420, the actual value is also lower than the planned value. However, in the example of reference numeral 420, by introducing an AC smaller than 1, the IL is reduced compared to the example of reference numeral 410. Therefore, by adopting an appropriate AC, the IL can be reduced.

[0046] The example of reference numeral 430 in FIG. 4 is a paired example with reference numeral 420. Unlike the example of reference numeral 420, the example of reference numeral 430 shows a case where a relatively small AC is employed. Therefore, in the example of reference numeral 430, P_plan=AC×P_predict <P_true<P_predict The relationship is established.

[0047] In the example of reference numeral 430, the actual value exceeds the planned value, so no IL occurs. If an excessively small AC is selected, the planned value is set smaller than the actual value, so that the penalty resulting from IL can be avoided.

[0048] However, in the example of reference numeral 430, the power P_true generated by the solar panel that exceeds P_plan cannot be used for self-transportation. In other words, in the example of reference numeral 430, the surplus power (P_true - P_plan) is wasted. In this way, if an AC that is too small is adopted, there is a risk of wasting surplus power. Therefore, adopting an AC that is too small is not necessarily a good idea.

[0049] Figure 5 is a diagram paired with Figure 4. Figure 5 illustrates a case where the actual value exceeds the predicted value, i.e., where P_true>P_predict. Reference numeral 510 in Figure 5 schematically illustrates the relationship between P_true, P_predict, and P_plan when P_plan is determined according to the above-mentioned formula (1).

[0050] In the example of code 510, P_true>P_plan=P_predict The following relationship holds. In the example of reference numeral 510, the actual value exceeds the planned value, so IL does not occur. However, even in the example of reference numeral 510, surplus power occurs.

[0051] Reference numeral 520 in FIG. 5 schematically shows the relationship between P_true, P_predict, and P_plan when P_plan is determined according to the above-mentioned formula (3). The example of reference numeral 520 shows a case where a relatively large AC is adopted. Therefore, in the example of reference numeral 520, P_plan=AC×P_predict <P_predict<P_true The relationship is established.

[0052] In the example of reference numeral 520, the actual value exceeds the planned value, so IL does not occur. However, in the example of reference numeral 520, the surplus power is increased compared to the example of reference numeral 510.

[0053] In conventional technology, a fixed value determined in advance based on, for example, the operator's experience has generally been adopted as the AC. However, as described above, the appropriate AC may vary depending on the relationship between P_true and P_predict. Therefore, by optimizing the AC while taking into account the effects of weather, for example, it is possible to obtain a more appropriate AC than conventional methods.

[0054] (Example of processing by the information processing system 200) Therefore, as described above, the information processing system 200 is provided with the weather information acquisition unit 16 and the location information acquisition unit 17. The weather information acquisition unit 16 acquires weather forecast information provided by a weather forecasting service. The weather information acquisition unit 16 supplies the acquired weather forecast information to the adjustment device 2.

[0055] In the example of the first embodiment, the weather forecast information includes a forecast value of the amount of solar radiation at the installation location of the solar panel. In this specification, the forecast value is referred to as L_forecast. On the other hand, in this specification, the theoretical maximum value of the amount of solar radiation on a clear day is referred to as L_theory. L_theory may be derived using any known theoretical formula.

[0056] In this specification, the ratio of L_forecast to L_theory is referred to as the clear weather index (see JP 2013-164286 A). In this specification, the clear weather index is represented as KS. As mentioned above, KS is KS=L_forecast / L_theory …(4) KS satisfies 0≦KS≦1. When L_forecast=L_theory, KS takes the maximum value of 1.

[0057] The location information acquisition unit 17 acquires location information indicating the installation location of the solar panel. The location information acquisition unit 17 may acquire the location information using, for example, a global positioning system (GPS). In the example of the first embodiment, the location information includes information indicating the latitude and longitude of the installation location of the solar panel. The location information acquisition unit 17 supplies the acquired location information to the adjustment device 2.

[0058] The adjustment device 2 generates (prepares) time series data indicating P_predict, P_true, L_forecast, and latitude and longitude at each time point by associating the time point information acquired from the RTC 14 with P_predict acquired from the prediction model ML, P_true acquired from the power generation measuring device 15, L_forecast acquired from the weather information acquisition unit 16, and latitude and longitude acquired from the location information acquisition unit 17. In this specification, this time series data is also referred to as adjustment target data DT. The adjustment target data DT is an example of time series data indicating the relationship between P_true and P_predict at each time point.

