Power generation plan creation program, power generation plan creation method, and power generation plan creation device

The power generation plan creation program uses machine learning to predict power generation and imbalance prices, addressing imbalance risks by optimizing power generation plans, thus enhancing operational efficiency and reducing financial losses in balancing groups and electricity distribution companies.

JP2024165890APending Publication Date: 2024-11-28TOHOKU ELECTRIC POWER
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
JP2023082467
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-05-18
Publication Date
2024-11-28

AI Technical Summary

Technical Problem

Conventional methods for creating power generation plans in balancing groups face challenges in accurately predicting power generation, leading to imbalances that result in increased costs, lost sales opportunities, tight demand, and curtailed renewable energy, making it difficult to create an optimal plan that minimizes imbalance risk.

Method used

A power generation plan creation program and device that utilize machine learning models to predict power generation and imbalance unit prices based on weather forecasts, wide-area reserve margins, and contract prices, allowing for the creation of a power generation plan that minimizes economic and opportunity losses by adjusting predicted power generation amounts.

Benefits of technology

Enables the creation of an optimal power generation plan that reduces imbalance risks, optimizing operations for balancing groups and electricity distribution companies by accurately predicting power generation and pricing, thereby enhancing operational efficiency and reducing financial losses.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024165890000001_ABST
    Figure 2024165890000001_ABST
Patent Text Reader

Abstract

To provide a power generation plan creation program, a power generation plan creation method, and a power generation plan creation device for creating an optimum power generation plan for minimizing an imbalance risk.SOLUTION: The present invention causes a computer to execute: a power generation amount prediction step in which the predictive value of power generation amount at a specific day is calculated using a power generation amount prediction model that takes weather prediction data as input and outputs a power generation amount; an imbalance unit cost prediction step in which weather prediction data at a specific day and a wide-area reserve margin are acquired, and the predictive value of an imbalance unit cost is calculated using an imbalance unit cost prediction model that takes the weather prediction data and the wide-area reserve margin as input and outputs an imbalance unit cost; and a power generation plan creation step in which a correction is added to the predictive value of power generation amount calculated in the power generation amount prediction step on the basis of the predictive value of imbalance unit cost calculated in the imbalance unit cost prediction step, so as to calculate a power generation plan value and create a power generation plan.SELECTED DRAWING: Figure 3
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to a power generation plan creating program, a power generation plan creating method, and a power generation plan creating device. [Background technology]

[0002] With the liberalization of the electric power market, power generation companies and retail electricity companies are focusing on operations that incorporate balancing groups in order to ensure a stable supply of electricity and reduce costs. A balancing group is a system in which multiple retail electricity companies form one group, and multiple power generation companies form one group, and each group enters into a wheeling supply contract with a local electricity transmission and distribution company.

[0003] In order to operate the balancing group appropriately, it is necessary for the operator to accurately predict the amount of power generated by the power generating source to be sold or procured. For solar and wind power generation in particular, predictions are made based on weather forecasts. Power generation forecasts are made daily in accordance with daily demand operations, and in order to create a power generation sales / demand procurement plan to be submitted to the Organization for Cross-regional Coordination of Transmission Operators, Japan, the amount of power generation must be predicted at the latest in the morning of the day before actual demand.

[0004] As a method of forecasting power generation for balancing group operation, specialized businesses provide systems and services that forecast power generation based on weather forecasts, satellite images, and power generation facility specifications. Some of the services that provide information on such power generation forecasts take into account the uncertainty of the forecast using an ensemble method. Balancing group operation businesses are expected to create optimal plan values ​​based on the power generation forecasts provided by specialized businesses.

[0005] Another technique proposed for balancing group operation involves estimating the power generation of a distributed power source equipped with remote measuring equipment to acquire power generation data, and calculating the correlation coefficient between the power generation of a distributed power source equipped with remote equipment and a distributed power source without remote equipment, thereby estimating the power generation of a distributed power source without remote equipment. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] JP 2022-184637 A Summary of the Invention [Problem to be solved by the invention]

[0007] However, in the past, in the operation of balancing groups, specialized businesses have provided systems and services that predict power generation based on weather forecasts, satellite photos, power generation equipment specifications, etc. However, for example, predictions of solar power generation and wind power generation are premised on the occurrence of errors, so there are errors in the prediction of power generation. Then, an imbalance, which is the difference between the planned value and the actual value, occurs due to errors in the prediction of the power generation amount. A state in which the actual demand is insufficient for the power generation amount is called a shortage imbalance, and a state in which the actual demand is excessive for the planned power generation amount is called a surplus imbalance. For balancing group operators, shortage imbalance is a risk of increased costs, and surplus imbalance is a risk of lost sales opportunities. For general power transmission and distribution companies, shortage imbalance is a risk of tight area demand, and surplus imbalance is a risk of suppressing the output of renewable energy. Therefore, it is desirable for balancing group operators to create planned values ​​by adjusting the predicted power generation amount while appropriately predicting the imbalance unit price, which corresponds to the value of electricity in the actual supply and demand, and the area demand.

[0008] In this regard, with the conventional method of creating a power generation plan based on the provided power generation forecast values, relying on the operator's experience and intuition to take into account the degree of tightness in supply and demand, the quality of the power generation plan depends on the ability of the plan creator, making it difficult to create an optimal power generation plan to minimize imbalance risk.

