Electric power market price prediction apparatus, electric power market price prediction method, and program

Abstract

To predict, with high accuracy, an electric power spot market price of JEPX on a prediction target date.SOLUTION: An intersection power source kind specifying and intersection capacity calculating unit 32 calculates an intersection capacity by specifying an intersection power source kind for each of actual values of an electric power spot market price on a comparison reference date by using an actual value of an electric power supply amount for each power source kind of each time slot of the comparison reference date and an actual value of a system demand thereof. An MOC creation unit 34 creates a merit order curve on a prediction target date by plotting each of the actual values of the electric power spot market price of each time slot of the comparison reference date predicted by the electric power supply amount prediction unit 33. A price prediction unit 36 predicts the electric power prediction target date by calculating an intersection of prices of the prediction target time slots on the target an electric power demand amount on the prediction target date predicted by an electric power demand prediction unit 35 and the created merit order curve.SELECTED DRAWING: Figure 6

Description

Technical Field

[0001] The present invention relates to a power market price prediction device, a power market price prediction method, and a program.

Background Art

[0002] With the liberalization of retail power, various operators have entered the power trading market as retail electricity providers or power generation providers. And these retail electricity providers and power generation providers conduct power trading in the power spot market (one-day-ahead market) operated by the Japan Electric Power Exchange (JEPX). In this power spot market, power trading is conducted every 30 minutes for 48 periods in a day. And in the power spot market, the transaction price and transaction volume are determined by an agreement method called a blind single price auction. And this transaction price becomes the power spot market price for each period, and the power transaction is executed.

[0003] Power generation providers can formulate an optimal power generation plan by predicting the power spot market price. Also, power generation providers and retail electricity providers can take economically optimal bidding actions by predicting the power spot market price.

[0004] Therefore, various methods for predicting the power spot market price have been proposed (see, for example, Patent Documents 1 and 2).

[0005] Patent Document 1 discloses a supply stack prediction model for the power generation capacity and power generation cost of each of a plurality of thermal power plants, creates a maximum power demand curve from past maximum power demand data and predicted maximum temperature value data, calculates the output value of nuclear power generation, etc. on the prediction date from past performance data, calculates the maximum power demand value on the prediction date from the intersection of the maximum power demand curve and the maximum temperature value on the prediction date, subtracts the output value of nuclear power generation, etc. from the maximum power demand value to calculate the maximum thermal power demand value on the prediction date, and predicts the electricity market price on the prediction date from the intersection of the maximum thermal power demand value and the supply stack prediction model.

[0006] Further, Patent Document 2 discloses a power market price prediction method including an acquisition unit that acquires an estimated supply amount of power generated by renewable energy on a future prediction target date, and a prediction unit that predicts the power market price on the prediction target date based on the estimated supply amount.

Prior Art Documents

Patent Documents

[0007]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0008] In the conventional prediction methods for predicting the electricity market price as disclosed in Patent Document 1 or Patent Document 2 as described above, the electricity market price on the prediction target date is predicted from the relationship between the electricity demand and supply on the prediction target date.

[0009] However, even if the power demand is the same value, the electricity spot market price fluctuates depending on the power generation amount of renewable energy and the power generation capacity of other power sources. Therefore, simply predicting the electricity spot market price on the target prediction date based on the relationship between the power demand and the power supply on the target prediction date cannot achieve a high-precision prediction.

[0010] An object of the present invention is to provide a power market price prediction device, a power market price prediction method, and a program capable of predicting the electricity spot market price of the Japan Electric Power Exchange (JEPX) on the target prediction date with high accuracy.

Means for Solving the Problems

[0011] The power market price prediction device according to the first aspect of the present invention uses a plurality of days within a preset past period in which the electricity spot market price is known as comparison reference days, and uses the actual values of the power supply amounts for each power source type and the actual values of the system demand for each tick of the comparison reference days. For each actual value of the electricity spot market price on the comparison reference day, identify the intersection power source type, which is the power source type to which the power generation facilities expected to be involved in the determination of the electricity spot market price belong. When arranging the power generation facilities included in the identified intersection power source type in ascending order of marginal cost, calculate the intersection capacity, which is the cumulative value of the power generation amounts from the power generation facility with the lowest marginal cost to the power generation facilities expected to be involved in the determination of the electricity spot market price. An intersection power source type identification and intersection capacity calculation unit; A supply amount prediction unit that predicts the power supply amount for each power source type on the target prediction date; On a graph using the predicted values of the power supply amounts for each power source type on the target prediction date predicted by the supply amount prediction unit, plot the actual values of the electricity spot market prices for each tick of the comparison reference day based on the intersection power source type identified by the intersection power source type identification and intersection capacity calculation unit and the calculated intersection capacity, thereby creating a merit order curve on the target prediction date. A creation unit; A demand amount prediction unit that predicts the power demand for each target tick on the target prediction date; A price prediction unit that predicts the electricity spot market price for each prediction target period on the prediction target day by obtaining the intersection of the predicted electricity demand by the demand prediction unit and the merit order curve created by the creation unit; is provided.

[0012] The electricity market price prediction device according to the second aspect of the present invention is the electricity market price prediction device according to the first aspect, wherein the creation unit creates merit order curves that differ in any time period with better prediction accuracy.

[0013] The electricity market price prediction device according to the third aspect of the present invention is the electricity market price prediction device according to the first aspect, wherein when the month to which the comparison reference date belongs is different from the month to which the prediction target date belongs, the creation unit corrects the fuel price for the actual value of the electricity spot market price for each period of the comparison reference date and then creates a merit order curve by plotting.