[0059] FIG. 6 shows an example of the data to be adjusted DT. In the example of FIG. 6, each numerical value for the daytime (08:30 to 15:30) of each day is shown. In the example of FIG. 6, the notation "lati_japan" represents the latitude of the installation location of the solar panels, and the notation "long_japan" represents the longitude of the installation location. In the example of FIG. 6, each numerical value at each time point is shown in 30-minute increments. As shown in FIG. 6, P_predict, P_true, and L_forecast can change over time. On the other hand, since the installation location of the solar panels remains unchanged, the latitude and longitude are constant values ​​regardless of time.

[0060] Next, the adjustment device 2 derives the above-mentioned L_theory for each time point belonging to the adjustment target data DT. Then, the adjustment device 2 calculates KS at each time point according to the above-mentioned formula (4). The adjustment device 2 generates data DT2 by adding the derived L_theory and KS to the adjustment target data DT.

[0061] 7 shows an example of data DT2 generated based on adjustment target data DT. As shown in Fig. 7, L_theory may change over time. Therefore, KS may also change over time.

[0062] Generally, the larger KS is, the higher the accuracy of P_predict is. This is because the larger KS is, the closer the solar radiation conditions on the solar panel are to the theoretically ideal conditions.

[0063] Therefore, the adjustment device 2 classifies multiple pairs of P_true and P_predict at each time point belonging to the data DT2 (in other words, multiple pairs of P_true and P_predict included in the data to be adjusted DT) into multiple groups according to KS. As an example, the adjustment device 2 sorts each row of the data DT2 in descending order of KS. Next, the adjustment device 2 classifies each row after sorting into classes of KS. The adjustment device outputs the classified data DT3.

[0064] Fig. 8 shows an example of data DT3 that is further generated based on the adjustment target data DT. In the example of Fig. 8, 41 classes from class 1 to class 41 are set for the numerical range of KS. In the example of Fig. 8, ·KS=1.000 …Class 1 ·0.975≦KS<1.000…Class 2 ·0.950≦KS<0.975…Class 3 ·0.925≦KS<0.950…Class 4 ·0.900≦KS<0.925…Class 5 ·0.875≦KS<0.900…Class 6 ·0.850≦KS<0.875…Class 7 ·0.825≦KS<0.850…Class 8 ·0.800≦KS<0.825…Class 9 ·0.775≦KS<0.800…Class 10 ·0.750≦KS<0.775…Class 11 ·0.725≦KS<0.750…Class 12 ·0.700≦KS<0.725…Class 13 ·0.675≦KS<0.700…Class 13 (…omitted…) ·0.000≦KS<0.025…Class 41 The numerical range of KS belonging to each class is set as follows.

[0065] In the example of Fig. 8, the step size of KS in each class except for class 1 is set to 0.025. However, the step size is not limited to this example. Therefore, the total number of classes is not limited to the example of Fig. 8.

[0066] In the following description, for example, a group of pairs of P_true and P_predict belonging to class i in data DT3 will be referred to as group i. i is a subscript indicating the class number. Note that in the example of Figure 8, there are no KS values ​​belonging to class 41, so group 41 is an empty set.

[0067] As an example, an objective function related to IL may be defined in the first embodiment. In this case, for example, it is possible to determine (derive) AC by optimizing the objective function.

[0068] As described above, in self-consignment, it is preferable to make IL as small as possible. VP1 = (1 - AC) × P_predict ... (5) The objective function may be defined based on a parameter VP1 defined by: VP1 is a parameter related to AC and may be referred to as a first evaluation parameter.

[0069] As can be seen from the examples of reference numerals 410 and 420 in FIG. 4 above, P_true <P_plan=AC×P_predict<P_predict When the above relationship holds, the amount of decrease in IL due to the introduction of AC is expressed as the right side of equation (5). Therefore, when the above inequality holds, it is preferable that VP1 at each time point is as large as possible.

[0070] In addition, in embodiment 1, VP2=|AC×P_predict-P_true| …(6) Preferably, the objective function is defined further based on another parameter VP2 defined by: VP2 is another parameter related to AC and may be referred to as a second evaluation parameter.

[0071] As described above with reference to the example of reference numeral 430 in FIG. 4 and the example of FIG. 5, in order to effectively utilize the power generated by the solar panels in self-consignment, it is desirable to prevent the generation of unnecessary surplus power. The degree of deviation between P_plan and P_true when AC is introduced is expressed as the right side of equation (6). Therefore, from the perspective of reducing surplus power, it is desirable to make VP2 at each time point as small as possible.