[0009] The disclosed technology has been made in consideration of the above, and aims to provide a power generation plan creation program, a power generation plan creation method, and a power generation plan creation device that create an optimal power generation plan for minimizing imbalance risk. [Means for solving the problem]

[0010] In one aspect of the power generation plan creation program, power generation plan creation method, and power generation plan creation device disclosed in the present application, a computer is caused to execute a power generation amount prediction step of acquiring weather forecast data for a specific day, and calculating a predicted value of the power generation amount on the specific day using a power generation amount prediction model that takes the weather forecast data as an input and outputs a power generation amount; an imbalance unit cost prediction step of acquiring weather forecast data and a wide-area reserve rate for the specific day, and calculating a predicted value of an imbalance unit cost using an imbalance unit cost prediction model that takes the weather forecast data and the wide-area reserve rate as input and outputs an imbalance unit cost; and a power generation plan creation step of calculating a power generation plan value by correcting the predicted value of the power generation amount calculated in the power generation amount prediction step based on the predicted value of the imbalance unit cost calculated in the imbalance unit cost prediction step, thereby creating a power generation plan. Effect of the Invention

[0011] In one aspect, the present invention can create an optimal power generation plan for minimizing an imbalance risk. [Brief description of the drawings]

[0012] [Figure 1] FIG. 1 is a block diagram of a power generation plan creating device. [Diagram 2] FIG. 2 is a diagram illustrating an example of correction to an estimated value of the power generation amount. [Diagram 3] FIG. 3 is a flowchart of a power generation plan creating process performed by the power generation plan creating device according to the embodiment. [Figure 4] FIG. 4 is a flowchart of the power generation amount prediction process performed by the power generation amount prediction unit according to the embodiment. [Diagram 5]FIG. 5 is a flowchart of an imbalance unit price prediction process performed by the imbalance unit price prediction unit according to the embodiment. [Figure 6] FIG. 6 is a flowchart of a power generation plan creation process by the power generation plan creation unit according to the embodiment. [Figure 7] FIG. 7 is a hardware configuration diagram of the power generation plan creating device. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] Hereinafter, the power generation plan creation program, the power generation plan creation method, and the power generation plan creation device disclosed in the present application will be described in detail with reference to the drawings. Note that the power generation plan creation program, the power generation plan creation method, and the power generation plan creation device disclosed in the present application are not limited to the following embodiments. EXAMPLES

[0014] FIG. 1 is a block diagram of a power generation plan creation device. The power generation plan creation device 1 calculates a predicted value of power generation amount at a predetermined time using weather forecast data, and calculates a predicted value of imbalance unit cost at the predetermined time using weather forecast data, wide-area reserve rate data, etc. Then, the power generation plan creation device 1 creates a power generation plan at the predetermined time using the calculated predicted value of power generation amount and the predicted value of imbalance unit cost. The power generation plan creation device 1 generates a power generation plan for each balancing group or for a specific balancing group. The power generation plan creation device 1 can generate a power generation plan every 30 minutes in accordance with fluctuations in the imbalance unit cost.

[0015] The power generation plan creation device 1 is connected to a weather forecast data providing system 2, a wide-area reserve margin information providing system 3, and a contract price providing system 4.

[0016] The weather forecast data providing system 2 provides weather forecast data including the amount of solar radiation, wind speed, etc. for a nationwide 5 km mesh. The weather forecast data providing system 2 provides weather forecast data at least as early as 3:00 a.m. every day.

[0017] The wide-area reserve margin information system 3 is a system managed by the cross-regional organization. The wide-area reserve margin information system 3 provides the wide-area reserve margin in 30-minute intervals every day. The wide-area reserve margin information system 3 also provides the weekly wide-area reserve margin in daily intervals every week.

[0018] Here, we will briefly explain the wide-area reserve margin. The supply region of a general electricity distribution utility is called an area. In order to operate the power supply to an area, multiple areas are operated together as an integrated unit when there is spare capacity in the interconnection lines connecting the areas. A wide-area block refers to this collection of multiple areas operated together. The wide-area reserve margin is the reserve margin in a wide-area block. The reserve margin is the ratio of supply reserve capacity to power usage, and is information that indicates how much supply capacity there is relative to power demand. In other words, the wide-area reserve margin is information that indicates how much supply capacity there is relative to power demand in multiple areas operated together.

[0019] The contract price providing system 4 provides the contract price of electricity determined at the power exchange every 30 minutes every day. In this embodiment, the contract price providing system 4 provides the contract price determined in the market for buying and selling electricity to be generated or sold the next day.

[0020] As shown in FIG. 1, the power generation plan creation device 1 has a power generation plan creation unit 11, an output unit 12, a power generation amount prediction model 13, a power generation amount prediction unit 14, an imbalance unit cost prediction unit 15, an imbalance unit cost prediction model 16, and a database 17.

[0021] Database 17 holds data on past weather forecasts. Database 17 also holds data on actual power generation corresponding to the weather forecasts for each day in the past. Database 17 also holds the weekly wide-area reserve rate for each day in the past and the wide-area reserve rate in 30-minute increments for each day in the past, as well as data on imbalance prices for the weekly wide-area reserve rate for each day in the past and the imbalance prices for the wide-area reserve rate in 30-minute increments for each day in the past. Database 17 also holds data on contract prices in 30-minute increments for each day in the past.