[0014] The electricity market price prediction device according to the fourth aspect of the present invention is the electricity market price prediction device according to the first aspect, wherein when there are a plurality of plots on the merit order curve whose difference from the predicted electricity demand is equal to or less than a preset electricity capacity, the price prediction unit weights the actual values of the electricity spot market prices of the plurality of plots such that the weighting coefficient increases as the number of days difference from the prediction target date decreases, and calculates a weighted average to predict the electricity spot market price for the prediction target period on the prediction target day.

[0015] The power market price prediction device according to the fifth aspect of the present invention is the power market price prediction device according to the first aspect, wherein the price prediction unit, on the merit order curve, when there are no multiple plots whose difference from the predicted power demand is equal to or less than a preset power capacity, performs cluster classification on the multiple plots constituting the merit order curve, and among the multiple classified clusters, selects the cluster with the smallest difference between the center of gravity position of each cluster and the predicted power demand, calculates an approximate straight line with the shortest distance to the multiple plots belonging to the selected cluster, and inputs the predicted power demand into the calculated approximate straight line, thereby predicting the power spot market price of the prediction target period on the prediction target day.

[0016] The power market price prediction method according to the sixth aspect of the present invention uses a plurality of days within a preset past period in which the power spot market price is known as comparison reference days, and uses the actual value of the power supply volume for each power source type and the actual value of the system demand for each period of the comparison reference days to identify the intersection power source type, which is the power source type to which the power generation facilities expected to be involved in determining the power spot market price belong, for each actual value of the power spot market price of the comparison reference days. When arranging the power generation facilities included in the identified intersection power source type in ascending order of marginal cost, an intersection capacity calculation step of calculating the intersection capacity, which is the cumulative value of the power generation volume from the power generation facility with the lowest marginal cost to the power generation facilities expected to be involved in determining the power spot market price; A supply volume prediction step of predicting the power supply volume for each power source type on the prediction target day; A creation step of creating a merit order curve on the prediction target day by plotting the actual values of the power spot market prices of each period of the comparison reference days based on the intersection power source type identified by the intersection power source type identification and intersection capacity calculation unit and the calculated intersection capacity on a graph using the predicted values of the power supply volume for each power source type on the prediction target day predicted in the supply volume prediction step; A demand volume prediction step of predicting the power demand for each prediction target period on the prediction target day; A price prediction step of predicting the electricity spot market prices of the prediction target periods on the prediction target day by obtaining the intersections between the predicted electricity demand amount in the demand prediction step and the merit order curve created in the creation step, respectively.

[0017] The program according to the seventh aspect of the present invention uses a plurality of days within a preset past period in which the electricity spot market price is known as comparison reference days, and uses the actual values of the power supply amounts for each power source type of each period of the comparison reference days and the actual values of the system demand to identify, for each actual value of the electricity spot market price of the comparison reference days, the intersection power source type which is the power source type to which the power generation facilities expected to be involved in the determination of the electricity spot market price belong. When arranging the power generation facilities included in the identified intersection power source type in ascending order of the marginal cost, an intersection capacity calculation step of calculating the cumulative value of the power generation amounts from the power generation facility with the lowest marginal cost to the power generation facilities expected to be involved in the determination of the electricity spot market price, which is the intersection capacity. A supply amount prediction step of predicting the power supply amounts for each power source type on the prediction target day. On a graph using the predicted values of the power supply amounts for each power source type on the prediction target day predicted in the supply amount prediction step, by plotting the actual values of the electricity spot market prices of each period of the comparison reference days based on the intersection power source type identified by the intersection power source type identification and intersection capacity calculation unit and the calculated intersection capacity, a creation step of creating a merit order curve on the prediction target day. A demand prediction step of predicting the electricity demand amounts of the prediction target periods on the prediction target day, respectively. A price prediction step of predicting the electricity spot market prices of the prediction target periods on the prediction target day by obtaining the intersections between the predicted electricity demand amount in the demand prediction step and the merit order curve created in the creation step, respectively. To be executed by a computer.

Advantages of the Invention

[0018] According to the present invention, it becomes possible to accurately predict the power spot market price of the Japan Electric Power Exchange on the prediction target date.

Brief Description of the Drawings

[0019]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16

Figure 17

Figure 18

Embodiments for Carrying Out the Invention

[0020] Next, embodiments of the present invention will be described in detail with reference to the drawings.

[0021] Before explaining the configuration of the power market price prediction device according to an embodiment of the present invention, the method for determining the power spot market price in JEPX will be explained. The method for determining the power spot market price in JEPX will be explained with reference to FIG. 1.

[0022] As described above, in the power spot market in JEPX, the contract price and contract quantity of power for the next day are determined by an auction method called a blind single price auction.

[0023] Specifically, the power generation company, which is the seller, makes a bid to the trading system, such as "sell 1000 kWh of electricity at a price of 15 yen / kWh or more". Also, the retail electricity company, which is the buyer, makes a bid to the trading system, such as "buy 2000 kWh of electricity at a price of 10 yen / kWh or less".

[0024] Then, in the JEPX trading system, as shown in Figure 1, at a fixed time every day, based on the bid details, a sell bid curve is created by stacking sell bids in ascending order of the bid price, and a buy bid curve is created by stacking buy bids in ascending order of the bid price. Then, by finding the intersection point of this sell bid curve and buy bid curve, the agreed price and the agreed quantity are determined. This agreed price is the electricity spot market price, and all electricity transactions will be carried out at this electricity spot market price. And the determination of such an electricity spot market price is carried out for each of the 48 frames obtained by dividing the next day into 30-minute units respectively.

[0025] Note that it is considered a desirable act for the power generation company to bid the sell bid price of electricity at the marginal cost (variable cost) at the time of power generation. Therefore, the sell bid price will be approximately the same as the marginal cost at the time of power generation.

