Information processing device, information processing method, and program

Abstract

To provide an information processing device, an information processing method, and a program for improving prediction accuracy of electricity pricing while ensuring consistency across a plurality of markets.SOLUTION: An information processing device 1 includes: an acquisition part 10 which acquires operating power supply information by applying power demand prediction to a power supply operation model; and a prediction part 20 which predicts electricity pricing in a first system by applying the operational power supply information to a first individual model, and predicts electricity pricing in a second system different from the first system by applying the operational power supply information to a second individual model.SELECTED DRAWING: Figure 8

Description

[Technical Field]

[0001] The present invention relates to an information processing device, an information processing method, and a program for predicting electricity prices. [Background technology]

[0002] Attempts have been made to predict electricity prices in various markets. For example, Patent Document 1 proposes an information processing device that includes: a model generation means that generates an optimization model having constraints on the amount of electricity that can be procured from each of a plurality of electricity procurement means, where the amount of electricity procured from each of the plurality of electricity procurement means is an manipulated variable and a value including the total cost required to procure an amount of electricity corresponding to predicted demand using the plurality of electricity procurement means is an objective function; and a determination means that determines a power procurement plan indicating the amount of electricity to be procured from each of the plurality of electricity procurement means by solving the optimization model using a mathematical optimization technique. The information processing device is disclosed to include an electricity price prediction means that predicts electricity prices based on a history of past electricity prices. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent Publication No. 2021-157724 Summary of the Invention [Problem to be solved by the invention]

[0004] In the conventional method as shown in Patent Document 1, it is possible to predict electricity prices in a single market such as a spot market, but it is not possible to achieve sufficient consistency between markets, and as a result, the prediction accuracy is not sufficient.

[0005] The present invention has been made in consideration of the above points, and provides an information processing device, an information processing method, and a program that can improve the accuracy of electricity price predictions while ensuring consistency between multiple market systems, etc. [Means for solving the problem]

[0006] [Concept 1] An information processing device according to a first aspect of the present invention includes: an acquisition unit that acquires information about operating power sources by applying the power demand forecast to a power source operation model; a prediction unit that predicts an electricity price in a first system by applying information about the operating power source to a first individual model, and predicts an electricity price in a second system different from the first system by applying information about the operating power source to a second individual model; may also be provided.

[0007] [Concept 2] In the information processing device according to concept 1, The electricity price in the first system may be any one of the agreed price in the spot market, the agreed price in the ΔkW supply and demand adjustment market, the registered price in the kWh supply and demand adjustment market, the future agreed price by area in the time-ahead market, and the imbalance price, and the electricity price in the second system may be any one of the agreed price in the spot market, the agreed price in the ΔkW supply and demand adjustment market, the registered price in the kWh supply and demand adjustment market, the future agreed price by area in the time-ahead market, and the imbalance price, and may be an electricity price in a second system different from the first system.

[0008] [Concept 3] In the information processing device according to concept 2, The electricity price in the first system is a contract price in the spot market or a contract price in the ΔkW supply and demand adjustment market, The electricity price in the second system is a registered price in the kWh supply and demand adjustment market, a future contract price by area in the time-ahead market, or an imbalance price, The prediction unit may predict the power generation plan and available adjustment capacity of each power source based on the price agreement history in the first system, or predict the power sources that are operating at the time of actual supply and demand in the spot market and predict the power sources that are waiting to operate at the time of actual supply and demand in the ΔkW supply and demand adjustment market, and use the results of the predictions to predict the electricity price in the second system.

[0009] [Concept 4] In an information processing device according to any one of concepts 1 to 3, The prediction unit may determine whether a continuously operating power source that operates continuously for a specified period of time is operating based on price agreement history, and if it determines that the continuously operating power source is operating, may predict the electricity price in the first system or the second system on the assumption that the continuously operating power source will be operated for the specified period of time.

[0010] [Concept 5] An information processing device according to a second aspect of the present invention includes: an acquisition unit that acquires information about operating power sources by applying the demand forecast to a power source operation model; a prediction unit that predicts an agreement price in the spot market by applying information about the operating power source to a spot market model, predicts an agreement price in the ΔkW supply and demand adjustment market by applying information about the operating power source to a ΔkW supply and demand adjustment market model, predicts a registered price in the kWh supply and demand adjustment market by applying information about the operating power source to a kWh supply and demand adjustment market model, predicts a future agreement price by area in the time-ahead market by applying information about the operating power source to an hour-ahead market model, and predicts an imbalance price by applying information about the operating power source to an imbalance price model; may also be provided.

[0011] [Concept 6] In the information processing device according to concept 5, The prediction unit may predict the power generation plan and available adjustment capacity of each power source from the price agreement records in the spot market and the supply and demand adjustment market, or may predict the power sources that will be operating at the time of actual supply and demand from the price agreement records in the spot market and predict the power sources that will be waiting to operate at the time of actual supply and demand from the price agreement records in the ΔkW supply and demand adjustment market, and may use the results of the predictions to predict the registered price in the kWh supply and demand adjustment market, and the future agreement price and imbalance price by area in the time-ahead market.

[0012] [Concept 7] In the information processing device according to Concept 5 or 6, The prediction unit may determine whether a continuously operating power source that operates continuously for a predetermined period of time is operating based on price agreement history, and if it determines that the continuously operating power source is operating, may predict the agreement price in the spot market, the agreement price in the ΔkW supply and demand adjustment market, the registered price in the kWh supply and demand adjustment market, and the future agreement price and imbalance price by area in the time-ahead market, assuming that the continuously operating power source will be operated for the predetermined period of time.