[0022] The power generation prediction model 13 is a machine learning model that inputs weather forecast data and outputs a predicted value of the power generation amount for each of 48 frames obtained by dividing 24 hours into 30-minute intervals. There is no particular limit to the machine learning algorithm used by the power generation prediction model 13. For example, the power generation prediction model 13 can use Light-GBM (Light Gradient Boosting Machine Learning) (registered trademark) as the machine learning algorithm. The power generation prediction model 13 exists for each of a plurality of prediction points for which the power generation amount is to be predicted.

[0023] There are two types of imbalance unit price prediction models 16, one for day-ahead prediction and one for day-to-day prediction. The imbalance unit price prediction model 16 for day-ahead prediction is a machine learning model that outputs a predicted value of the imbalance unit price for every 48 frames for input data created from weather forecast data and data on the weekly wide-area reserve rate. The imbalance unit price prediction model 16 for day-to-day prediction is a machine learning model that outputs a predicted value of the imbalance unit price for every 48 frames for input data created from weather forecast data, data on the wide-area reserve rate in 30-minute units, and the contract price. There is no particular restriction on the machine learning algorithm used for the imbalance unit price prediction model 16, and for example, Light-GBM can be used. The imbalance unit price exists for each area to which the balancing group for which the power generation plan is to be created belongs.

[0024] The imbalance unit price is selected from the higher of the limit kWh price of the adjustment power or the supply-demand tightness compensation imbalance charge. In daily life, the imbalance unit price is the limit kWh price of the adjustment power. The limit kWh price of the adjustment power used to calculate the imbalance charge is the highest value of the adjustment power operated in a wide area. In addition, since the wide area operation of the adjustment power is performed in 15-minute increments, the imbalance charge for each of the 48 frames is calculated by weighting the limit kWh prices of the first 15 minutes and the last 15 minutes with the command value of the adjustment power. However, in the case of specific conditions such as when the output of solar or wind power is suppressed, when a generator such as a thermal power plant is suppressed, or when a blackout occurs, a predetermined imbalance unit price is applied. In other words, the imbalance unit price prediction model 16 actually predicts the imbalance unit price of each frame, which is determined according to the above procedure, from weather forecast data, wide area reserve rate data, etc.

[0025] The power generation prediction unit 14 has information on a plurality of prediction points for predicting the power generation amount in advance. When learning the power generation prediction model 13, the power generation prediction unit 14 acquires past weather forecast data for each prediction point and data on power generation results corresponding to the past weather forecast from the database 17. Then, the power generation prediction unit 14 uses the acquired data on past weather forecasts and data on power generation results corresponding to the past weather forecasts as teacher data to execute learning of each of the power generation prediction models 13 for each prediction point and perform parameter tuning. At this time, the power generation prediction unit 14 performs feature engineering processing on the weather forecast data to extract features such as time difference, moving average, categorization, etc., to generate input data, and executes learning using the input data. In addition, the parameter tuning may involve manual work by an engineer. As a result, the power generation prediction unit 14 generates the learned power generation prediction model 13.

[0026] Furthermore, the power generation amount prediction unit 14 obtains nationwide weather forecast data at 3:00 every day from the weather forecast data providing system 2. Then, when creating an actual power generation plan, the power generation amount prediction unit 14 uses the obtained weather forecast data to perform a day-ahead prediction and a day-to-day prediction of the power generation amount at each prediction point.

[0027] Specifically, the power generation prediction unit 14 extracts weather forecast data for each prediction point from nationwide weather forecast data. In the case of a day-ahead prediction, the power generation prediction unit 14 extracts data as of 3:00 on the day before the actual supply and demand. In the case of a day-to-day prediction, the power generation prediction unit 14 extracts data as of 3:00 on the day of the actual supply and demand.

[0028] Next, the power generation amount prediction unit 14 performs feature engineering processing on the extracted weather forecast data to generate input data. For example, the power generation amount prediction unit 14 generates input data using time difference, moving average, categorization, etc. as features. Then, the power generation amount prediction unit 14 inputs the generated input data to the trained power generation amount prediction model 13 to obtain its output value. Next, the power generation amount prediction unit 14 makes fine adjustments to each output value from the power generation amount prediction model 13 for each prediction point, such as changing a negative value to 0. Thereafter, the power generation amount prediction unit 14 aggregates each output value from the finely adjusted power generation amount prediction model 13 for each prediction point in units of balancing groups to generate a predicted value of the power generation amount. Then, the power generation amount prediction unit 14 outputs the predicted value of the power generation amount to the power generation plan creation unit 11.

[0029] The imbalance unit price prediction unit 15 has in advance information on the area for predicting the imbalance unit price and its representative point. When learning of the imbalance unit price prediction model 16 is executed, the imbalance unit price prediction unit 15 acquires past weather forecast data for each representative point of each area from the database 17. In addition, the imbalance unit price prediction unit 15 acquires information on the imbalance unit price for the past weekly wide-area reserve rate and information on the imbalance unit price for the past wide-area reserve rate in 30-minute increments from the database 17. Furthermore, the imbalance unit price prediction unit 15 acquires the past contract price of electricity in 30-minute increments from the database 17.

[0030] The imbalance unit cost prediction unit 15 extracts features of the weather forecast data by performing feature engineering processes such as time difference, moving average, categorization, etc. on the acquired past weather forecast data. Next, the imbalance unit cost prediction unit 15 generates input data by integrating the features of the weather forecast data and the weekly wide-area reserve rate. Then, the imbalance unit cost prediction unit 15 uses the generated input data and the corresponding imbalance unit cost as teacher data to execute learning of each of the imbalance unit cost prediction models 16 for day-ahead forecast for each area, and performs parameter tuning.