[0072] Therefore, in the first embodiment, a case where the objective function is determined based on both VP1 and VP2 will be exemplified. Hereinafter, the objective function in the first embodiment will be denoted as G. G in the first embodiment is defined as follows: G=VP1_SUM / VP2_SUM …(7) VP1_SUM represents the sum of VP1 at each time point. VP2_SUM represents the sum of VP2 at each time point.

[0073] Maximizing G defined by equation (7) corresponds to making VP1 as large as possible at each time point and making VP2 as small as possible at each time point. The AC that maximizes G can be derived using any optimization method. Examples of optimization methods include gradient descent, Nelder-Mead algorithm, Newton's method, and golden section method. As an example, the adjustment device 2 may use gradient descent to derive the AC that maximizes G. Then, the adjustment device 2 may determine the derived AC as the official AC (final AC) to be used for determining P_plan.

[0074] As described above, in the first embodiment, multiple pairs of P_true and P_predict can be classified into multiple groups according to KS. Therefore, in the first embodiment, it is preferable to derive AC for each of the multiple groups. This allows AC to be derived while taking into account the influence of KS. As an example, the adjustment device 2 may determine a final AC for each of the multiple groups based on the above-mentioned G.

[0075] Next, the adjustment device 2 generates an adjustment coefficient table TB indicating the final AC determined for each of the multiple groups, and outputs the adjustment coefficient table TB. Fig. 9 shows an example of the adjustment coefficient table TB. The example in Fig. 9 shows a representative value of KS in each class and the AC corresponding to the representative value.

[0076] In the first embodiment, there is no pair of P_true and P_predict corresponding to KS≈0. Therefore, in the example of FIG. 9, a dummy value of AC is assigned to each item where KS≈0. The dummy value in the example of FIG. 9 is 0. However, the dummy value is not limited to this example. For example, a predetermined non-zero value may be assigned as the dummy value. As an example, the smallest non-zero value among the multiple ACs derived so far may be assigned as the dummy value.

[0077] FIG. 10 is a graph showing an example of the relationship between KS and AC obtained for a solar panel of a certain PV system. In this graph, the horizontal axis represents KS, and the vertical axis represents AC. In the example of FIG. 10, the positive side of the horizontal axis of the graph is defined as the right side of the graph. Therefore, the left side, which is the opposite side to the right side, is defined as the negative side of the horizontal axis of the graph. The positive side of the vertical axis of the graph is defined as the upper side of the graph. Therefore, the lower side, which is the opposite side to the upper side, is defined as the negative side of the vertical axis of the graph.

[0078] In the example of Fig. 10, unlike the example of Fig. 9, the step size of KS is set to 0.01. In addition, in the example of Fig. 10, unlike the example of Fig. 9, the dummy value of AC is set to a non-zero value. In the example of Fig. 10, the above-mentioned non-zero minimum value is set as the dummy value of AC. The minimum value is approximately 0.4.

[0079] According to the example of Fig. 10, AC has a positive correlation with KS. According to the examples of Fig. 9 and Fig. 10, when KS is close to the maximum value of 1, AC also approaches the maximum value of 1. This means that when KS is close to 1, the relationship P_true ≒ P_predict holds.

[0080] Since AC has a positive correlation with KS, AC tends to decrease as KS decreases. This means that when KS is small enough, the relationship P_true ≒ P_predict no longer holds.

[0081] Therefore, by deriving AC for each class of KS, it becomes possible to determine P_plan according to the actual operating state of the solar panel. For example, the adjustment device 2 may acquire L_forecast for each future point in time at which a planned value should be determined from the weather information acquisition unit 16. Then, the adjustment device 2 may derive KS for each point in time.

[0082] In this case, the adjustment device 2 can determine the AC corresponding to the KS at each time point by referring to the adjustment coefficient table TB. Therefore, the adjustment device 2 can determine the P_plan at each time point using the AC corresponding to each time point.

[0083] (Example of a flowchart in the information processing system 200) 11 is a flowchart illustrating the flow of processing up to the generation of the adjustment coefficient table TB in the information processing system 200. In step S11, the adjustment device 2 prepares adjustment target data (for example, adjustment target data DT in FIG. 6) as described above.

[0084] Next, in step S12, the adjustment device 2 derives the above-mentioned L_theory for each time point belonging to the adjustment target data DT. Then, in step S13, the adjustment device 2 calculates KS (clearness index) at each time point according to the above-mentioned equation (4).

[0085] Next, in step S14, the adjustment device 2 classifies a plurality of pairs of P_true and P_predict included in the data to be adjusted into a plurality of groups according to the KS.