[0026] Here, the marginal cost means the additional cost incurred when the power generation amount is increased. That is, at a certain power plant, the additional cost generated to increase the power generation amount by 1 kWh is the marginal cost at that power plant.

[0027] And if the electricity spot market price determined by the above method can be predicted in advance, the power generation company and the retail electricity company can optimize the sell bid price, the buy bid price, and the amount of electricity to be bid. Also, the power generation company can create plans such as operating the power plant during time periods when the electricity spot market price is likely to increase.

[0028] Therefore, in the electricity market price prediction device of this embodiment, a merit order curve (hereinafter may be abbreviated as MOC) on the prediction target date for predicting the electricity spot market price is created, and the electricity spot market price on the prediction target date is predicted using the created merit order curve.

[0029] Here, the merit order means arranging power plants (power generation facilities) of various power source types in ascending order of marginal cost. For example, the power source types will be described by dividing them into the following five types.

[0030] (1) Renewable energy: Power generation using renewable energy such as solar power generation, wind power generation, geothermal power generation, and hydroelectric power generation (2) Nuclear power: Nuclear power generation (3) Coal-fired power: Thermal power generation using coal (4) Gas-fired power: Thermal power generation using natural gas, etc. (5) Oil-fired power: Thermal power generation using oil

[0031] An example of a merit order curve when arranging power plants of these five types of power source types in ascending order of marginal cost is shown in FIG. 2. It should be noted that the power sources of the above five types of power source types generally increase in marginal cost in the order of (1) renewable energy, (2) nuclear power, (3) coal-fired power, (4) gas-fired power, and (5) oil-fired power.

[0032] And the supply curve created by accumulating the power generation amounts of the power plants arranged in merit order is the merit order curve (or merit order curve). That is, the merit order curve corresponds to the sell bid curve in FIG. 1. And from the agreement principle of the electricity spot market price in JEPX as described above, it is considered that the intersection of this merit order curve and the straight line indicating the system demand is the electricity spot market price.

[0033] Here, the system demand refers to the total amount of power demand in a certain power system. Also, the power system refers to a system that transmits and distributes electricity for each area of Hokkaido, Tohoku, Tokyo, Chubu, Hokuriku, Kansai, Chugoku, Shikoku, Kyushu, and Okinawa. For each of these power systems (excluding Okinawa), the power spot market price is also determined.

[0034] Even within a day, the power demand changes moment by moment. Therefore, in order to meet this power demand, it is most economical to operate the power generation facilities in the order of the lowest generation cost (= operate the power plants in the above merit order). That is, generally, power generated by "renewable energy" that does not require fuel costs is preferentially used, followed by power generated by "nuclear power", and power is supplied to meet the power demand by operating power plants in the order of coal-fired, gas-fired, and oil-fired power generation.

[0035] Note that in Figure 2, a simple MOC is illustrated with the marginal costs of power generation facilities of the same power source type set to the same value. However, in reality, even for power generation facilities of the same power source type, the marginal costs vary for each power generation facility. An example of a merit order curve when the marginal costs vary for each power generation facility is shown in Figure 3.

[0036] Referring to Figure 3, it can be seen that even for the same power source type of "gas-fired power", the marginal costs are different for each of power plants A to I. And in Figure 3, in order to meet the system demand in a certain power system, it is necessary to operate power generation facilities A to E among "renewable energy", "nuclear power", "coal-fired power", and "gas-fired power".

[0037] Here, among the power generation facilities operated to meet the system demand, it is considered that the marginal cost of the power generation facility with the highest marginal cost (marginal plant) is approximately the same as the power spot market price. That is, in Figure 3, it is considered that the marginal cost of power plant E is the power spot market price at a certain time on a certain day in a certain power system.

[0038] Therefore, in the electricity market price prediction device of the present embodiment, based on the idea that "the past electricity spot market price ≒ the marginal cost of a certain power plant", the MOC on the prediction target date is generated, and the intersection with the electricity demand on the prediction target date is obtained, thereby predicting the electricity spot market price on the prediction target date.

[0039] First, the system configuration of a system including the electricity market price prediction device 10 according to an embodiment of the present invention is shown in FIG. 4.

[0040] The electricity market price prediction device 10 of the present embodiment is connected to a plurality of server devices 20 capable of acquiring various information via a network such as the Internet 30.

[0041] Specifically, the plurality of server devices 20 are server devices of JEPX capable of acquiring the actual values of past electricity spot market prices, server devices of the Japan Meteorological Agency capable of acquiring the weather information on the prediction target date, and server devices in HJKS (Power Generation Information Disclosure System) capable of acquiring the operation information of power generation facilities on the prediction target date and in the past.

[0042] Next, the hardware configuration of the electricity market price prediction device 10 of the present embodiment is shown in FIG. 5.

[0043] As shown in FIG. 5, the electricity market price prediction device 10 includes a CPU 11, a memory 12, a storage device 13 such as a hard disk drive, a communication interface (abbreviated as IF) 14 that transmits and receives data to and from an external server device 20 via the Internet 30, and a user interface (abbreviated as UI) device 15 including a touch panel or a liquid crystal display and a keyboard. These components are connected to each other via a control bus 16.

[0044] The CPU 11 is a processor that executes predetermined processing based on a control program stored in the memory 12 or the storage device 13 to control the operation of the electricity market price prediction device 10. In the present embodiment, the CPU 11 has been described as reading and executing a control program stored in the memory 12 or the storage device 13, but it is not limited thereto. This control program may be provided in a form recorded on a computer-readable recording medium.