[0031] The imbalance unit price prediction unit 15 also generates input data by integrating the feature quantities of the weather forecast data with the weekly wide-area reserve rate and the contract price. The imbalance unit price prediction unit 15 then uses the generated input data and the corresponding imbalance unit prices as teacher data to execute learning of each of the imbalance unit price prediction models 16 for the day's forecast for each area and performs parameter tuning. Here, the parameter tuning may involve manual work by an engineer. In this way, the imbalance unit price prediction unit 15 generates the learned imbalance unit price prediction models 16 for the day-ahead forecast and the day's forecast.

[0032] Furthermore, the imbalance unit price prediction unit 15 obtains nationwide weather forecast data at 3:00 every day from the weather forecast data providing system 2. Furthermore, the imbalance unit price prediction unit 15 obtains wide-area reserve margin data in 30-minute units every day from the wide-area reserve margin information providing system 3. Furthermore, the imbalance unit price prediction unit 15 obtains weekly wide-area reserve margin data every week from the wide-area reserve margin information providing system 3. Furthermore, the imbalance unit price prediction unit 15 obtains electricity contract prices in 30-minute units every day from the contract price providing system 4.

[0033] In the case of a day-ahead forecast, the imbalance unit price forecasting unit 15 extracts data of actual supply and demand at a representative point of the forecast target area at 3:00 a.m. on the day before from the national weather forecast data. Next, the imbalance unit price forecasting unit 15 performs feature engineering processing on the extracted weather forecast data to extract features of the weather forecast data. Furthermore, the imbalance unit price forecasting unit 15 performs integration processing on the features of the weather forecast data and the weekly wide-area reserve rate to create input data. Next, the imbalance unit price forecasting unit 15 inputs each of the generated input data for each forecast target area to the trained imbalance unit price forecasting model 16 for the day-ahead forecast for each forecast target area. Then, the imbalance unit price forecasting unit 15 obtains an output value from the imbalance unit price forecasting model 16 for the day-ahead forecast for each forecast target area. Next, the imbalance unit price forecasting unit 15 fine-tunes the output value from the imbalance unit price forecasting model 16 for the day-ahead forecast for each forecast target area to create a forecast value of the day-ahead forecast of the imbalance unit price. Thereafter, the imbalance unit cost prediction unit 15 outputs the predicted value of the day-ahead prediction of the imbalance unit cost to the power generation plan creation unit 11 .

[0034] In the case of same-day forecasting, the imbalance unit price prediction unit 15 extracts data at 3:00 a.m. on the day before the actual supply and demand at a representative point of the forecast target area from the nationwide weather forecast data. Next, the imbalance unit price prediction unit 15 performs feature engineering processing on the extracted weather forecast data to extract features of the weather forecast data. Furthermore, the imbalance unit price prediction unit 15 creates input data by integrating the features of the weather forecast data, the weekly wide-area reserve rate, and the contract price. Next, the imbalance unit price prediction unit 15 inputs each of the generated input data for each forecast target area to the trained imbalance unit price prediction model 16 for same-day forecast for each forecast target area. Then, the imbalance unit price prediction unit 15 obtains an output value from the imbalance unit price prediction model 16 for same-day forecast for each forecast target area. Next, the imbalance unit price prediction unit 15 creates a forecast value of the same-day forecast of the imbalance unit price by fine-tuning the output value from the imbalance unit price prediction model 16 for same-day forecast for each forecast target area. Thereafter, the imbalance unit cost prediction unit 15 outputs the predicted value of the imbalance unit cost for that day to the power generation plan creation unit 11 .

[0035] Here, in this embodiment, two imbalance unit price prediction models 16, one for the day-ahead prediction and one for the current day prediction, are used, but these can also be combined into one imbalance unit price prediction model 16. In that case, when making a day-ahead prediction, it becomes possible to make a prediction by setting the contract price in the input data to a fixed value, for example.

[0036] The power generation plan creation unit 11 receives an input of a predicted value of the power generation amount for every 48 frames from the power generation amount prediction unit 14. In addition, the power generation plan creation unit 11 receives an input of a predicted value of the imbalance unit cost for every 48 frames from the imbalance unit cost prediction unit 15.

[0037] Then, the power generation plan preparation unit 11 corrects the predicted value of the power generation amount so as to reduce the economic loss and opportunity loss due to the imbalance as much as possible, and prepares a power generation plan as a power generation plan value. That is, the power generation plan preparation unit 11 calculates a predicted value of the imbalance amount from the predicted value of the power generation amount. Then, the power generation plan preparation unit 11 corrects the predicted value of the power generation amount so as to minimize the difference between (actual power generation amount x power generation unit cost) and (actual power generation amount x power generation unit cost ± imbalance amount x imbalance unit cost). In other words, the power generation plan preparation unit 11 calculates the power generation plan value by correcting the predicted value of the power generation amount so as to minimize the product of the imbalance amount and the imbalance unit cost.

[0038] More specifically, the power generation plan creation unit 11 corrects the predicted value of the power generation amount based on the following principle. For the balancing group operator, if a shortage imbalance occurs during a time period when the imbalance unit price is high, a large loss will occur. On the other hand, for the general electricity transmission and distribution company, since the area supply and demand and the increase adjustment capacity are tight during a time period when the imbalance unit price is high, if a balancing group operator causes a shortage imbalance, the system will be even more tight. Therefore, when the predicted value of the imbalance unit price is high, the power generation plan creation unit 11 creates a power generation plan value by slightly lowering the predicted power generation amount.