[0013] [Concept 8] The information processing method according to the present invention comprises: acquiring information about the operating power sources by applying the demand forecast to a power source operation model by an acquisition unit; a step of predicting, by a prediction unit, a contract price in the spot market by applying information about the operating power source to a spot market model, predicting a contract price in the ΔkW supply and demand adjustment market by applying information about the operating power source to a ΔkW supply and demand adjustment market model, predicting a registered price in the kWh supply and demand adjustment market by applying information about the operating power source to a kWh supply and demand adjustment market model, predicting a future contract price by area in the time-ahead market by applying information about the operating power source to an hour-ahead market model, and predicting an imbalance price by applying information about the operating power source to an imbalance price model; may also be provided.

[0014] [Concept 9] The program according to the present invention comprises: A program to be installed on an information processing device, On the information processing device on which the program is installed, The function of applying demand forecasts to a power generation operation model to obtain information about the power generation in operation; a function of predicting a contract price in the spot market by applying information about the operating power source to a spot market model, predicting a contract price in the ΔkW supply and demand adjustment market by applying information about the operating power source to a ΔkW supply and demand adjustment market model, predicting a registered price in the kWh supply and demand adjustment market by applying information about the operating power source to a kWh supply and demand adjustment market model, predicting future contract prices by area in the hour-ahead market by applying information about the operating power source to an hour-ahead market model, and predicting an imbalance price by applying information about the operating power source to an imbalance price model; may be realized. [Effects of the Invention]

[0015] In the present invention, if an aspect is adopted in which information about operating power sources is obtained by applying an electricity demand forecast to a power source operation model, the information about the operating power sources is applied to a first individual model to predict an electricity price in a first system, and the information about the operating power sources is applied to a second individual model to predict an electricity price in a second system different from the first system, it is possible to improve the accuracy of electricity price prediction while achieving consistency between multiple systems. Also, in the present invention, if the prediction unit applies information about the operating power sources to a spot market model to predict an agreement price in the spot market, applies information about the operating power sources to a ΔkW balancing market model to predict an agreement price in the ΔkW balancing market, applies information about the operating power sources to a kWh balancing market model to predict a registered price in the kWh balancing market, applies information about the operating power sources to an hour-ahead market model to predict a future agreement price by area in the hour-ahead market, and applies information about the operating power sources to an imbalance price model to predict an imbalance price, it is possible to improve the accuracy of prediction of electricity prices and imbalance prices in the spot market, the ΔkW balancing market, the kWh balancing market, and the hour-ahead market. [Brief explanation of the drawings]

[0016] [Figure 1] FIG. 4 is a diagram showing an example of information flow in a power supply operation model and an individual model according to the embodiment. [Figure 2] 10A and 10B are diagrams illustrating one mode of determining the startup state of a power supply using a power supply operation model according to an embodiment. [Figure 3] FIG. 2 is a diagram showing an example of information flow in forecasting spot market prices according to an embodiment. [Figure 4] FIG. 10 is a diagram showing an example of a graph used in predicting an imbalance price in the embodiment. [Figure 5] FIG. 10 is a diagram showing an example of the flow of information in predicting the bid price-contract probability in the ΔkW supply and demand balancing market according to the embodiment. [Figure 6]FIG. 2 is a diagram for explaining one mode of predicting a registration price-response probability in a kWh supply and demand adjustment market according to an embodiment. [Figure 7] 10 is a diagram for explaining one mode of predicting future contract prices by area in the time-ahead market according to an embodiment. FIG. [Figure 8] 1 is a schematic diagram showing a configuration of an information processing system according to an embodiment of the present invention; DETAILED DESCRIPTION OF THE INVENTION

[0017] Embodiment Hereinafter, an information processing device and an information processing method according to the present invention will be described. In the present embodiment, a program that can be installed in a computer such as a personal computer to enable the computer to execute the information processing method according to the present embodiment, and a recording medium on which the program is recorded are also provided.

[0018] The information processing device of this embodiment may be installed anywhere, may be a server, or may be used in a cloud environment. The information processing device of this embodiment may be composed of one device or multiple devices. Furthermore, when an information processing device is composed of multiple devices, the devices do not need to be installed in the same space such as the same room, but may be installed in different rooms, different buildings, different regions, etc. Furthermore, when an information processing device is composed of multiple devices, some of the devices may be owned and / or managed by one institution, and the rest may be owned and / or managed by another institution.

[0019] The information processing device of this embodiment is generated, for example, by installing a program. This program may be distributed by email, may be available by accessing a predetermined URL and logging in, or may be recorded on a recording medium. The program of this embodiment is used to generate the information processing device described below, and the recording medium of this embodiment is used to record the program. Furthermore, the information processing method of this embodiment is implemented by an information processing device in which the program is installed. The information processing device may execute the information processing method of this embodiment by executing an application installed on the information processing device.

[0020] As shown in Figure 8, the information processing device 1 of this embodiment has an acquisition unit 10 that acquires information about operating power sources at the time of supply and demand, including operating power sources (this ``operating power sources'' includes ``power sources on standby for operation''), non-operating power sources, and the equipment utilization rates of operating power sources, by applying power demand forecasts to a power source operation model; a prediction unit 20 that predicts electricity prices in a first system by applying the information about the operating power sources to a certain individual model (first individual model), and predicts electricity prices in a second system different from the first system by applying the information about the operating power sources to another individual model (second individual model); and an output unit 90 that outputs the prediction results by the prediction unit 20.

[0021] The information processing device 1 may be capable of communicating with multiple user terminals 100 and an administrator terminal 150. The user terminal 100 is a terminal such as a personal computer, a smartphone, a tablet terminal, or a wearable computer such as a smartwatch or smartglasses, and is typically a mobile terminal such as a smartphone or tablet terminal. The user terminal 100 may have an operation unit 120 through which a user inputs information and a display unit 110 for displaying information. In a smartphone or tablet terminal, the operation display unit has the functions of the operation unit 120 and the display unit 110. The administrator terminal 150 is also, for example, a terminal such as a personal computer, a smartphone, a tablet terminal, or a wearable computer such as a smartwatch or smartglasses. The administrator terminal 150 may have an operation unit 170 through which an administrator inputs information and a display unit 160 for displaying information. In a smartphone or tablet terminal, the operation display unit has the functions of the operation unit 170 and the display unit 160. Input from users and administrators may be performed by text input or voice input. The content output by the output unit 90 can be viewed on the display unit 110 of the user terminal 100 or the display unit 160 of the administrator terminal 150. Furthermore, information input from the operation unit 120 of the user terminal 100 or the operation unit 170 of the administrator terminal 150 may be stored in the storage unit 80 in association with user identification information or administrator identification information.