[0045] FIG. 6 is a block diagram showing the functional configuration of the electricity market price prediction device 10 realized by executing the above control program.

[0046] As shown in FIG. 6, the electricity market price prediction device 10 of the present embodiment includes an information acquisition unit 31, an intersection power source type identification and intersection capacity calculation unit 32, a power supply amount prediction unit 33, an MOC creation unit 34, a power demand amount prediction unit 35, a price prediction unit 36, and a display unit 37.

[0047] The information acquisition unit 31 acquires various information such as information on past actual values of electricity spot market prices, weather information such as weather forecast information on the prediction target date, operation information of power generation facilities, and information on past grid demands from the server device 20.

[0048] The intersection power source type identification and intersection capacity calculation unit 32 uses, as comparison reference dates, a plurality of days within a preset past period in which the electricity spot market price is known, and uses the actual values of the power supply amounts for each power source type and the actual values of the grid demands for each tick of the comparison reference dates to identify, for each actual value of the electricity spot market price of the comparison reference dates, the intersection power source type, which is the power source type to which the power generation facilities expected to be involved in the determination of the electricity spot market price belong.

[0049] Then, the intersection power source type identification and intersection capacity calculation unit 32 calculates the intersection capacity, which is the cumulative value of the power generation amounts from the power generation facility with the lowest marginal cost to the power generation facilities expected to be involved in the determination of the electricity spot market price when the power generation facilities included in the identified intersection power source type are arranged in ascending order of marginal cost.

[0050] The power supply amount prediction unit 33 predicts the power supply amount for each power source type on the prediction target date. Specifically, the power supply amount prediction unit 33 divides power sources into natural variable power sources such as solar power generation, wind power generation, and hydroelectric power generation, whose power generation amount fluctuates depending on weather conditions, and whose power source type is "renewable energy", and non-variable power sources such as thermal power generation and nuclear power generation, and predicts the power supply amount by the following method.

[0051] (1) Natural variable power sources (solar power generation, wind power generation, hydroelectric power generation, etc.) For natural variable power sources, the information acquisition unit 31 acquires weather information such as solar radiation amount and wind speed from the server device 20. Then, the power supply amount prediction unit 33 performs machine learning using the actual values of the weather information such as solar radiation amount and wind speed acquired by the information acquisition unit 31 as feature quantities and the power generation amount as the target variable to generate a prediction model. Then, the power supply amount prediction unit 33 inputs the weather information on the prediction target date to the generated learned prediction model to obtain a predicted value of the power generation amount.

[0052] (2) Non-variable power sources (thermal power generation, nuclear power generation, etc.) For non-variable power sources, the information acquisition unit 31 acquires the operation information of thermal power and nuclear power plants in the prediction target date and prediction target area from the HJKS (Power Generation Information Disclosure System). Then, the power supply amount prediction unit 33 collects information on the power generation capacity on the prediction target date for each of thermal power generation such as coal-fired thermal power, gas-fired thermal power, and oil-fired thermal power, and nuclear power generation from the operation information acquired by the information acquisition unit 31.

[0053] The MOC creation unit 34 plots the actual values of the power spot market prices for each cell on the comparison reference date on a graph using the predicted values of the power supply amounts for each power source type on the prediction target date predicted by the power supply amount prediction unit 33, based on the intersection power source type specified by the intersection power source type identification and intersection capacity calculation unit 32 and the calculated intersection capacity, thereby creating a merit order curve on the prediction target date.

[0054] The power demand prediction unit 35 predicts the power demand amounts for the prediction target time period and the prediction target area for each prediction target time slot using the actual power demand values and weather information acquired by the information acquisition unit 31. For example, the power demand prediction unit 35 predicts the power demand amounts for the prediction target time period and the prediction target area for each prediction target time slot using the power demand prediction method disclosed in Japanese Patent No. 7062144.

[0055] Specifically, the power demand prediction unit 35 captures weather data groups including at least air temperature and solar radiation amount in units obtained by subdividing this fixed period, and actual power demand values, from a time period going back by a fixed period, and classifies them into patterns in which the changes in these actual values are similar. Then, for each classified pattern, the power demand prediction unit 35 creates a long-term regression model with the weather data group as the explanatory variable and the actual power demand value as the objective variable. Then, the power demand prediction unit 35 selects a predetermined number of short-term reference days close to the prediction target day from patterns similar to the prediction target day for power demand, captures the weather data group and the actual power demand values, and weights and blends a plurality of power demand prediction values for the prediction target day predicted by each of a plurality of prediction models using the created long-term regression model and the weather data group and the actual power demand values of the short-term reference days at a predetermined ratio, to obtain the power demand prediction value for the prediction target day.

[0056] The price prediction unit 36 predicts the electricity spot market prices for each prediction target time slot on the prediction target day by obtaining the intersections of the power demand amounts predicted by the power demand prediction unit 35 and the merit order curve created by the MOC creation unit 34.

[0057] The display unit 37 displays the predicted values of the electricity spot market prices for each prediction target time slot on the prediction target day predicted by the price prediction unit 36.

[0058] Next, the operation of the electricity market price prediction device 10 of the present embodiment will be described in detail with reference to the drawings.

[0059] In the electricity market price prediction device 10 of this embodiment, the overall operation when predicting the electricity spot market price on the prediction target day is shown in the flowchart of FIG. 7.

[0060] First, the intersection power source type identification and intersection capacity calculation unit 32 uses the past days (the period is set in advance) when the electricity spot market price is known as the comparison reference days, and for each actual value of the electricity spot market on this comparison reference day, identifies the intersection power source type and calculates the intersection capacity (step S101). Note that the intersection power source type identification and intersection capacity calculation unit 32 performs the identification of the intersection power source type and the calculation of the intersection capacity for all frames of the entire comparison reference day.