[0039] Furthermore, creating a low power generation plan value tends to result in a surplus imbalance, which may result in an opportunity loss compared to the amount of electricity that could be sold if the imbalance was zero. Therefore, the power generation plan creation unit 11 compares the electricity sales income and the imbalance risk, and determines the power generation plan value based on whichever is more economical. Furthermore, if the imbalance unit price is not extremely high, the power generation plan creation unit 11 sets the predicted value of the power generation amount as the power generation plan value so that an imbalance does not occur. After that, the power generation plan creation unit 11 outputs the power generation plan created using the calculated power generation plan value to the output unit 12.

[0040] FIG. 2 is a diagram showing an example of correction to the estimated value of the power generation amount. For example, in case #1 in FIG. 2, the imbalance unit price is higher than in other cases. Therefore, the power generation plan creation unit 11 creates the power generation plan value by correcting the predicted value of the power generation amount to a lower value. In case #2, the imbalance unit price is lower than in other cases and the imbalance amount is also smaller than in other cases. Therefore, the power generation plan creation unit 11 sets the predicted value of the power generation amount as the power generation plan value. In case #3, the imbalance unit price is higher than in other cases and the imbalance unit price is lower than in other cases. Therefore, the power generation plan creation unit 11 creates the power generation plan value by correcting the predicted value of the power generation amount to a higher value with emphasis on the sales income of electricity. In case #4, the imbalance is insufficient and the imbalance unit price is about the middle of the cases. Therefore, the power generation plan creation unit 11 creates the power generation plan value by correcting the predicted value of the power generation amount to a lower value so as to suppress losses. In case #5, the imbalance unit price is higher than in other cases, but the imbalance is in a surplus state. Therefore, the power generation plan creation unit 11 sets the predicted value of the power generation amount as the power generation plan value so as to prevent a shortage imbalance.

[0041] Returning to Fig. 1, the explanation will be continued. The output unit 12 receives an input of a power generation plan from the power generation plan creation unit 11. Then, the output unit 12 outputs the power generation plan to a monitor or the like to notify a user. Alternatively, the output unit 12 may transmit the power generation plan to a terminal of the balancing group operating company to provide the balancing group operating company with the power generation plan.

[0042] 3 is a flowchart of the power generation plan creation process by the power generation plan creation device 1 according to the embodiment. Next, the flow of the power generation plan creation process by the power generation plan creation device 1 according to the embodiment will be described with reference to FIG.

[0043] The power generation amount prediction unit 14 acquires past weather forecast data for each prediction location and actual power generation data corresponding to the past weather forecast from the database 17. Then, the power generation amount prediction unit 14 executes learning of each of the power generation amount prediction models 13 for each prediction location and performs parameter tuning using the acquired past weather forecast data and the actual power generation data corresponding to the past weather forecast as teacher data (step S1).

[0044] The imbalance unit price prediction unit 15 acquires past weather forecast data for each representative point in each area from the database 17. The imbalance unit price prediction unit 15 also acquires information on the imbalance unit price for the past weekly wide-area reserve rate and information on the imbalance unit price for the past 30-minute wide-area reserve rate from the database 17. Furthermore, the imbalance unit price prediction unit 15 acquires the past 30-minute contract price of electricity from the database 17. Then, using the acquired past weather forecast data, the weekly wide-area reserve rate, and the corresponding imbalance unit price as teacher data, the imbalance unit price prediction unit 15 executes learning of each of the imbalance unit price prediction models 16 for day-ahead prediction for each area, and performs parameter tuning (step S2).

[0045] The power generation amount prediction unit 14 obtains nationwide weather forecast data at 3:00 every day from the weather forecast data providing system 2. Then, on the day before the actual supply and demand, the power generation amount prediction unit 14 performs a day-ahead prediction of the power generation amount at each prediction point using input data created from the weather forecast data at 3:00 on the day before the actual supply and demand and the power generation amount prediction model 13 (step S3).

[0046] The imbalance unit cost prediction unit 15 obtains nationwide weather forecast data at 3:00 every day from the weather forecast data providing system 2. In addition, the imbalance unit cost prediction unit 15 obtains weekly wide-area reserve rate data every week from the wide-area reserve rate information providing system 3. Then, on the day before actual supply and demand, the imbalance unit cost prediction unit 15 executes a day-ahead prediction of the imbalance unit cost using input data created from the weather forecast data as of 3:00 on the day before actual supply and demand and the weekly wide-area reserve rate, and the imbalance unit cost prediction model 16 for day-ahead prediction (step S4).

[0047] The power generation plan creation unit 11 receives an input of a predicted value of power generation amount based on the day-ahead prediction from the power generation amount prediction unit 14 on the day before actual supply and demand. The power generation plan creation unit 11 also receives an input of a predicted value of the imbalance unit price based on the day-ahead prediction from the imbalance unit price prediction unit 15. Then, using the predicted value of power generation amount based on the day-ahead prediction and the predicted value of the imbalance unit price, the power generation plan creation unit 11 corrects the predicted value of power generation amount so as to minimize the difference between (actual power generation amount x power generation unit price) and (actual power generation amount x power generation unit price ± imbalance amount x imbalance unit price) as a power generation plan value, and creates a power generation plan for the day before. The output unit 12 outputs the power generation plan for the day before created by the power generation plan creation unit 11 to a monitor or the like (step S5).