[0022] The term "system" refers to electricity markets such as spot markets, balancing markets, and time-ahead markets, as well as pricing systems for imbalance pricing. According to this embodiment, electricity prices in multiple systems can be estimated using information about operating power sources (including power sources on standby for operation), power sources that are not operating, and information about operating power sources, including their utilization rates. It is useful to obtain information about operating power sources using a single power source operation model and use this information about operating power sources to predict electricity prices in multiple systems. Regarding prices for each system, multiple individual models may be stored in the storage unit 80, and electricity prices in each system may be predicted by applying information about operating power sources to the individual models. Each individual model may be a model generated or corrected using machine learning using past performance data. The start / stop characteristics and fuel cost characteristics of power sources are particularly important in predicting electricity prices in each system. These parameters may be updated appropriately using information about the actual electricity prices. In addition, when correcting each individual model using machine learning, one example is to generate a model for each system in advance for each individual model based on electrical engineering, etc., and then correct each model using machine learning using past performance data to improve accuracy.

[0023] Existing algorithms such as dynamic programming and branch and bound can be used to determine the start / stop status of each power source using a power source operation model, and the equal incremental fuel cost method can be used to calculate marginal costs based on fuel cost characteristics (see Figure 2). The equal incremental fuel cost method is a method in which the fuel cost of each generator is expressed as a specified function of generator output, and if the generators are operated so that the incremental fuel cost of each generator is equal in each time period, the generator output with the lowest fuel cost can be obtained for the entire group of generators.

[0024] The following can be used as input data when constructing the power supply operation model and the individual models (see FIG. 1). The individual models may be generated or corrected by machine learning using past performance data related to the information shown below, and stored in the storage unit 80. Spot bidding and execution results: supply and demand curve, block bidding and execution volume, market fragmentation occurrence status, etc. Power supply: capacity, thermal efficiency, startup costs, shutdown information (HJKS), etc. Supply and demand: demand by time of day, renewable energy and nuclear power generation, pumped storage operation, etc. Interconnection lines: power flow performance, operational capacity, etc. ·Fuel costs: coal, LNG, oil, etc. - Former electric power companies (former general electric utilities): company demand, company power sources, demand and power generation forecasts, etc. Imbalance: Price, generation volume, merit order of adjustment capacity kWh, etc.

[0025] In markets and systems related to kWh prices (other than imbalance prices), power producers are expected to bid or register in the system the excess capacity from their power generation plans at each point in time at marginal cost. Therefore, from the power source startup status calculated based on demand forecasts such as demand estimates (see Figure 2), it is possible to calculate marginal cost curves using the equal incremental fuel cost method (such as the adjustment capacity kWh merit order shown in Figures 4 and 6) and to estimate bids that require startup costs to be factored in. Publicly known technologies can be used to forecast electricity demand.

[0026] The power supply operation model may be a pre-prepared model, and may be configured to operate power supplies in order of lowest price to fill the power demand forecast (see Figure 2). In this configuration, among multiple available power supplies, the lowest-priced power supplies are activated first, allowing for identification of power supplies operating at a certain percentage of capacity utilization and unused power supplies, and the amount of power generated by unused power supplies is calculated as unused power. A model may also be created to operate power supplies of similar price levels evenly. Similar prices may be appropriately defined by the administrator via the administrator terminal 150, for example, by a price difference of within 5% of the price. The power supply operation model may be constructed based on generator information registered in the power generation information disclosure system, published materials from power generation companies, power generation performance by power source, etc. A bidding curve may be generated by inputting input data into the power supply operation model.

[0027] Balancing markets may include ΔkW balancing markets and kWh balancing markets. Products such as JPEG2025152395000002.jpg11160 exist.

[0028] The electricity price in the first system is one of the contract price in the spot market, the contract price-contract probability in the ΔkW supply and demand adjustment market (meaning the contract price and the contract probability for each contract price), the registered price-response probability in the kWh supply and demand adjustment market (meaning the registered price and the response probability for each registered price), the future contract price by area in the time-ahead market, and the imbalance price, and the electricity price in the second system is one of the contract price in the spot market, the contract price-contract probability in the ΔkW supply and demand adjustment market, the registered price-response probability in the kWh supply and demand adjustment market, the future contract price by area in the time-ahead market, and may be an electricity price in a system different from the first system.

[0029] The electricity price in the first system may be the contract price in the spot market or the contract price minus the contract probability in the ΔkW balancing market, and the electricity price in the second system may be the registered price minus the response probability in the kWh balancing market, or the future contract price by area or the imbalance price in the time-ahead market. Since the contract price in the spot market and the contract price in the ΔkW balancing market are contracted and fixed by the day before the actual supply and demand date, the prediction unit 20 may predict the power generation plan and available adjustment capacity of each power source based on the contract history of the contract price in the spot market or the contract price in the ΔkW balancing market included in the first system, or may predict the power sources operating at the time of actual supply and demand in the spot market and the power sources waiting to operate at the time of actual supply and demand in the ΔkW balancing market, and use the results of the predictions to predict the electricity price in the second system, such as the registered price minus the response probability in the kWh balancing market, or the future contract price by area or the imbalance price in the time-ahead market. When such an embodiment is adopted, the electricity price in the second system is predicted using actually confirmed information (contract price, etc.), so that the electricity price can be predicted with higher accuracy. In this case, the prediction unit 20 may predict the electricity price in the second system using not only the contract results in the spot market and the supply and demand adjustment market (ΔkW contract price - contract probability) but also related published information.