[0061] For example, if the last 7 days are used as the comparison reference days, when the prediction target day is August 1, 2023, the intersection power source type identification and intersection capacity calculation unit 32 uses the 7 days from July 25 to July 31, 2023 as the comparison reference days to identify the intersection power source type and calculate the intersection capacity.

[0062] An example of the identification of the intersection power source type and the calculation of the intersection capacity performed by the intersection power source type identification and intersection capacity calculation unit 32 in this way is shown in FIG. 8. FIG. 8 explains the case where the electricity spot market price of the 20:00 frame on July 30, 2023 of the comparison reference day is 28 yen / kWh.

[0063] First, on a graph with the cumulative capacity on the X-axis and the price on the Y-axis, arrange the actual values of the power generation amounts for each power source type on the comparison reference day in ascending order of the marginal cost. Then, plot on this graph at the location specified by the actual value of the electricity spot market price of 28 yen / kWh on the comparison reference day and the actual value of the system demand. If the actual value of the system demand on the comparison reference day is not published, the system demand on the comparison reference day is predicted by the power demand prediction unit 35, and the predicted value is used as the value of the system demand.

[0064] Then, the intersection power source type identification and intersection capacity calculation unit 32 identifies the intersection power source type from the intersection of the vertical line dropped from this plot location to the X-axis and the actual values of the arranged power source types. In the specific example shown in FIG. 8, the intersection power source type is identified as "gas-fired power".

[0065] Next, the intersection power source type identification and intersection capacity calculation unit 32 calculates the distance from the left end of the identified power source type to the plotted position as the intersection capacity. In the specific example shown in FIG. 8, the intersection capacity is calculated to be 21 million kW. That is, the intersection capacity is a value indicating where the power generation facilities involved in determining the electricity spot market price on the comparison reference date are located when the power generation facilities of the intersection power source type are arranged in ascending order of marginal cost.

[0066] Next, the power supply amount prediction unit 33 predicts the power supply amount for each power source type on the prediction target date (step S102).

[0067] Then, the MOC creation unit 34 plots the actual values of the electricity spot market price for each frame on the comparison reference date on a graph using the predicted values of the power supply amount for each power source type on the prediction target date predicted by the power supply amount prediction unit 33, based on the intersection power source type identified by the intersection power source type identification and intersection capacity calculation unit 32 and the calculated intersection capacity, thereby creating a merit order curve on the prediction target date (step S103). The details of the method for creating this merit order curve on the prediction target date will be described later.

[0068] Here, when the comparison reference date is 7 days, the MOC creation unit 34 creates a merit order curve with 336 plots (= 7 days × 48 frames).

[0069] Next, the power demand amount prediction unit 35 predicts the power demand amount for each prediction target frame on the prediction target date and in the prediction target area using the actual values of the power demand and the weather information acquired by the information acquisition unit 31 (step S104).

[0070] Then, the price prediction unit 36 predicts the electricity spot market prices of the target periods on the target date by obtaining the intersection points between the electricity demand volume predicted by the electricity demand volume prediction unit 35 for the target date and the merit order curve created by the MOC creation unit 34 (step S105). Note that the details of the method for predicting the electricity spot market price on the target date will also be described later.

[0071] [Method for Creating MOC] Next, the details of the method for creating the merit order curve on the target date described in step S103 above will be described.

[0072] FIG. 9 is a flowchart for explaining the method for creating the merit order curve on the target date.

[0073] First, the MOC creation unit 34 arranges the predicted values of the power supply volume for each power source type on the target date predicted by the power supply volume prediction unit 33 in ascending order of marginal cost (lowest price) on a graph with the cumulative capacity, which is the cumulative value of the power volume, on the X-axis and the price on the Y-axis (step S201). The state in which the predicted values of the power supply volume for each power source type are arranged in this way is shown in FIG. 10.

[0074] Next, the MOC creation unit 34 selects one actual value of the electricity spot market price of a certain target period on a certain comparison reference date (step S202). Here, for example, it will be described as if the actual value of the electricity spot market price of the 20:00 period on July 30, 2023, which was described in FIG. 8, 28 yen / kWh, was selected.

[0075] Next, when the month to which the comparison reference date belongs is different from the month to which the target date belongs, the MOC creation unit 34 performs the fuel price correction described below on the actual value of the electricity spot market price on the selected comparison reference date (step S203).

[0076] [Fuel Price Correction] Since fuel prices may vary from month to month, when the month to which the prediction target date belongs is different from the month to which the comparison reference date belongs, the MOC creation unit 34 corrects the fuel price by the following method. Specifically, when the month to which the comparison reference date belongs is different from the month to which the prediction target date belongs, the MOC creation unit 34 corrects the fuel price for the actual value of the power spot market price for each period of the comparison reference date and then plots it to create a merit order curve.

[0077] Specifically, when the month to which the comparison reference date belongs is different from the month to which the prediction target date belongs, the MOC creation unit 34 creates a merit order curve using the actual value of the power spot market price after fuel price correction calculated by the following formula.

[0078] Actual value of power spot market price (after fuel price correction) = Actual value of power spot market price (before fuel price correction) × (Power generation cost of the month to which the prediction target date belongs / Power generation cost of the month to which the comparison reference date belongs)

[0079] For example, a specific example of the fuel price correction method will be described with reference to the case where the intersection power source type is "gas-fired power generation". Here, it is assumed that the power generation cost in general gas-fired power generation in the month to which the prediction target date belongs is 12 yen / kWh, and the power generation cost in general gas-fired power generation in the month to which the comparison reference date belongs is 10 yen / kWh. That is, the fuel price is higher in the month to which the prediction target date belongs than in the month to which the comparison reference date belongs. And assuming that the actual value of the power spot market price before fuel price correction is 20 yen / kWh, the actual value of the power spot market price after fuel price correction is 24 yen / kWh according to the following calculation formula.