[0086] Next, in step S15, the adjustment device 2 derives an AC (adjustment coefficient) for each of the multiple groups classified according to the KS. That is, the adjustment device 2 derives an AC for each class of the KS. As described above, the AC may be derived by applying any optimization method to a predetermined objective function.

[0087] In step S16, the adjustment device 2 generates an adjustment coefficient table (for example, the adjustment coefficient table TB in FIG. 9) indicating the AC determined for each of the multiple groups, and outputs the adjustment coefficient table.

[0088] By using the adjustment coefficient table obtained by the processing of FIG. 11, as described above, the adjustment device 2 can determine P_plan at each future point in time when a planned value should be determined.

[0089] (effect) As described above, the adjustment device 2 can determine AC according to, for example, KS. That is, unlike conventional techniques, the adjustment device 2 can determine AC after taking into more detailed consideration the impact of weather on the power generation performance of the solar panel. Then, the adjustment device 2 can derive P_plan from P_predict using the AC. Therefore, the adjustment device 2 can determine P_plan based on P_predict so as to reduce IL.

[0090] As described above, according to the first embodiment, it is possible to reduce the IL when self-consigning power generated by solar panels by using a method different from the conventional method.

[0091] As described above, in the first embodiment, an objective function for IL may be determined based on VP1 and VP2. By determining AC through optimization of the objective function, it becomes possible to (i) reduce the penalty due to IL and (ii) reduce the amount of surplus power generated by the solar panels during self-consignment.

[0092] In the first embodiment, AC can be determined for each installation position of a plurality of solar panels. As a result, the more accurate the weather forecast for the installation position of a certain solar panel, the more reliable AC can be obtained for that solar panel.

[0093] (Supplementary information on the first embodiment) The power generation costs of electric power companies generally change over time. Therefore, the fines due to IL also generally change over time. Therefore, for example, a function F(t) may be introduced to represent the time dependency of the fine, where t represents a point in time. F(t) may be set in advance, for example, based on time series data of fines previously imposed on the above-mentioned business operator.

[0094] As an example, F(t) can be mapped to the above-mentioned VP1. VP1_fine=F(t)×VP1 =F(t)×(1-AC)×P_predict …(8) The objective function may be determined based on the parameter VP1_fine defined by:

[0095] Then, F(t) can be mapped to VP2 as described above. VP2_fine=F(t)×VP2 =F(t)×|AC×P_predict-P_true| …(9) The objective function may be defined further based on VP2_fine defined by:

[0096] The objective function in this example is denoted as G2. For example, G2 is G2=VP1_fine_SUM / VP2_fine_SUM …(10) VP1_fine_SUM represents the sum of VP1_fine at each time point. VP2_fine_SUM represents the sum of VP2_fine at each time point.

[0097] Maximizing G2 defined by Equation (10) corresponds to making VP1_fine as large as possible at each time point and making VP2_fine as small as possible at each time point. In this example, AC that maximizes G2 may be derived using any optimization method. By determining AC based on G2, the penalty due to IL can be reduced more effectively.

[0098] [Embodiment 2] FIG. 12 shows an example configuration of an information processing system 200A in embodiment 2. The adjustment device of the information processing system 200A is referred to as adjustment device 2A. The adjustment device 2A can generate a second table TB2, which will be described later. The second table TB2 is a table that forms a pair with the adjustment coefficient table TB. For this reason, the adjustment coefficient table TB may also be referred to as a first table. The second table TB2 may also be referred to as an approximate adjustment coefficient table.

[0099] The adjustment device 2A also executes steps S11 to S15 described in FIG. 11 of the first embodiment. Therefore, the adjustment device 2A can also derive a graph (e.g., FIG. 10 described above) on which the correspondence relationship between KS and AC is plotted. Next, the adjustment device 2A derives an approximate curve that approximately represents the correspondence relationship in at least a partial range of the graph. That is, the adjustment device 2A derives an approximate curve that approximately represents the correspondence relationship between KS and AC based on the AC derived in at least a partial range of KS.

[0100] Fig. 13 shows an example of a graph including an approximation curve that approximately represents the correspondence relationship between KS and AC. Fig. 13 is a diagram that is paired with the above-mentioned Fig. 10. The symbol KK in the example of Fig. 13 represents the approximation curve.

[0101] The range in which the approximate curve should be derived may be set arbitrarily. In the second embodiment, the graph in Fig. 10 illustrates a case in which the approximate curve KK is derived in (i) a range in which a dummy value for AC is not set and (ii) a range in which AC is less than a predetermined threshold. In the example of the second embodiment, the threshold for AC is set to 0.7.