[0030] As in the present embodiment, by using the results of a commonly used power generation operation model and then predicting the power prices of each system using individual models, discrepancies between the results of each power price forecast are reduced, enabling consistent explanations. More specifically, the mechanisms that form the electricity market include spot markets and time-ahead markets, where electricity is bought and sold to achieve simultaneous balancing of planned values, supply-demand balancing markets, where adjustment capacity is procured to compensate for deviations from planned values, and imbalance prices, which settle discrepancies between final planned and actual values. Meanwhile, power generation operation plans, which are important in forming each power price, are updated sequentially at the bidding and contract timing of each power market, taking into account constraints imposed by the power generation plan at that time, so all of the above power prices are correlated with each other. In this regard, using a commonly used power generation operation model reduces discrepancies between the results of each power price forecast and enables consistent explanations.

[0031] Please also refer to pages 7 and 11 of the document "Issues in the Electricity Market System from the Perspective of Thermal Power Plant Operators" (https: / / www.meti.go.jp / shingikai / energy_environment / oroshi_jukyu / pdf / 001_09_00.pdf) dated December 28, 2021, which provides examples of each market and the timing of its execution, as well as the purpose and execution method of each market. Also, please refer to the document "Regarding the Imbalance Pricing System, etc." (https: / / www.emsc.meti.go.jp / info / public / pdf / 20220117001b.pdf) dated January 28, 2022, which explains imbalance pricing.

[0032] In the spot market and the ΔkW supply and demand balancing market, it is necessary to simulate the operation of power sources prior to the day before the actual supply and demand, when the start and stop of power sources is decided, and the bid amount in the spot market also increases or decreases in order to secure the ΔkW to be provided to the supply and demand balancing market. Therefore, when simulating the operation of a power generation company aiming for the most economical operation under market rules, it is possible to calculate the amount to be provided to each market taking such correlation into account. Furthermore, in the imbalance price, supply and demand adjustment market (kWh price - command probability), and hour-ahead market, it is important to estimate the required adjustment capacity and the adjustment capacity kWh price merit order. By generating linked forecast values ​​using a common power source operation model, for example, if the imbalance forecast price is high in a certain forecast target frame, the price that can be traded in the hour-ahead market will also be high, and the probability of a command being issued in the supply and demand adjustment market will also be high. This makes it possible to create operational plans that refer to multiple electricity prices based on rational forecast results.

[0033] The prediction unit 20 may use an operation prediction model to predict the power generation plan and available adjustment capacity of each power source based on the price contract history, as well as the power sources operating at the time of actual supply and demand and / or standby for operation at the time of actual supply and demand. Because the price of each power source is generally fixed, it is beneficial to use the contracted price. However, predictions may also be made using information such as the contracted volume and the use history of interconnection lines in addition to the price contract history. As an example, the operation prediction model is a model that associates the price contract history, the contracted volume, the use history of interconnection lines, etc. with power sources estimated to be operating, and is stored in the storage unit 80. The operation prediction model may be a model generated using the contract price and the power sources operating or standby for operation at that contract price. For example, it may be a model generated using the contract price in a target area and the power sources operating or standby for operation in that area at that contract price. Furthermore, the operation prediction model may be a model generated or corrected by machine learning using the contract price in a target area and the past performance of the power sources operating or standby for operation at that contract price. The operating power sources or power sources on standby for operation associated with the contract price may be publicly available information, or may be information appropriately input by the administrator via the administrator terminal 150. The size of the area may be determined by prefecture, or may be divided into predetermined regions such as Hokkaido, Tohoku, Kanto, Chubu, Kinki, Chugoku, Shikoku, and Kyushu (including Okinawa), or may be determined by the area under the jurisdiction of a general electricity transmission and distribution utility, such as the divisions shown in FIG. 5 (Hokkaido, Tohoku, Tokyo, Chubu, Hokuriku, Kansai, Chugoku, Shikoku, and Kyushu). When correcting the operation prediction model using machine learning, for example, an operation prediction model may be generated in a rational manner based on electrical engineering or the like using the contract price and the power sources operating at that contract price, and then corrections may be made using machine learning using past performance data to improve accuracy.

[0034] Note that, once a power source using a large turbine is put into operation, it is generally operated for a certain period of time, such as several hours or a day, and is not suddenly stopped. Therefore, when the prediction unit 20 determines (predicts) that a predetermined power source (continuously operating power source) such as a power source using a large turbine is operating at a certain point in time (the actual supply and demand point) based on the price agreement records, the agreed amount, the usage record of the interconnection line, etc. in the spot market, the ΔkW supply and demand adjustment market, the kWh supply and demand adjustment market, and the time-ahead market, the contract price in the spot market, the contract price-contract probability in the ΔkW supply and demand adjustment market, the registered price-response probability in the kWh supply and demand adjustment market, and the future contract price and imbalance price by area in the time-ahead market at another future point in time (the actual supply and demand point) may be predicted, assuming that the continuously operating power source will continue to operate (using the condition that the continuously operating power source will continue to operate for a predetermined period of time, such as 4 hours, 12 hours, 1 day, or 2 days from the start of the actual supply and demand in question). By adopting such an embodiment, it is possible to further improve the accuracy of electricity price prediction.

[0035] In a typical example of this embodiment, the prediction unit 20 (1) predicts the contract price in the spot market by applying information about the operating power sources (see FIG. 2) to a spot market model, (2) predicts the contract price-contract probability in the ΔkW balancing market by applying information about the operating power sources to a ΔkW balancing market model, (3) predicts the registration price-response probability in the kWh balancing market by applying information about the operating power sources to a kWh balancing market model, (4) predicts future contract prices by area in the hour-ahead market by applying information about the operating power sources to an hour-ahead market model, and (5) predicts the imbalance price by applying information about the operating power sources to an imbalance price model, thereby predicting electricity prices in five different systems (see FIG. 1). When such an embodiment is adopted, it is possible to obtain information about the operating power sources using a single power source operation model and consistently predict electricity prices in five different major systems using this information about the operating power sources, thereby achieving an extremely beneficial effect.