[0080] Actual value of power spot market price (after fuel price correction) = 20 yen / kWh × (12 yen / kWh / 10 yen / kWh) = 24 yen / kWh

[0081] Next, the MOC creation unit 34 plots the actual value of the power spot market price after fuel price correction at the position based on the intersection power source type and intersection capacity on the graph shown in FIG. 10 (step S204).

[0082] In this way, Fig. 11 shows how the actual values of the past electricity spot market prices are plotted. Referring to Fig. 11, the actual value of the electricity spot market price at 20:00 on July 30, 2023, which is 28 yen / kWh, is shown to be plotted at the position where the intersection power source type is gas-fired and the intersection capacity is 21 million kW.

[0083] Here, in Fig. 8, although the actual value of the electricity spot market price of 28 yen / kWh is plotted at the position of the system demand of 45 million kW, in Fig. 11, it is plotted at the position of the cumulative capacity of approximately 55 million kW. This is because in the predicted value of the electricity supply amount on the prediction target date, the power generation amount of the power generation facilities with the power source type of "renewable energy" has increased, so even with the same intersection power source type and intersection capacity, the plotted position changes. That is, on July 30, 2023, which is the comparison reference date, even though the system demand was 45 million kW, the electricity spot market price was 28 yen / kWh, but on August 1, 2023, which is the prediction target date, it means that the electricity spot market price will not be 28 yen / kWh unless the system demand reaches about 55 million kW.

[0084] Then, the MOC creation unit 34 repeats the processing from steps S202 to S204 until the processing for all frames on the comparison reference date is completed (step S205). That is, the MOC creation unit 34 creates a merit order curve by repeating 336 plots (= 7 days × 48 frames).

[0085] An example of the merit order curve created by the MOC creation unit 34 in this way is shown in Fig. 12. Note that in Fig. 12, the number of plots is reduced compared to the actual number for easier viewing of the drawing.

[0086] Note that the MOC creation unit 34 may create different merit order curves in any time period with better prediction accuracy.

[0087] [Prediction of Electricity Spot Market Price on the Date to be Predicted] Next, the details of the method for predicting the electricity spot market price on the date to be predicted, which were described in step S105 of the flowchart in FIG. 7, will be explained.

[0088] The price prediction unit 36 predicts the electricity spot market price of each prediction target period on the date to be predicted by obtaining the intersection point between the electricity demand (system demand) on the date to be predicted predicted by the electricity demand prediction unit 35 and the merit order curve as shown in FIG. 12 created by the MOC creation unit 34.

[0089] FIG. 13 shows a state when a line indicating the system demand of a certain prediction target period on the date to be predicted is drawn on the merit order curve created by the MOC creation unit 34. As shown in FIG. 13, the intersection point between the line indicating this system demand and the merit order curve becomes the predicted value of the electricity spot market price of that prediction target period on the date to be predicted.

[0090] However, the merit order curve created by the MOC creation unit 34 is composed of a plurality of plots and is not defined as a continuous function. Therefore, there are cases where the intersection point between the electricity demand on the date to be predicted and the merit order curve cannot be directly obtained.

[0091] Therefore, the price prediction unit 36 pseudo-obtains the intersection point between the electricity demand on the date to be predicted and the merit order curve by the method described below, and predicts the electricity spot market price on the date to be predicted.

[0092] Specifically, the price prediction unit 36 predicts the electricity spot market price on the date to be predicted by the prediction method shown in the flowchart of FIG. 14.

[0093] First, the price prediction unit 36 extracts plots on the merit order curve that are close to the predicted value of the system demand on the prediction target date (step S301). For example, the price prediction unit 36 extracts plots from among a plurality of plots constituting the merit order curve, where the difference from the predicted value of the system demand is equal to or less than a preset power capacity.

[0094] Then, the price prediction unit 36 determines whether there are a plurality of plots on the merit order curve whose difference from the predicted value of the system demand is equal to or less than a preset power capacity (step S302).

[0095] When there are a plurality of plots on the merit order curve whose difference from the predicted power demand is equal to or less than a preset power capacity (yes in step S302), the price prediction unit 36 weights the actual values of the power spot market prices of the plurality of plots such that the weighting coefficient increases as the number of days difference from the prediction target date decreases, and calculates a weighted average to predict the power spot market price of the prediction target period on the prediction target date (step S303).

[0096] For example, as shown in FIG. 15, a case where there are three plots whose difference from the predicted value of the system demand is equal to or less than a preset power capacity will be described. Here, the price prediction unit 36 predicts the power spot market price of the prediction target date by weighting each comparison reference date using a weighting coefficient calculated by an expression as shown in FIG. 16 and calculating a weighted average.

[0097] Specifically, based on the idea that the operating status of the power generation facilities and the fuel price conditions, which are the premises, are closer as the number of days difference between the prediction target date and the comparison reference date is smaller, the weight at the time of prediction should be increased, and conversely, as the number of days difference is larger, these conditions are different, so the weight should be decreased, a weighting coefficient is set. That is, weighting is performed such that the weighting coefficient increases as the number of days difference from the prediction target date decreases.