[0102] Any interpolation algorithm may be applied to derive the approximate curve KK. Therefore, the type of the approximate curve KK is not particularly limited. The approximate curve KK in the example of FIG. 13 is derived as a quadratic function.

[0103] The shape of the approximate curve KK is determined according to the distribution of KS and AC in the region from which the approximate curve KK is to be derived. As an example, the approximate curve KK may include a curved portion that is convex downward. Therefore, the approximate curve KK may have a minimum point of AC. As described above, the approximate curve KK in the example of FIG. 13 is derived as a quadratic function. Therefore, the approximate curve KK in the example of FIG. 13 is a curved portion that is convex downward and has only one minimum point of AC.

[0104] As described in the first embodiment, K depends on the forecast value of the amount of solar radiation. Therefore, K depends on the accuracy of the prediction of the amount of solar radiation. Furthermore, AC is derived from the prediction of the amount of solar power generation based on the prediction of the amount of solar radiation. Therefore, AC is also related to the accuracy of the prediction of the amount of solar radiation.

[0105] Generally, when KS is very large, it is thought that there is little cloud cover, so the accuracy of solar radiation forecasts tends to be high. Also, when KS is very small, it is thought that there is a lot of cloud cover, so the accuracy of solar radiation forecasts tends to be slightly higher than when KS is somewhat small. From this, it is thought that there are pairs of KS and AC where the accuracy of solar radiation forecasts is particularly low. Therefore, the above-mentioned approximation curve KK can be a downward convex curve. And the local minimum of AC on the approximation curve KK can be located at the inflection point of the downward convex curve.

[0106] The position of the minimum point of AC on the approximate curve KK is also determined according to the distribution of KS and AC in the target region from which the approximate curve KK is to be derived. For example, KS at the minimum point can be greater than 0 and less than 1. In the example of FIG. 13, KS at the minimum point is approximately 0.35.

[0107] The adjustment device 2A outputs, as a second table TB2, a table showing the correspondence between KS and AC in the graph of Fig. 13. As can be understood from the description of the first embodiment, the P_plan for each future point in time can also be determined by using the second table TB2. In other words, the P_plan can also be determined by using the approximation curve KK.

[0108] In the second embodiment, the adjustment device 2A derives an approximate value of AC corresponding to KS according to the approximate curve KK in the range where the approximate curve KK exists. In this specification, the approximate value of AC derived according to the approximate curve KK is denoted as AC_KK.

[0109] In the range where the approximate curve KK exists, P_plan can be determined by using AC_KK instead of AC. In the second embodiment, in the range where the approximate curve KK exists, the adjustment device 2A derives the value obtained by multiplying P_predict by AC_KK as P_plan.

[0110] That is, in the range where the approximate curve KK exists, the adjustment device 2A P_plan=AC_KK×P_predict …(11) P_plan is determined according to the following formula (11): Formula (11) corresponds to the above formula (3) in the first embodiment.

[0111] On the other hand, in a range where the approximation curve KK does not exist, the adjustment device 2A determines P_plan by the same method as in embodiment 1. That is, in a range where the approximation curve KK does not exist, the adjustment device 2A determines P_plan according to equation (3). Specifically, in a range where the approximation curve KK does not exist, the adjustment device 2A derives P_plan as a value obtained by multiplying P_predict by AC.

[0112] Fig. 14 shows an example of the results of a demonstration conducted by the inventors of the present application (hereinafter abbreviated as "the inventors"). The inventors derived the values ​​shown in Fig. 14 from data obtained at a certain facility using the P_plan determination method in the reference embodiment, the P_plan determination method in the first embodiment, and the P_plan determination method in the second embodiment.

[0113] Each power in Figure 14 is expressed in the unit of electric energy "kWh". However, as will be clear to those skilled in the art, electric energy can be converted into electric power. For example, the unit of electric energy "kWh" can be converted into the unit of electric power "kW".

[0114] 14 is defined as the value obtained by dividing IL by P_plan. Therefore, the loss degree is an index value that represents IL normalized by P_plan. Using the loss degree makes it easier to compare multiple cases with different P_plans.

[0115] 14, in the reference embodiment, AC is fixed at a constant value of 0.7. Therefore, in the reference embodiment, unlike the first and second embodiments, P_plan cannot be determined by fully considering the influence of weather on the power generation performance of the solar panel. For this reason, the reference embodiment generates a higher degree of loss than the first and second embodiments.