[0036] However, the present invention is not limited to this embodiment, and electricity prices may be predicted in two, three, or four of these five systems. Each of the individual models, namely, the spot market model, the ΔkW balancing market model, the kWh balancing market model, the time-ahead market model, and the imbalance price model, is stored in the storage unit 80.

[0037] Electricity prices are determined by different methods depending on the system, such as in the balancing market, where electricity prices are determined by a multi-price auction, in the spot market, where electricity prices are determined by a single-price auction, and in the hourly market, where electricity prices are determined by intraday trading.In order to improve the interpretability of predictions and to logically calculate phenomena such as price hikes, individual models may use prediction models based on the contract rules of each market.

[0038] The spot market model may be generated using information about historical operating power sources and data about electricity prices including block bids and regular bids in the spot market, and stored in the storage unit 80. Similarly, the ΔkW balancing market model may be generated using information about historical operating power sources and data about electricity prices in the historical ΔkW balancing market, and stored in the storage unit 80. The kWh balancing market model may be generated using information about historical operating power sources and data about electricity prices in the historical kWh balancing market, and stored in the storage unit 80. The hour-ahead market model may be generated using information about historical operating power sources and data about electricity prices in the historical hour-ahead market, and stored in the storage unit 80. The imbalance price model may be generated using information about historical operating power sources and data about imbalance prices, and stored in the storage unit 80.

[0039] Machine learning may also be used, and the spot market model may be generated by machine learning using learning data related to information about past operating power sources and past electricity prices in the spot market, including block bids and regular bids, and stored in the storage unit 80. Similarly, the ΔkW balancing market model may be generated by machine learning using learning data related to information about past operating power sources and past electricity prices in the ΔkW balancing market, and stored in the storage unit 80. The kWh balancing market model may be generated by machine learning using learning data related to information about past operating power sources and past electricity prices in the kWh balancing market, and stored in the storage unit 80. The hour-ahead market model may be generated by machine learning using learning data related to information about past operating power sources and past electricity prices in the hour-ahead market, and stored in the storage unit 80. The imbalance price model may be generated by machine learning using learning data related to information about past operating power sources and past imbalance prices, and stored in the storage unit 80.

[0040] The spot market model may be designed to simulate market rules that cause price spikes, such as block contracts and market fragmentation (see Figure 1). Refined power supply information based on actual results can also be used for other forecasts. The imbalance price model may incorporate demand and renewable energy forecasts provided by the former general electric utilities (former electric utilities) on an ongoing basis. The price may also be calculated based on the merit order of adjustment capacity kWh based on market rules. The ΔkW supply and demand balancing market model may be a model that predicts the probability of a contract for a bid price. The contract process may be simulated by calculating the wide-area procurement of balancing capacity that takes into account the upper limit of interconnection lines. The kWh supply and demand adjustment market model may be a model that predicts the command probability for the registered kWh unit price. The adjustment capacity kWh merit order assumed in the imbalance price forecast may be used. In the hour-ahead market model, it is also possible to separate the contract price trends (nationwide) by area using quote information, etc. If such a configuration is adopted, it becomes possible to make bids that take into account price increases and decreases several hours later by predicting price trends by area.

[0041] [Spot market price forecast] When forecasting block bids in the spot market, dynamic programming, branch and bound methods, etc. may be used to simulate the power plant startup and shutdown plans of major power generation companies that are required to provide all of their major power plants, especially surplus power, using a power plant operation model, and block bid quantities and prices may be estimated (see Figure 3, "Power Plant Operation Model"). Block bids for power plants that are currently shut down may include startup costs, while block bids due to step constraints (constraints related to the rate of change in generator output) may be estimated without startup costs. Because block bids are executed based on the weighted average price during the bidding time period, even block bids equivalent to the daily average price in the spot market may not necessarily be executed, resulting in a shortage of selling contracts and higher prices. However, this method can simulate such phenomena.

[0042] When forecasting regular bids in the spot market, a marginal cost curve for power sources with excess supply capacity can be created using the equal incremental fuel cost method or other methods based on the results of calculations of start-up and shutdown plans, and regular bid volumes and prices can be estimated (see Figure 3, "Power Source Operation Model").

[0043] In the spot market model, in order to predict other bidding information, it may be configured to predict the interconnection line capacity available for buying bids and the spot market, the amount of 0 yen selling bids (renewable energy), etc. (See "AI model" in Figure 3.) The AI ​​model in Figure 3 is a model included in the spot market model, and is a model generated by machine learning using past performance as learning data.

[0044] Using assumptions related to bidding information, block transactions, interconnection line interchange, and market segmentation processing can be simulated in accordance with market rules, and supply and demand curves can be created for each segmented area, with the final area-specific price calculated from their intersections (see "Agreement Processing Model" in Figure 3). Adopting this approach makes it possible to simulate and forecast price hikes due to sellouts caused by block transactions, market segmentation, etc. The agreement processing model may be a model generated or corrected by machine learning using past block bid volumes and prices, regular bid volumes and prices, buy curves, 0 yen sell bids, and the operating capacity of interconnection lines, as well as past agreement prices in the spot market, as learning data.

[0045] Area-specific prices can also be predicted probabilistically by assuming the input bidding information. When this type of method is adopted, the forecast assumes supply and demand curves, so it is possible to simulate the respective price elasticities, and when the input bidding information is assumed probabilistically, it is possible to generate an asymmetric probability distribution that reflects the distribution of actual contracts, such that the probability of occurrence gradually decreases as the contract price increases.

[0046] If you want to further improve the accuracy of your forecasts, you can use various explanatory variables to generate price forecasts using AI (spot market models) to perform trend corrections (see Figure 3).