[0098] In the formula shown in FIG. 16, X is the number of days difference from the target prediction date, and a and b are constants. The values of a and b are determined by performing parameter search so that the prediction error becomes the smallest when predicting the known latest power spot market price. An example of the weighting coefficient calculated in this way is shown in FIG. 16. Referring to the example of the weighting coefficient shown in FIG. 16, it can be seen that the closer the number of days difference from the target prediction date, August 1, 2023, is, the closer the weighting coefficient is to 1, and as the number of days difference increases, it approaches 0.

[0099] For example, in the example shown in FIG. 15, a specific calculation method will be described assuming that there are the following three plots close to the predicted value of the system demand.

[0100] Plot 1: The comparison reference date is July 30, 2023, and the actual value of the power spot market price is 11.5 yen / kWh Plot 2: The comparison reference date is July 30, 2023, and the actual value of the power spot market price is 11 yen / kWh Plot 3: The comparison reference date is July 29, 2023, and the actual value of the power spot market price is 10 yen / kWh

[0101] And when the weighting coefficient for July 30, 2023 is 1.0 and the weighting coefficient for July 29, 2023 is 0.9, the price prediction unit 36 calculates the weighted average of the three actual values by the following formula to calculate the predicted value of the power spot market price for the target prediction date, August 1, 2023.

[0102] Predicted value of power spot market price = Cumulative value of (actual price of adjacent plot × weighting coefficient of that plot) / Total of weighting coefficients = (11.5×1 + 11×1 + 10×0.9) / (1 + 1 + 0.9) ≈ 10.9 yen / kWh

[0103] Further, when there are no multiple plots on the merit order curve whose difference from the predicted power demand is equal to or less than a preset power capacity (no in step S302), the merit order curve is classified into clusters for a plurality of plots constituting the curve. Among the plurality of classified clusters, the cluster with the smallest difference between the centroid position of each cluster and the predicted power demand is selected, and an approximate straight line with the shortest distance to the plurality of plots belonging to the selected cluster is calculated. By inputting the predicted power demand into the calculated approximate straight line, the electricity spot market price of the target period on the target date is predicted (step S304).

[0104] When there are no multiple plots close to the system demand in this way, a specific example of the method for calculating the predicted value of the electricity spot market price on the target date is shown in FIG. 17.

[0105] (1) First, a plurality of plots constituting the merit order curve created by the MOC creation unit 34 are classified into a plurality of clusters. In the example shown in FIG. 17, a plurality of plots constituting the merit order curve are classified into three clusters, clusters 1 to 3.

[0106] (2) Next, the centroid positions of the three clusters 1 to 3 are obtained respectively, and the distance from each centroid position to the predicted value of the system demand is calculated.

[0107] (3) Then, one cluster with the shortest distance from the centroid position to the predicted value of the system demand is selected. That is, it is considered that the electricity spot market price at the target period also belongs to the cluster with the shortest distance from the centroid position to the predicted value of the system demand. In FIG. 17, it is described that cluster 1 is selected.

[0108] (4) Based on the information of the plurality of plots constituting the selected cluster, an approximate straight line of the cluster is obtained. Specifically, an approximate straight line with the shortest distance to the plurality of plots belonging to the cluster is calculated by obtaining an equation of Y (price) = aX (cumulative capacity) + b (a and b are constants).

[0109] (5) By inputting the predicted value of the system demand for the target frame on the prediction date into X in the equation of the calculated approximate straight line, the value of Y calculated is the predicted value of the electricity spot market price for the target frame on the prediction date.

[0110] Finally, the display unit 37 displays the predicted value of the electricity spot market price for each target frame on the prediction date predicted by the price prediction unit 36. An example of the display of the predicted value of the electricity spot market price displayed in this way is shown in FIG. 18. Referring to FIG. 18, it can be seen that the predicted price for each target frame on August 1, 2023, which is the prediction date, is displayed.

[0111] As described above, in the method for predicting the electricity spot market price by the electricity market price prediction device 10 of the present embodiment, it is specified which power generation equipment of which power source type the past actual value of the electricity spot market price was determined based on the marginal cost, and a merit order curve for the prediction date is created based on the specified information.

[0112] Generally explained, the merit order curve is, as shown in FIG. 3, a stepped broken line graph in which each power plant is arranged in ascending order of marginal cost. Therefore, within the same power source type, as the electricity demand increases, the marginal cost should increase because which power generation equipment is operated up to changes. Furthermore, even with the same electricity demand, the electricity spot market price fluctuates depending on the renewable energy generation amount and the capacity (generable amount) of other power sources.

[0113] In this way, on the merit order curve, even when intersecting with a power source type with system demand, the electricity spot market price changes depending on which power generation equipment included in that power source type it intersects with. That is, it is considered that by specifying which power generation equipment satisfies the system demand when the generated amounts of a plurality of power generation equipment with different marginal costs are accumulated, the electricity spot market price can be predicted with higher accuracy.

[0114] Therefore, by creating a merit order curve using the intersection power source type and the intersection capacitance as in this embodiment, and predicting the electricity spot market price based on the created merit order curve and the predicted value of the grid demand on the prediction target date, compared with the conventional prediction method that predicts the electricity market price on the prediction target date from the relationship between the electricity demand and supply on the prediction target date, it becomes possible to predict the spot market price on the prediction target date with high accuracy.