[0048] Next, on the day of actual supply and demand, the power generation prediction unit 14 uses input data created from weather forecast data at 3:00 on the day of actual supply and demand and the power generation prediction model 13 to make a prediction of the power generation amount for that day at each prediction point (step S6).

[0049] Furthermore, the imbalance unit price prediction unit 15 obtains nationwide weather forecast data at 3:00 every day from the weather forecast data providing system 2. Furthermore, the imbalance unit price prediction unit 15 obtains wide-area reserve rate data in 30-minute units every day from the wide-area reserve rate information providing system 3. Furthermore, the imbalance unit price prediction unit 15 obtains electricity contract prices in 30-minute units every day from the contract price providing system 4. Then, on the day of actual supply and demand, the imbalance unit price prediction unit 15 executes a day's prediction of the imbalance unit price using input data created from the weather forecast data, wide-area reserve rate data, and contract prices as of 3:00 on the day of actual supply and demand, and the imbalance unit price prediction model 16 for day's prediction (step S7).

[0050] On the actual demand day, the power generation plan creation unit 11 receives an input of a predicted value of power generation amount based on the day's prediction from the power generation amount prediction unit 14. The power generation plan creation unit 11 also receives an input of a predicted value of the imbalance unit price based on the day's prediction from the imbalance unit price prediction unit 15. Then, using the predicted value of power generation amount based on the day's prediction and the predicted value of the imbalance unit price, the power generation plan creation unit 11 corrects the predicted value of power generation amount so as to minimize the difference between (actual power generation amount x power generation unit price) and (actual power generation amount x power generation unit price ± imbalance amount x imbalance unit price), and creates a power generation plan for the day as a power generation plan value. The output unit 12 outputs the power generation plan for the day created by the power generation plan creation unit 11 to a monitor or the like (step S8).

[0051] Fig. 4 is a flowchart of the power generation prediction process by the power generation prediction unit according to the embodiment. Next, the flow of the power generation prediction process by the power generation prediction unit 14 according to the embodiment will be described with reference to Fig. 4. Each process in the flow of Fig. 4 corresponds to an example of the process executed in steps S3 and S6 of the flow of Fig. 3.

[0052] The power generation amount prediction unit 14 acquires nationwide weather forecast data for 3:00 a.m. from the weather forecast data providing system 2 (step S101). In the case of a day-ahead prediction, the power generation amount prediction unit 14 acquires data as of 3:00 a.m. on the day before the actual supply and demand. In the case of a day-ahead prediction, the power generation amount prediction unit 14 acquires data as of 3:00 a.m. on the day of the actual supply and demand.

[0053] Next, the power generation amount prediction unit 14 extracts weather forecast data for each prediction point from the nationwide weather forecast data (step S102).

[0054] Next, the power generation amount prediction unit 14 performs feature engineering processing on the extracted weather forecast data to generate input data (step S103).

[0055] Then, the power generation prediction unit 14 inputs each of the generated input data for each prediction location to the power generation prediction model 13 for each prediction location, and obtains each output value from the power generation prediction model 13 for each prediction location (step S104).

[0056] Next, the power generation amount prediction unit 14 performs fine adjustments on the output values ​​from the power generation amount prediction model 13 for each prediction point (step S105).

[0057] Thereafter, the power generation amount prediction unit 14 aggregates the output values ​​from the power generation amount prediction model 13 for each prediction point that has been finely adjusted, for each balancing group, to generate a predicted value of the power generation amount (step S106).

[0058] Fig. 5 is a flowchart of an imbalance unit price prediction process by the imbalance unit price prediction unit according to the embodiment. Next, a flow of the imbalance unit price prediction process by the imbalance unit price prediction unit 15 according to the embodiment will be described with reference to Fig. 5. Each process in the flow of Fig. 5 corresponds to an example of the process executed in steps S4 and S7 of the flow of Fig. 3.

[0059] The imbalance unit price prediction unit 15 acquires nationwide weather forecast data as of 3:00 a.m. from the weather forecast data providing system 2 (step S201). In the case of a day-ahead prediction, the imbalance unit price prediction unit 15 acquires data as of 3:00 a.m. on the day before the actual supply and demand. In the case of a day-ahead prediction, the imbalance unit price prediction unit 15 acquires data as of 3:00 a.m. on the day of the actual supply and demand.

[0060] Next, the imbalance unit price prediction unit 15 extracts weather forecast data of a representative point in the prediction target area from the nationwide weather forecast data (step S202).

[0061] Next, the imbalance unit price prediction unit 15 performs feature engineering processing on the extracted weather forecast data to extract features (step S203).

[0062] Here, the imbalance unit price prediction unit 15 judges whether or not the prediction is for the current day (step S204).

[0063] In the case of the same day prediction (step S204: Yes), the imbalance cost prediction unit 15 acquires the wide-area reserve margin data in 30-minute units from the wide-area reserve margin information providing system 3 (step S205).

[0064] Moreover, the imbalance unit price prediction unit 15 acquires the contract price of electricity in 30-minute units from the contract price providing system 4 (step S206).

[0065] On the other hand, in the case of the same day prediction (step S204: No), the imbalance unit cost prediction unit 15 acquires data on the weekly wide-area reserve margin from the wide-area reserve margin information providing system 3 (step S207).

[0066] Then, the imbalance unit price prediction unit 15 performs integration processing of the feature amount of the weather forecast data and each of the acquired data to create input data (step S208).