[0047] In the embodiment shown in Figure 3, input data is applied to a power generation operation model and a spot market model to predict block bid volumes and prices, as well as regular bid volumes and prices. Furthermore, input data is applied to an AI model included in the spot market model to predict buy curves, 0 yen sell bids, the operating capacity of interconnection lines, and price trends. Then, the predicted block bid volumes and prices, regular bid volumes and prices, buy curves, 0 yen sell bids, and the operating capacity of interconnection lines are applied to an agreement processing model to predict agreement prices in the spot market.

[0048] In the embodiment shown in Figure 3, the supply curve generated from the power generation operation model is refined using relevant public information from the spot market, which can also contribute to improving the accuracy of kWh price assumptions in other electricity price forecasts.

[0049] [Imbalance Price Prediction] The imbalance pricing model may use AI or other methods to predict the amount of imbalance in each area based on electricity demand forecasts. A kWh price merit order for balancing capacity (see the dashed arrow in Figure 4) may be created based on the kWh price marginal cost curve derived from the power source operation model and public information on balancing capacity (see the dotted arrow in Figure 4). The marginal balancing capacity kWh price for the predicted amount of imbalance may be used as the predicted value. The predicted amount of imbalance in each area may be updated as appropriate based on updates to weather forecasts and area supply and demand forecasts. When creating a balancing capacity kWh price merit order based on the power source operation model, the equal incremental fuel cost method may be used, as in the spot market, or it may be adjusted as appropriate using information on supply and demand released by general electricity transmission and distribution companies. This also applies to forecasting contract prices in the ΔkW balancing market, future area-specific contract prices in the hourly market, and imbalance prices. Under normal circumstances, the forecast result may be an intersection shifted by the predicted amount of imbalance from the total power generation plan within the area (see the dotted line extending vertically in Figure 4).

[0050] If a method is adopted in which the amount of imbalance generated and the adjustment capacity kWh price merit order are broken down and the predicted value is calculated based on a calculation formula established in the system design, it will be useful in that it will be possible to explain the cause, for example, when prices soar, whether the amount of imbalance generated was large or the slope of the adjustment capacity kWh price merit order was large.

[0051] When supply and demand are tight, a correction fee calculation index can be calculated from the results of a power source operation model for area-specific demand forecasts predicted by AI, and information on supply and demand regularly released by general electricity transmission and distribution companies, etc., and this value can be used to calculate a predicted value based on the ``corrected imbalance fee when supply and demand are tight'' defined in the imbalance fee system.

[0052] When photovoltaic and wind power output is suppressed, when power source III is suppressed, or when a blackout occurs, the charge may be calculated using a formula defined in the imbalance charge system based on continuously updated supply and demand forecasts and system information.

[0053] Since there is uncertainty as to which calculation method will be applied at the time of prediction, an ensemble prediction may be made using the probability that each calculation method will be applied. For example, if the probability of normal times at the time of prediction is 50% and the probability of output curtailment occurring is 50%, the final prediction value may be obtained by taking the average of the predicted values ​​for normal times and output curtailment. Also, as with the spot market, trend correction may be performed using price predictions such as AI.

[0054] Because the amount of imbalance generated is predicted, when using interconnection line information, it is possible to simulate the effect of imbalance netting (reducing the amount of imbalance generated) through wide-area operation.

[0055] The adjustment capacity kWh price merit order may be estimated by sequentially incorporating spot market supply and demand curve information that is made public at noon the previous day. When such an embodiment is adopted, it is possible to reflect the most recent power source operation status, thereby improving the prediction accuracy and explainability.

[0056] [Forecast for the ΔkW supply and demand adjustment market] For forecasts in the ΔkW balancing market, the following input data may be used (see Figure 5): Past performance data on contract prices, solicitation volume, bidding volume, upper limit of interconnection lines, etc. Spot market execution results (unexecuted amounts by area and their price ranges, etc.) Renewable energy-related forecast values ​​as of two days ago Past performance data may be used as input information when generating or correcting a ΔkW supply and demand adjustment market model using machine learning.

[0057] The area-specific solicitation volume shown in Figure 5 may be predicted using AI (specifically, a ΔkW supply and demand adjustment market model) based on the characteristics of each product and information released by general electricity transmission and distribution companies. For example, JPEG2025152395000003.jpg1254 provides the adjustment power to respond to renewable energy forecast errors, so weather forecasts such as solar radiation and wind speed can be used as explanatory variables. The area-specific bidding curves shown in Figure 5 may be created for each area based on the supply and demand adjustment market guidelines, using a power source operation model, contract records for each electricity market price, etc. The bidding curves created appropriately from the contract records for each product may be corrected using AI (specifically, a ΔkW supply and demand adjustment market model). The upper limit of interconnection lines shown in Figure 5 may be predicted based on the amount of interconnection lines secured as determined by the system design, past performance, etc. The upper limit of interconnection lines may also be predicted using AI (specifically, a ΔkW supply and demand adjustment market model).

[0058] The maximum contract price for each area (power source area) may be calculated by simulating wide-area procurement that minimizes the adjustment capacity cost while taking into account the upper limit of the interconnection line (see the "Table" at the bottom of Figure 5). JPEG2025152395000004.jpg1153 If wide-area procurement is difficult due to DC interconnection, this calculation may be performed for each area group where wide-area procurement is possible.

[0059] Because the ΔkW supply-demand balancing market uses a multi-price auction, bids below the highest price are expected to be executed. Predictions using the ΔkW supply-demand balancing market model can be performed probabilistically (see the graph in the lower right of Figure 5). Alternatively, the relationship between ΔkW and execution probability can be quantitatively modeled by calculating the highest execution price by area (power source location) under multiple scenarios. This approach provides information useful for considering bidding strategies, such as the execution probability when bidding at a certain price, or, conversely, how much lower the bid price should be if a execution probability of, say, 80% or higher is desired. Furthermore, the expected-value-optimal bid price can be automatically calculated by calculating "ΔkW bid price x execution probability."