Explanation of Signs

[0115] 10 Power Market Price Prediction Device 11 CPU 12 Memory 13 Storage Device 14 Communication Interface 15 User Interface Device 16 Control Bus 20 Server Device 30 Internet 31 Information Acquisition Unit 32 Intersection Power Source Type Identification and Intersection Capacitance Calculation Unit 33 Electricity Supply Quantity Prediction Unit 34 MOC Creation Unit 35 Electricity Demand Quantity Prediction Unit 36 Price Prediction Unit 37 Display Unit

Claims

1. Using a plurality of days within a preset past period where the electricity spot market price is known as comparison reference days, and using the actual values of the electricity supply volume for each power source type and the actual value of the grid demand for each period of the comparison reference days, for each actual value of the electricity spot market price of the comparison reference days, identify the intersection power source type, which is the power source type to which the power generation facilities expected to be involved in determining the electricity spot market price belong. When arranging the power generation facilities included in the identified intersection power source type in ascending order of marginal cost, calculate the intersection capacity, which is the cumulative value of the power generation volume from the power generation facility with the lowest marginal cost to the power generation facilities expected to be involved in determining the electricity spot market price. An intersection power source type identification and intersection capacity calculation unit; A supply volume prediction unit that predicts the electricity supply volume for each power source type on the prediction target day; Based on the intersection power source type identified by the intersection power source type identification and intersection capacity calculation unit and the calculated intersection capacity, plot the actual values of the electricity spot market price for each period of the comparison reference days on a graph using the predicted values of the electricity supply volume for each power source type on the prediction target day predicted by the supply volume prediction unit, thereby creating a merit order curve on the prediction target day. A creation unit; A demand volume prediction unit that predicts the electricity demand volume for each prediction target period on the prediction target day; A price prediction unit that predicts the electricity spot market price for each prediction target period on the prediction target day by finding the intersection of the electricity demand volume predicted by the demand volume prediction unit and the merit order curve created by the creation unit; An electricity market price prediction device comprising the above.

2. The creation unit creates different merit order curves in any time division with better prediction accuracy. The electricity market price prediction device according to Claim 1.

3. When the month to which the comparison reference day belongs is different from the month to which the prediction target day belongs, the creation unit creates a merit order curve by performing fuel price correction on the actual values of the electricity spot market price for each period of the comparison reference day and then plotting. The electricity market price prediction device according to Claim 1.

4. When there are a plurality of plots on the merit order curve whose difference from the predicted power demand is equal to or less than a preset power capacity, the price prediction unit weights the actual values of the power spot market prices of the plurality of plots such that the weighting coefficient increases as the number of days difference from the prediction target date decreases, and calculates a weighted average, thereby predicting the power spot market price of the prediction target period on the prediction target date. The power market price prediction device according to claim 1.

5. When there are no plurality of plots on the merit order curve whose difference from the predicted power demand is equal to or less than a preset power capacity, the price prediction unit performs cluster classification on the plurality of plots constituting the merit order curve, and selects, from among the plurality of classified clusters, the cluster with the smallest difference between the center of gravity position of each cluster and the predicted power demand. Then, an approximate straight line with the shortest distance to the plurality of plots belonging to the selected cluster is calculated, and the predicted power demand is input to the calculated approximate straight line, thereby predicting the power spot market price of the prediction target period on the prediction target date. The power market price prediction device according to claim 1.

6. Using a plurality of days within a preset past period in which the power spot market price is known as comparison reference dates, using the actual values of the power supply amounts for each power source type and the actual values of the system demand for each period of these comparison reference dates, for each actual value of the power spot market price of the comparison reference date, identifying the intersection power source type, which is the power source type to which the power generation facilities expected to be involved in determining the power spot market price belong, and calculating the intersection capacity, which is the cumulative value of the power generation amounts from the power generation facility with the lowest marginal cost to the power generation facilities expected to be involved in determining the power spot market price when the power generation facilities included in the identified intersection power source type are arranged in ascending order of marginal cost; an intersection power source type identification and intersection capacity calculation step; A supply amount prediction step of predicting the power supply amount for each power source type on the prediction target date; A creating step of creating a merit order curve for the prediction target date by plotting the actual values of the power spot market prices for each frame of the comparison reference date on a graph using the predicted values of the power supply amounts for each power source type on the prediction target date predicted in the supply amount prediction step, based on the intersection power source type specified by the intersection power source type specifying and intersection capacity calculating unit and the calculated intersection capacity; A demand amount prediction step of predicting the power demand amounts for the prediction target frames on the prediction target date respectively; A price prediction step of predicting the power spot market prices for the prediction target frames on the prediction target date respectively by obtaining the intersections between the power demand amounts predicted in the demand amount prediction step and the merit order curve created in the creating step; A power market price prediction method comprising the above.

7. An intersection power source type specifying and intersection capacity calculating step of specifying, as the intersection power source type, the power source type to which the power generation facilities expected to be involved in the determination of the power spot market price belong, for each actual value of the power spot market price on the comparison reference date, using the actual values of the power supply amounts for each power source type and the actual values of the system demand for each frame of the comparison reference date within a preset past period in which the power spot market price is known, and calculating the intersection capacity, which is the cumulative value of the power generation amounts from the power generation facility with the lowest marginal cost to the power generation facilities expected to be involved in the determination of the power spot market price when the power generation facilities included in the specified intersection power source type are arranged in ascending order of marginal cost; A supply amount prediction step of predicting the power supply amounts for each power source type on the prediction target date; A creating step of creating a merit order curve for the prediction target date by plotting the actual values of the power spot market prices for each frame of the comparison reference date on a graph using the predicted values of the power supply amounts for each power source type on the prediction target date predicted in the supply amount prediction step, based on the intersection power source type specified by the intersection power source type specifying and intersection capacity calculating unit and the calculated intersection capacity; A demand amount prediction step of predicting the power demand amounts for the prediction target frames on the prediction target date respectively; A price prediction step of predicting the electricity spot market prices of the prediction target periods on the prediction target date by obtaining the intersection points between the predicted electricity demand volume in the demand volume prediction step and the merit order curve created in the creation step; A program for causing a computer to execute the same.

Images