[0067] Next, the imbalance unit price prediction unit 15 inputs the generated input data to the trained imbalance unit price prediction model 16 for each prediction target area, and obtains an output value from the imbalance unit price prediction model 16 for each prediction target area (step S209).

[0068] Thereafter, the imbalance unit price prediction unit 15 creates a predicted value of the imbalance unit price by making fine adjustments to the output value from the imbalance unit price prediction model 16 for each prediction target area (step S210).

[0069] FIG. 6 is a flowchart of the power generation plan creation process by the power generation plan creation unit according to the embodiment. Next, with reference to FIG. 6, a flow of the imbalance unit price prediction process by the power generation plan creation unit 11 according to the embodiment will be described. Each process in the flow of FIG. 6 corresponds to an example of the process executed in steps S5 and S8 of the flow of FIG. 3. Here, the power generation plan creation unit 11 has a first threshold for determining whether the imbalance amount is low or not and a second threshold for determining whether the imbalance amount is high or not. The first threshold can be set to, for example, 5% of the power generation amount. Also, the second threshold can be set to, for example, 20% of the power generation amount. Also, the power generation plan creation unit 11 has a unit price threshold for determining whether the imbalance unit price is high or not. The unit price threshold can be set to, for example, 20 yen.

[0070] The power generation plan creation unit 11 receives an input of a predicted value of the power generation amount from the power generation amount prediction unit 14 (step S301).

[0071] Moreover, the power generation plan creation unit 11 receives an input of a predicted value of the imbalance unit cost from the imbalance unit cost prediction unit 15 (step S302).

[0072] Next, the power generation plan creation unit 11 determines whether the imbalance amount is equal to or less than the first threshold (step S303). If the imbalance amount is equal to or less than the first threshold (step S303: Yes), the power generation plan creation unit 11 creates a power generation plan using the predicted value of the power generation amount as the power generation plan value (step S304).

[0073] On the other hand, if the imbalance amount is greater than the first threshold (step S303: No), the power generation plan creation unit 11 determines whether the imbalance amount is greater than or equal to the second threshold (step S305). If the imbalance amount is greater than or equal to the second threshold (step S305: Yes), the power generation plan creation unit 11 adds an upward correction to the predicted value of the power generation amount and creates a power generation plan as a power generation plan value (step S306).

[0074] On the other hand, if the imbalance amount is less than the second threshold (step S305: No), the power generation plan creation unit 11 determines whether the imbalance unit cost is equal to or greater than the unit cost threshold (step S307).

[0075] If the imbalance unit cost is equal to or greater than the unit cost threshold (step S307: Yes), the power generation plan creation unit 11 determines whether the imbalance is an excess imbalance (step S308).

[0076] If there is a surplus imbalance (step S308: Yes), the power generation plan creating unit 11 creates a power generation plan using the predicted value of the power generation amount as the power generation plan value (step S304).

[0077] If the imbalance is not a surplus imbalance (step S308: No), the power generation plan preparation unit 11 applies a downward correction to the predicted value of the power generation amount to prepare a power generation plan as a power generation plan value (step S309).

[0078] On the other hand, if the imbalance unit cost is less than the unit cost threshold (step S307: No), the power generation plan creation unit 11 determines whether or not the imbalance is a shortage imbalance (step S310).

[0079] If the imbalance is insufficient (step S310: Yes), the power generation plan preparation unit 11 applies a downward correction to the predicted value of the power generation amount to prepare a power generation plan as a power generation plan value (step S309).

[0080] If the imbalance is not a shortage (step S310: No), the power generation plan preparation unit 11 applies an upward correction to the predicted value of the power generation amount to prepare a power generation plan as a power generation plan value (step S306).

[0081] As described above, the power generation plan creation device according to the present embodiment uses a machine learning model to predict the amount of power generation every 48 frames using weather forecast data as an input. By predicting this amount of power generation, the power generation plan creation device can predict the amount of power generation from renewable energy sources such as solar power generation and wind power generation. In addition, the power generation plan creation device uses a machine learning model to predict the imbalance unit price of the day-ahead forecast using weather forecast data and a wide-area reserve rate as an input, and predicts the imbalance unit price of the day forecast using weather forecast data, a wide-area reserve rate, and a contract price as an input. Then, the power generation plan creation device uses the predicted value of the power generation amount and the predicted value of the imbalance unit price to create a power generation plan so as to reduce economic loss and opportunity loss due to imbalance as much as possible.

[0082] This makes it possible to calculate a power generation plan value that takes into account the imbalance unit price, in addition to the prediction function for power generation including renewable energy power generation such as solar power generation and wind power generation. Therefore, it becomes possible to create a power generation plan that is optimal for the balancing group operator and the general power distribution company. It also becomes possible to predict the imbalance unit price corresponding to the calculation method of the imbalance unit price using the limit kWh price of the adjustment power procured by the general power distribution company over a wide area. In addition to formulating the plan value to be submitted to the cross-regional organization as of the day before the actual supply and demand, it is possible to predict the imbalance unit price of the power generation with higher accuracy based on information close to the actual supply and demand on the day, and update and provide the plan value. In this way, it becomes possible to create an optimal power generation plan to minimize the imbalance risk.

[0083] (Hardware configuration) Fig. 7 is a hardware configuration diagram of a power generation plan creation device. The power generation plan creation device 1 has, for example, a hardware configuration as shown in Fig. 7. That is, the power generation plan creation device 1 has a CPU (Central Processing Unit) 91, a memory 92, a hard disk 93, and a network interface 94. The CPU 91 is connected to the memory 92, the hard disk 93, and the network interface 94 via a bus.