[0060] In the aspect shown in Figure 5, information related to bidding in the supply and demand balancing market is predicted and the ΔkW contract price is calculated by simulating actual contract rules, so it is possible to provide a model that can explain the causes, such as whether the amount solicited was high when the price soared, whether the amount of bids received was low, or whether efficient wide-area procurement was not achieved due to upper limit constraints on interconnection lines.

[0061] In the supply and demand adjustment market, it is expected that changes to the system design will continue in the future, such as dividing the bidding unit into 30 parts and introducing composite agreements. However, if a method is adopted in which the contract rules are simulated and predicted after assuming the solicitation volume and bid volume of each product, it will be beneficial in that even if the system is changed, it can be quickly responded to by changing the processing on the contract rule side.

[0062] [Forecast for the kWh supply and demand adjustment market] The required adjustment capacity for JPEG2025152395000005.jpg1052 and tertiary (tertiary adjustment capacity) is the adjustment capacity for the 30-minute average value of the forecast error and is in the area of ​​EDC (Economic Load Dispatching Control), so as with the amount of imbalance generation, appropriate predictions can be made using AI, etc. in accordance with updates to weather forecasts and area supply and demand forecasts, etc. Primary (primary control capacity) and The required adjustment capacity for JPEG2025152395000006.jpg1154 is a fine adjustment capacity within 30 minutes, and is in the area of ​​LFC (Load Frequency Control) and GF (Governor Free), so it is difficult to predict based on publicly available information. However, it is possible to estimate it by narrowing the range to a certain extent, for example, from the required adjustment capacity for EDC mentioned above.

[0063] Regarding the adjustment capacity kWh price merit order, while using the model assumed in the imbalance price model for the overall adjustment capacity kWh (see the dashed arrow in the "graph" on the left side of Figure 6), it is also possible to make estimates for each product by aggregating the marginal costs of the power source groups corresponding to each adjustment capacity from the power source operation model. When using a model similar to the imbalance price model to forecast the required adjustment capacity and estimate the adjustment capacity kWh price merit order, it is beneficial in that it can eliminate discrepancies in the forecast results between the imbalance price model and the kWh supply and demand adjustment market model.

[0064] Because forecasts of required control capacity for each product involve a certain degree of uncertainty, probabilistic forecasts can be made based on past forecast errors. It is also possible to aggregate the probability of receiving a command when a certain control capacity kWh price is registered (see "Control Capacity Occurrence Probability" in Figure 6). In this case, it is also possible to model the relationship between the control capacity kWh price and command probability (see "Graph" on the right side of Figure 6). Uncertainty in commands in the supply and demand balancing market is a major challenge in formulating operational plans, especially for pumped storage power generation and storage batteries, which have continuous output constraints. However, being able to predict future supply surpluses and shortages based on command probability is beneficial in that it allows for operations that minimize penalties without compromising business viability by utilizing the time-ahead market.

[0065] Businesses participating in the adjustment power kWh market can determine whether or not a command has been issued for the kWh price they have registered, and so this data can be learned using AI or other means to correct the kWh supply and demand adjustment market model.

[0066] The prediction may be made by simulating actual commands from general electricity transmission and distribution companies by breaking down the forecast of required control capacity and the control capacity kWh price merit order. In this case, for example, if a command is received despite a high kWh unit price being registered, the model can explain the cause, such as whether the required control capacity was large or the slope of the control capacity kWh merit order was steep.

[0067] [Pre-market forecast] In the time-ahead market model, future contract trends may be predicted for each predetermined area. As mentioned above, the size of the area may be determined by prefecture, or by predetermined regions such as Hokkaido, Tohoku, Kanto, Chubu, Kinki, Chugoku, Shikoku, and Kyushu (including Okinawa), or by the areas under the jurisdiction of general electricity transmission and distribution companies, as shown in Figure 5 (Hokkaido, Tohoku, Tokyo, Chubu, Hokuriku, Kansai, Chugoku, Shikoku, and Kyushu). Regarding the assumption of execution results by area, participants in the pre-hours market can obtain quote information (bid volume and bid price), but because area information is unknown, market fragmentation may result in no execution even if an offer price lower than the minimum bid price is received (see "graph" in the upper left of Figure 7). In this regard, the pre-hours market model can prevent such a situation (non-execution) from occurring by assuming the trend of execution results by area using quote information, spot market-related information for the same day, interconnection line information, etc. (see "graph" in the lower left of Figure 7). Considering that the time-ahead market is a market for achieving simultaneous balancing of planned values ​​and reducing imbalances, it is expected that there will be a certain correlation with the prediction of imbalance prices using an imbalance price model and the prediction results using related information, etc. Therefore, it is possible to construct an area-specific execution price prediction model using AI, etc., and to generate reliable area-specific execution performance data (correct values ​​in this prediction) by appropriately adjusting the parameters of the area-specific execution performance assumption logic (by using machine learning) so that the correlation between the prediction results and actual performance (prediction accuracy) is high.

[0068] Currently, transaction results by area are not made public. However, due to the uneven regional distribution of power generation and demand, market segmentation has a large impact on pre-hour market transactions. Therefore, by making plausible assumptions and forecasts of transaction results by area based on the high correlation with information such as the results of imbalance price forecasts, it is possible to provide important information for pre-hour market transactions.

[0069] Market participants can obtain quote information and execution results as they occur, but because the pre-hours market model uses a continuous trading method, there is a possibility that they will be able to sell at a higher price or buy at a lower price in the future. By predicting future execution trends by area, it is possible to improve profitability.

[0070] The acquisition unit 10, prediction unit 20, output unit 80, etc. shown in FIG. 8 may be realized by one unit (control unit) or by different units. The functions of multiple "units" may be integrated. Furthermore, the acquisition unit 10, prediction unit 20, output unit 80, etc. may be realized by a circuit configuration. The information processing device 1 has a processor, and the processor may execute a program to realize various functions of the acquisition unit 10, prediction unit 20, output unit 80, etc. of the information processing device 1 of this embodiment.