[0116] In the first embodiment, unlike the reference embodiment, AC can be determined according to KS. In the first embodiment, AC is determined according to KS, for example, according to the graph in FIG. 10. Therefore, in the first embodiment, P_plan can be determined by fully considering the influence of weather on the power generation performance of the solar panel, unlike the reference embodiment. As a result, a lower degree of loss is obtained in the first embodiment compared to the reference embodiment.

[0117] In the second embodiment, unlike the first embodiment, AC according to KS can be determined using the above-described approximation curve. In the first embodiment, AC according to KS is determined, for example, according to the graph in Fig. 13. As shown in Fig. 14, the second embodiment achieves a lower degree of loss than the first embodiment.

[0118] In the graph of Fig. 10, for example, the variation of AC with respect to KS is relatively large around a clear weather index of 0.5. On the other hand, as shown in Fig. 13, by using the approximation curve KK, the variation can be eliminated. From this, it can be considered that a lower loss degree is obtained in the second embodiment than in the first embodiment.

[0119] Furthermore, by using the approximation curve KK, it is possible to derive an appropriate AC according to KS even in a range of KS where the number of data showing the correspondence between KS and AC is small. From this point of view, according to the second embodiment, it is possible to determine P_plan more appropriately than in the first embodiment.

[0120] (Supplementary information on the second embodiment) (1) In the second embodiment, the range in which the approximate curve is to be derived is set by setting a threshold value for AC. As another example, the range in which the approximate curve is to be derived may be set by setting a threshold value for KS.

[0121] For example, in the graph of Fig. 10, an approximate curve KK may be derived in (i) the range where no dummy value of AC is set and (ii) the range where KS is equal to or greater than a predetermined threshold value. As an example, the threshold value of KS may be set to 0.9.

[0122] (2) In the second embodiment, the case where the approximate curve KK is not derived in the range where AC is equal to or greater than a predetermined threshold has been exemplified. However, in the range where AC is equal to or greater than a predetermined threshold, an approximate curve other than the approximate curve KK may be derived. In the example of Fig. 10, the other approximate curve may be an upward convex curve (e.g., an upward convex quadratic function).

[0123] (3) In the second embodiment, the case where the approximate curve KK is derived in a partial range of the graph in Fig. 10 has been exemplified. However, the approximate curve KK may be derived in the entire range of KS in the graph in Fig. 10.

[0124] [Software implementation example] The functions of the information processing systems 100 to 200A (hereinafter referred to as "devices" for convenience) can be realized by a program that causes a computer to function as the device, and a program that causes a computer to function as each control block of the device (in particular, the learning device 1 and the adjustment devices 2 to 2A).

[0125] In this case, the device includes a computer having at least one control device (e.g., a processor) and at least one storage device (e.g., a memory) as hardware for executing the program. The control device and storage device execute the program, thereby realizing the functions described in each of the above embodiments.

[0126] The program may be non-transitory and may be recorded on one or more computer-readable recording media. The recording media may or may not be included in the device. In the latter case, the program may be supplied to the device via any wired or wireless transmission medium.

[0127] In addition, some or all of the functions of the control blocks can be realized by logic circuits. For example, an integrated circuit in which a logic circuit that functions as each of the control blocks is formed is also included in the scope of one aspect of the present invention. In addition, the functions of the control blocks can be realized by, for example, a quantum computer.

[0128] The processes described in the above embodiments may be executed by AI (Artificial Intelligence). In this case, the AI ​​may run on the control device or on another device (for example, an edge computer or a cloud server).

[0129] 〔summary〕 An information processing device according to a first aspect of the present invention is an information processing device that determines a planned value of power to be applied for in advance when power generated by a solar panel is self-consignment, and an imbalance shortage amount in the self-consignment of power is determined as a value obtained by subtracting an actual value of the power from the planned value when the actual value of the power is lower than the planned value, and the information processing device derives the planned value based on a predicted value of the power obtained from a prediction model generated in advance so as to reduce the imbalance shortage amount, and a coefficient for converting the predicted value to the planned value is referred to as an adjustment coefficient, and a ratio of a forecast value of the amount of solar radiation to a theoretical maximum value of the amount of solar radiation in clear weather is referred to as a degree of weather clarity, and the information processing device calculates a value between the actual value and the The system derives the degree of weather at the installation site of the solar panel for each time point belonging to time series data showing the actual value and the predicted value, classifies the pairs of actual value and predicted value at each time point belonging to the time series data into a plurality of groups according to the degree of weather, derives the adjustment coefficient for each of the plurality of groups classified according to the degree of weather, derives an approximation curve that approximately represents the correspondence between the degree of weather and the adjustment coefficient based on the adjustment coefficient derived for at least a part of the range of the degree of weather, derives an approximation of the adjustment coefficient corresponding to the degree of weather according to the approximation curve within the range in which the approximation curve exists, and derives the value obtained by multiplying the predicted value by the approximation of the adjustment coefficient as the planned value.

