JP6507734B2 - Power trading volume determination system, power trading volume determining method and program - Google Patents

Description

  The present invention relates to a power trading volume determination system, a power trading volume determination method, and a program.

  In recent years, in the power market, it has become possible to freely buy and sell power. When buying and selling the power, the trading price of the power is determined for a certain future period to buy and sell the power (for example, buying and selling every 30 minutes two days ago). For example, Patent Document 1 proposes a method of predicting the amount of power demand on a transaction target date (for example, two days later) so that the power supplier can perform risk management in the power transaction.

JP, 2004-252967, A

  However, although data on the amount of power demand in the past has been accumulated, and it has become possible to forecast the amount of power demand at plant facilities etc. accurately to some extent in advance, consideration of the risk due to demand fluctuation has been made. In addition, it was not possible to determine the optimal power transaction volume according to risk and return.

  The present invention has been made in view of such background, and it is an object of the present invention to be able to determine the amount of traded power according to risk and return.

  The main invention of the present invention for solving the above problems is a system for determining the sales volume for each time zone to be sold in the power transaction, and acquiring prediction error acquiring the prediction error of the past air temperature for each time zone A correlation value calculation unit that calculates a correlation value representing the correlation of the prediction error for each of a plurality of pairs and a plurality of pairs of time zones, and a sum of profits from the power transaction for each time zone according to the correlation value And a power transaction amount determination unit that determines the transaction amount for each time period so that a value obtained by subtracting the risk is maximized.

  Further, in the power transaction amount determination system of the present invention, the transaction amount determination unit determines a second power demand amount larger than the first power demand amount from a first profit corresponding to the first power demand amount. The risk may be calculated in accordance with the difference obtained by subtracting the second profit corresponding to and the correlation value.

  Further, in the power transaction amount determination system of the present invention, the transaction amount determination unit is a first vector in which the difference obtained by subtracting the second income from the first income is arranged in one row for each time period. And calculating a root of a value obtained by multiplying a correlation matrix, which is a matrix of the correlation values corresponding to the pair of time zones, and a second vector obtained by transposing the first vector as the risk It is also good.

  Further, in the power transaction amount determination system according to the present invention, the power demand amount is given as a probability distribution, and the transaction amount decision unit sets the 50th percentile value of the probability distribution as the first power demand amount. The first and second revenues may be calculated using a percentile value larger than 50 as the second power demand amount.

  Further, in the power transaction amount determination system of the present invention, the power generation cost is obtained by a function corresponding to the power generation amount, and the transaction amount determination unit determines each of the first and second power demand amounts obtained by the function. A second power demand corresponding to the first or second power demand from a first power generation cost according to a first power generation amount obtained by adding the transaction amount to the first or second power demand The increased cost minus the generation cost may be subtracted from the sales price multiplied by the transaction volume to calculate the first and second revenues.

  Further, another aspect of the present invention is a method of determining the amount of power traded for each time zone to be sold in a power transaction, wherein the computer acquires a prediction error of past air temperature for each time slot, Calculating a correlation value representing the correlation of the prediction error for each of the plurality of pairs of time zones, and a value obtained by subtracting a risk according to the correlation value from a sum of profits from the power transaction for each time zone Determining the trading volume for each time zone to be maximum.

  Moreover, the other aspect of this invention is a program for determining the power transaction amount for every time slot | zone sold in power transaction, Comprising: The computer acquires the prediction error of the past air temperature for every said time slot | zone And calculating a correlation value representing the correlation of the prediction error for each of the plurality of pairs of time zones, and subtracting the risk according to the correlation value from the sum of revenues from the power transaction for each time zone Determining the transaction volume for each of the time zones so that the value is maximized.

  Other problems disclosed in the present application and their solutions will be made clear by the section of the embodiments of the present invention and the drawings.

  According to the present invention, the amount of power traded can be determined according to the risk and the return.

It is a figure for demonstrating optimization of the transaction volume for every time slot | zone. FIG. 2 is a diagram showing an example of a hardware configuration of a transaction amount determination device 10. FIG. 2 is a diagram showing an example of the software configuration of the transaction amount determination device 10. It is a figure which shows the structural example of the air temperature performance value memory | storage part 131. FIG. It is a figure which shows the structural example of the air temperature estimated value memory | storage part 132. FIG. It is a figure which shows the structural example of the reference | standard curve memory | storage part 133. As shown in FIG. FIG. 6 is a diagram showing an example of the configuration of a weight storage unit 134. FIG. 16 is a diagram showing an example of configuration of a correlation matrix storage unit 135. FIG. 6 is a diagram showing an example of a configuration of a power price storage unit 136. It is a figure which shows the structural example of the transaction volume memory | storage part 137. As shown in FIG. It is a figure which shows the flow of pre-processing. It is a figure which shows the flow of preparation processing of a reference | standard curve. It is a figure which shows the flow of past temperature prediction processing. It is a figure which shows the flow of creation processing of a correlation matrix. It is a figure which shows the flow of the process which determines the transaction volume for every time slot | zone. It is a figure explaining the predicted value of the time slot | zone of a unit of 30 minutes.

== Overview = =
The transaction volume determination device 10 according to an embodiment of the present invention is a time zone (in the present embodiment, a time zone so as to maximize the revenue from power sales transactions performed by a power supplier such as a power company in a power transaction market). Is assumed to be in units of one hour) to determine the power transaction volume (hereinafter referred to as w).