[0071] The above description of the embodiment and the disclosure of the drawings are merely examples for explaining the invention described in the claims, and the invention described in the claims is not limited by the above description of the embodiment or the disclosure of the drawings. [Explanation of symbols]

[0072] 10 Acquisition Department 20 Prediction Department 80 Storage section 90 Output section

Claims

1. an acquisition unit that acquires information about operating power sources by applying the power demand forecast to a power source operation model; a prediction unit that predicts an electricity price in a first system by applying information about the operating power source to a first individual model, and predicts an electricity price in a second system different from the first system by applying information about the operating power source to a second individual model; An information processing device comprising:

2. 2. The information processing device according to claim 1, wherein the electricity price in the first system is any one of an agreement price in the spot market, an agreement price in the ΔkW supply and demand adjustment market, a registered price in the kWh supply and demand adjustment market, a future agreement price by area in the time-ahead market, and an imbalance price, and the electricity price in the second system is any one of an agreement price in the spot market, an agreement price in the ΔkW supply and demand adjustment market, a registered price in the kWh supply and demand adjustment market, a future agreement price by area in the time-ahead market, and an imbalance price, and is an electricity price in a second system different from the first system.

3. the electricity price in the first system is a contract price in a spot market or a contract price in a ΔkW supply and demand adjustment market, The electricity price in the second system is a registered price in the kWh supply and demand adjustment market, a future contract price by area in the time-ahead market, or an imbalance price, 3. The information processing device according to claim 2, wherein the prediction unit predicts the power generation plan and available adjustment capacity of each power source from price agreement records in the first system, and predicts the electricity price in the second system using the results of the prediction.

4. the electricity price in the first system is a contract price in a spot market or a contract price in a ΔkW supply and demand adjustment market, The electricity price in the second system is a registered price in the kWh supply and demand adjustment market, a future contract price by area in the time-ahead market, or an imbalance price, 3. The information processing device according to claim 2, wherein the prediction unit predicts power sources operating at the time of actual supply and demand in the spot market based on price agreement records in the first system, and predicts power sources waiting to operate at the time of actual supply and demand in the ΔkW supply and demand adjustment market, and predicts the electricity price in the second system using the results of the predictions.

5. 5. The information processing device according to claim 1, wherein the prediction unit determines whether a continuously operating power source that operates continuously for a predetermined period of time is operating based on price agreement history, and if it determines that the continuously operating power source is operating, predicts the electricity price in the first system or the second system on the assumption that the continuously operating power source will be operated for the predetermined period of time.

6. an acquisition unit that acquires information about operating power sources by applying the demand forecast to a power source operation model; a prediction unit that predicts electricity prices in the spot market by applying information about the operating power source to a spot market model, predicts contract prices in the ΔkW supply and demand adjustment market by applying information about the operating power source to a ΔkW supply and demand adjustment market model, predicts registered prices in the kWh supply and demand adjustment market by applying information about the operating power source to a kWh supply and demand adjustment market model, predicts future contract prices by area in the time-ahead market by applying information about the operating power source to an hour-ahead market model, and predicts imbalance prices by applying information about the operating power source to an imbalance price model; An information processing device comprising:

7. 7. The information processing device according to claim 6, wherein the prediction unit predicts the power generation plan and available adjustment capacity of each power source from price agreement records in the spot market and the ΔkW supply and demand adjustment market, and predicts the registered price in the kWh supply and demand adjustment market, the future agreement price by area in the time-ahead market, and the imbalance price using the results of the prediction.

8. 7. The information processing device according to claim 6, wherein the prediction unit predicts power sources that are operating at the time of actual supply and demand based on price agreement records in the spot market, and predicts power sources that are waiting to operate at the time of actual supply and demand based on price agreement records in the ΔkW supply and demand adjustment market, and predicts registered prices in the kWh supply and demand adjustment market, future agreement prices by area in the time-ahead market, and imbalance prices using the results of the predictions.

9. 9. The information processing device according to claim 6, wherein the prediction unit determines whether a continuously operating power source that operates continuously for a predetermined time is operating based on price agreement history, and if it determines that the continuously operating power source is operating, predicts the agreement price in the spot market, the agreement price in the ΔkW supply and demand adjustment market, the registered price in the kWh supply and demand adjustment market, and the future agreement price and imbalance price by area in the time-ahead market, assuming that the continuously operating power source will be operated for the predetermined time.

10. acquiring information about the operating power sources by applying the demand forecast to a power source operation model by an acquisition unit; a step of predicting, by a prediction unit, a contract price in the spot market by applying information about the operating power source to a spot market model, predicting a contract price in the ΔkW supply and demand adjustment market by applying information about the operating power source to a ΔkW supply and demand adjustment market model, predicting a registered price in the kWh supply and demand adjustment market by applying information about the operating power source to a kWh supply and demand adjustment market model, predicting a future contract price by area in the time-ahead market by applying information about the operating power source to an hour-ahead market model, and predicting an imbalance price by applying information about the operating power source to an imbalance price model; An information processing method comprising:

11. A program to be installed on an information processing device, On the information processing device on which the program is installed, The function of applying demand forecasts to a power generation operation model to obtain information about the power generation in operation; a function of predicting a contract price in the spot market by applying information about the operating power source to a spot market model, predicting a contract price in the ΔkW supply and demand adjustment market by applying information about the operating power source to a ΔkW supply and demand adjustment market model, predicting a registered price in the kWh supply and demand adjustment market by applying information about the operating power source to a kWh supply and demand adjustment market model, predicting future contract prices by area in the time-ahead market by applying information about the operating power source to an hour-ahead market model, and predicting an imbalance price by applying information about the operating power source to an imbalance price model; A program that makes this happen.

Images