[0084] The network interface 94 is a communication interface between the power generation plan creation device 1 and an external device. The network interface 94 relays communication between the CPU 91 and the weather forecast data providing system 2, the wide-area reserve margin information providing system 3, and the contract price providing system 4, for example.

[0085] The hard disk 93 is an auxiliary storage device. The hard disk 93 can realize the function of the database 17. The hard disk 93 may store the power generation amount prediction model 13 and the imbalance unit price prediction model 16. The hard disk 93 also stores various programs including programs for realizing the functions of each of the power generation plan creation unit 11, the output unit 12, the power generation amount prediction unit 14, and the imbalance unit price prediction unit 15 illustrated in FIG. 1.

[0086] The memory 92 is a main storage device and may be a dynamic random access memory (DRAM).

[0087] The CPU 91 reads out various programs from the hard disk 93, and deploys and executes them in the memory 92. In this way, the CPU 91 realizes the functions of the power generation plan creation unit 11, the output unit 12, the power generation amount prediction unit 14, and the imbalance unit price prediction unit 15 illustrated in FIG.

[0088] The power generation plan creation device 1 can also realize the same functions as those of the above-mentioned embodiment by reading the above-mentioned program from a recording medium using a medium reading device and executing the read program. Note that the program referred to here is not limited to being executed by a specific power generation plan creation device 1. For example, the present invention can be similarly applied to a case where another computer or server executes a program, or where these cooperate to execute a program.

[0089] This program can be distributed via a network such as the Internet. This program can also be recorded on a computer-readable recording medium such as a hard disk, a flexible disk (FD), a CD-ROM, a magneto-optical disk (MO), or a digital versatile disk (DVD), and can be executed by being read out from the recording medium by a computer. [Explanation of symbols]

[0090] 1. Power generation planning device 2. Weather forecast data provision system 3. Wide-area Reserve Margin Information System 4. Contract Price Providing System 11 Power Generation Planning Department 12 Output section 13 Power generation forecast model 14 Power generation forecast section 15 Imbalance Unit Price Prediction Department 16 Imbalance unit price prediction model 17 Database

Claims

1. a power generation prediction step of acquiring weather forecast data for a specific day, and calculating a predicted value of the power generation amount on the specific day using a power generation prediction model that inputs the weather forecast data and outputs a power generation amount; an imbalance unit cost forecasting step of acquiring weather forecast data and a wide-area reserve rate for the specific day, and calculating a predicted value of the imbalance unit cost using an imbalance unit cost forecasting model that inputs the weather forecast data and the wide-area reserve rate and outputs an imbalance unit cost; a power generation plan creating step of correcting the predicted value of the power generation amount calculated in the power generation amount predicting step based on the predicted value of the imbalance unit price calculated in the imbalance unit price predicting step to calculate a power generation plan value and create a power generation plan; A power generation plan creation program that causes a computer to execute the above steps.

2. The imbalance unit price prediction step includes: a day-ahead prediction step of predicting the imbalance unit price for the specific day based on weather forecast data for the specific day as of the day before the specific day and the wide-area reserve margin on a daily basis; a current day prediction step of further acquiring information on the contract price of electricity for the specific day, and predicting the imbalance unit price for the specific day based on weather forecast data for the specific day as of the specific day, the wide-area reserve margin for a predetermined minute, and the contract price; 2. The power generation plan creating program according to claim 1, further comprising:

3. 2. The power generation plan creation program according to claim 1, wherein the power generation plan creation step predicts an imbalance amount for the specific day from a predicted value of the power generation amount for the specific day, and calculates the power generation plan value by correcting the predicted value of the power generation amount so that a product of the predicted imbalance amount and the imbalance unit price is as small as possible.

4. a power generation prediction step of acquiring weather forecast data for a specific day, and calculating a predicted value of the power generation amount on the specific day using a power generation prediction model that inputs the weather forecast data and outputs a power generation amount; an imbalance unit cost forecasting step of acquiring weather forecast data and a wide-area reserve rate for the specific day, and calculating a predicted value of the imbalance unit cost using an imbalance unit cost forecasting model that inputs the weather forecast data and the wide-area reserve rate and outputs an imbalance unit cost; a power generation plan creating step of correcting the predicted value of the power generation amount calculated in the power generation amount predicting step based on the predicted value of the imbalance unit price calculated in the imbalance unit price predicting step to calculate a power generation plan value and create a power generation plan; A power generation plan creating method comprising the steps of:

5. A power generation prediction model that inputs weather forecast data and outputs power generation amount; an imbalance unit cost forecast model that receives weather forecast data and a wide-area reserve margin as inputs and outputs an imbalance unit cost; an energy generation prediction unit that obtains weather forecast data for a specific day and calculates a predicted value of the energy generation amount for the specific day using the energy generation prediction model; an imbalance unit cost prediction unit that obtains weather forecast data and a wide-area reserve margin for the specific day and calculates a predicted value of the imbalance unit cost for the specific day; a power generation plan creation unit that creates a power generation plan by correcting the predicted value of the power generation amount based on the imbalance unit price to calculate a power generation plan value; A power generation plan creation device comprising:

Citation Information

Patent Citations

  • Supply-demand management system, supply-demand management method and program

    JP2022184637A