[0130] The information processing device according to aspect 2 of the present invention may be configured in such a way that, in the first aspect, in a range where the approximation curve does not exist, the information processing device derives, as the planned value, a value obtained by multiplying the predicted value by the adjustment coefficient.

[0131] In an information processing device according to aspect 3 of the present invention, in aspect 1 or 2, when the horizontal axis of a graph on which the correspondence between the degree of weather and the adjustment coefficient is plotted indicates the degree of weather, the vertical axis of the graph indicates the adjustment coefficient, and the positive side of the vertical axis is defined as the upper side of the graph, and the lower side, which is the opposite side to the upper side, is defined as the negative side of the vertical axis, the approximation curve may include a curved portion that is convex downward.

[0132] In an information processing device according to a fourth aspect of the present invention, in the third aspect, the clarity index at the minimum point of the adjustment coefficient on the approximation curve may be greater than 0 and less than 1.

[0133] In the information processing device according to a fifth aspect of the present invention, in any one of the first to fourth aspects, the adjustment coefficient may be derived by optimizing an objective function relating to the imbalance shortage amount.

[0134] In the information processing device according to the sixth aspect of the present invention, in the fifth aspect, The actual value is denoted as P_true, The predicted value is denoted as P_predict, When the adjustment coefficient is expressed as AC, The objective function may be determined based on VP1 determined by the above equation (5) and VP2 determined by the above equation (6).

[0135] In the information processing device according to aspect 7 of the present invention, in the information processing device according to aspect 6, The sum of VP1 at each time point belonging to the time series data indicating the actual value and the predicted value is denoted as VP1_SUM, The sum of VP2 at each time point belonging to the time series data is denoted as VP2_SUM, When the objective function is denoted as G, G may be determined by equation (7) above.

[0136] The information processing device according to an eighth aspect of the present invention may be configured in any one of the fifth to seventh aspects to derive the adjustment coefficient that gives a maximum value of the objective function.

[0137] An information processing method according to a ninth aspect of the present invention is an information processing method for determining a planned value of power to be applied for in advance when power generated by a solar panel is self-consignment, wherein an imbalance shortage amount in the self-consignment of power is determined as a value obtained by subtracting an actual value of the power from the planned value when the actual value of the power is lower than the planned value, and the information processing method includes a step of deriving the planned value based on a predicted value of the power obtained from a prediction model generated in advance so as to reduce the imbalance shortage amount, a coefficient for converting the predicted value to the planned value is referred to as an adjustment coefficient, and a ratio of a forecast value of the amount of solar radiation to a theoretical maximum value of the amount of solar radiation in clear weather is referred to as a degree of weather clarity, and the information processing method includes a step of deriving the planned value based on a predicted value of the power obtained from a prediction model generated in advance so as to reduce the imbalance shortage amount, the adjustment coefficient for each of the groups classified according to the degree of weather; deriving an approximation curve that approximately represents the correspondence between the degree of weather and the adjustment coefficient based on the adjustment coefficient derived for at least a part of the range of the degree of weather; and, within the range where the approximation curve exists, deriving an approximation of the adjustment coefficient corresponding to the degree of weather according to the approximation curve, and deriving the planned value by multiplying the predicted value by the approximation of the adjustment coefficient.

[0138] [Additional Notes] One aspect of the present invention is not limited to the above-described embodiments, and various modifications are possible within the scope of the claims. Embodiments obtained by appropriately combining the technical means disclosed in different embodiments are also included in the technical scope of one aspect of the present invention. [Explanation of symbols]

[0139] 1 Learning device 2,2A Adjustment device (information processing device that determines the planned value of the power generated by the solar panel) 12 Solar radiation meter 13 Temperature Meter 14 RTC 15 Power generation measuring device 16 Weather Information Acquisition Department 17 Location information acquisition section 100 Information processing system (information processing system including learning device) 200, 200A Information processing system (information processing system including adjustment device) ML predictive models DT adjustment target data (time series data showing actual and predicted values ​​of power generated by solar panels) TB Adjustment Factor Table TB2 Table 2 KK trendline

Claims

1. An information processing device that determines a planned value of power to be applied for in advance when self-consigning power generated by a solar panel, The imbalance deficiency amount in the self-transportation of the power is determined as a value obtained by subtracting the actual value from the planned value when the actual value of the power is lower than the planned value, the information processing device derives the planned value based on a predicted value of the power acquired from a prediction model generated in advance so as to reduce the imbalance shortage amount; A coefficient for converting the predicted value into the planned value is called an adjustment coefficient, The ratio of the forecast value of the amount of solar radiation to the theoretical maximum amount of solar radiation in a clear day is called the degree of clearness, The information processing device includes: deriving the degree of weather at the installation point of the solar panel for each time point belonging to the time series data indicating the actual value and the predicted value; classifying pairs of the actual value and the predicted value at each time point in the time series data into a plurality of groups according to the degree of weather; deriving the adjustment coefficient for each of the plurality of groups classified according to the degree of weather; deriving an approximation curve that approximately represents the correspondence relationship between the degree of weather and the adjustment coefficient based on the adjustment coefficient derived for at least a part of the range of the degree of weather; In the range where the approximation curve exists, an approximation value of the adjustment coefficient corresponding to the degree of weather is derived according to the approximation curve, and the value obtained by multiplying the predicted value by the approximation value of the adjustment coefficient is derived as the planned value.

2. The information processing device according to claim 1 , wherein the information processing device derives, as the planned value, a value obtained by multiplying the predicted value by the adjustment coefficient in a range where the approximation curve does not exist.

3. The horizontal axis of the graph on which the correspondence relationship between the degree of weather and the adjustment coefficient is plotted represents the degree of weather, The vertical axis of the graph represents the adjustment coefficient, The positive side of the vertical axis is defined as the upper side of the graph; When the lower side, which is the opposite side to the upper side, is defined as the negative side of the vertical axis, The information processing device according to claim 1 , wherein the approximation curve includes a curved portion that is convex downward.

4. The information processing apparatus according to claim 3 , wherein the clarity index at the minimum point of the adjustment coefficient on the approximation curve is greater than 0 and less than 1.

5. The information processing device according to claim 1 , wherein the information processing device derives the adjustment coefficient by optimizing an objective function relating to the imbalance deficiency amount.

6. The actual value is denoted as P_true, The predicted value is denoted as P_predict, When the adjustment coefficient is expressed as AC, The following formula: VP1=(1-AC)×P_predict VP1 defined by The following formula: VP2=|AC×P_predict−P_true| VP2 defined by The information processing apparatus according to claim 5 , wherein the objective function is determined based on the following:

7. The sum of VP1 at each time point belonging to the time series data indicating the actual value and the predicted value is denoted as VP1_SUM, The sum of VP2 at each time point belonging to the time series data is denoted as VP2_SUM, When the objective function is denoted as G, The following formula: G=VP1_SUM / VP2_SUM The information processing device according to claim 6 , wherein G is determined by the following equation:

8. The information processing device according to claim 5 , wherein the information processing device derives the adjustment coefficient that gives a maximum value of the objective function.

9. An information processing method for determining a planned value of power to be applied for in advance when self-consigning power generated by solar panels, The imbalance deficiency amount in the self-transportation of the power is determined as a value obtained by subtracting the actual value from the planned value when the actual value of the power is lower than the planned value, the information processing method includes a step of deriving the planned value based on a predicted value of the power acquired from a prediction model generated in advance so as to reduce the imbalance shortage amount; A coefficient for converting the predicted value into the planned value is called an adjustment coefficient, The ratio of the forecast value of the amount of solar radiation to the theoretical maximum amount of solar radiation in a clear day is called the degree of clearness, The information processing method includes: deriving the degree of weather clarity at the installation point of the solar panel for each time point belonging to the time series data indicating the actual value and the predicted value; classifying pairs of the actual value and the predicted value at each time point belonging to the time series data into a plurality of groups according to the degree of weather; deriving the adjustment coefficient for each of the plurality of groups classified according to the degree of weather; deriving an approximation curve that approximately represents the correspondence relationship between the degree of weather and the adjustment factor based on the adjustment factor derived for at least a part of the range of the degree of weather; an information processing method including a step of deriving an approximate value of the adjustment coefficient corresponding to the degree of weather in accordance with the approximate curve in a range where the approximate curve exists, and deriving the value obtained by multiplying the predicted value by the approximate value of the adjustment coefficient as the planned value.

Images