  Revenue from electricity sales transaction is the amount obtained by multiplying the transaction volume w by the expected cash flow of the electricity sales transaction (hereinafter referred to as return and expressed as R. Expected cash flow is the revenue per expected unit transaction volume) , Reduced power generation costs for power sales transactions. The power sold in the power exchange market is generated by using the surplus power generation capacity after generating the power demand amount. Generally, there are a plurality of generators, and the power generation cost limit cost (increase per unit power amount) is operated in order from the smallest one, so it is assumed that the limit cost also increases as the power generation amount increases. Since the power transaction accommodates power at a future time, the generation cost is calculated based on the predicted value of the power demand. Therefore, if the actual amount of power demand is higher than the forecast value, the marginal cost will be higher than expected and the return will drop.

Therefore, in the present embodiment, the power transaction amount for each time zone is determined in consideration of such a risk of upward fluctuation of demand (that is, downward fluctuation of return). Specifically, the forecast of the power demand is performed based on the forecast of the air temperature, and it is assumed that the demand for the power also increases when the air temperature becomes higher than the forecast value, and the forecast error of the air temperature is a risk ( Hereinafter, it is expressed as σ _P. ).

  Furthermore, when predicting the temperature, for example, when the prediction of the morning temperature deviates, the prediction of the daytime temperature also tends to deviate, but the prediction about the temperature of the night does not deviate so much; There is a correlation. Therefore, it is possible to suppress the profit decrease due to the prediction error if the power transaction is performed in the time zone where the correlation of the prediction error is low, rather than performing many power transactions in the time zone where the correlation of the prediction error of the temperature is high. Become. So, in this embodiment, it considers also about correlation of a prediction error of air temperature for every time zone. For example, if the temperature in the evening (the first time zone) is higher than the predicted value (the prediction error is a positive value), it is also predicted in another time zone in the evening (the second time zone) While the temperature is often higher than the value (the prediction error is a positive value), the temperature in the early morning (the third time zone) is often lower than the predicted value (the prediction error Is a negative value), if the transaction volume in the first time zone is the same, the transaction volume in the third time zone is made to be larger than that in the second time zone.

In the present embodiment, the risk σ _P related to the power transaction is a value (correlation risk) which is determined to be higher as there are more transactions in the time zone where the correlation of the prediction error of the air temperature is high. It shall be decided according to the corresponding amount of decrease in revenue (Risk fluctuation risk), and maximize the revenue (R R _i × w _i ) with a penalty of risk (σ _P ).

Furthermore, a risk aversion degree (hereinafter referred to as λ) is set, and the risk is weighted using this. Risk aversion λ is grasped utility function according to the return as those set with U = R-λσ _P. Therefore, U is changed with respect to a straight line of R = λσ _P + U in which U = R−λσ _P is transformed into an efficient frontier, and a portfolio at a point where the straight line is in contact with the efficient frontier is obtained.

Specifically, the transaction amount wi for each time zone _i is determined so that the following equation (1) is maximized.

As shown in FIG. 1, using portfolio theory, power trading for each time zone constitutes one product, and the portfolio is selected as the optimum trading volume at the point where the line of λ inclines on the efficient frontier. It is synonymous with to do. The optimal point on the efficient frontier changes according to the size of the risk aversion degree λ.

  The details will be described below.

== Hardware configuration ==
FIG. 2 is a diagram showing an example of the hardware configuration of the transaction amount determination device 10 according to the present embodiment. The transaction amount determination device includes a CPU 101, a memory 102, a storage device 103, a communication interface 104, an input device 105, and an output device 106. The storage device 103 stores various data and programs, and is, for example, a hard disk drive, a solid state drive, or a flash memory. The communication interface 104 is an interface for connecting to a communication network, for example, an adapter for connecting to Ethernet (registered trademark), a modem for connecting to a public telephone network, a wireless communication device for performing wireless communication, It is a USB (Universal Serial Bus) connector or RS232C connector for serial communication. The transaction volume determination apparatus 10 accesses, for example, the computer of the Japan Meteorological Agency or a meteorological company via the communication interface 104 to receive weather forecast data, or accesses a computer that is forecasting the demand volume to forecast the demand volume You can get it. The input device 105 is, for example, a keyboard, a mouse, a touch panel, a button, a microphone, and the like for inputting data. The output device 106 is, for example, a display, a printer, or a speaker that outputs data.

== Software Configuration ==
FIG. 3 is a view showing an example of the software configuration of the transaction amount determination device 10 according to the present embodiment. The transaction amount determination device 10 includes an actual value acquisition unit 111, a predicted value acquisition unit 112, a reference curve generation unit 113, an air temperature prediction unit 114, a correlation matrix generation unit 115, a total transaction amount acquisition unit 116, a transaction amount determination unit 117, and a transaction. The volume output unit 118, the air temperature actual value storage unit 131, the air temperature predicted value storage unit 132, the reference curve storage unit 133, the weight storage unit 134, the correlation matrix storage unit 135, the power price storage unit 136, and the transaction amount storage unit 137.

  The actual value acquisition unit 111, the predicted value acquisition unit 112, the reference curve creation unit 113, the air temperature prediction unit 114, the correlation matrix creation unit 115, the total transaction amount acquisition unit 116, the transaction amount determination unit 117, and the transaction amount output unit 118 This is realized by the CPU 101 included in the transaction amount determination apparatus 10 reading out and executing the program stored in the storage device 103 to the memory 102, and the air temperature actual value storage unit 131, the air temperature predicted value storage unit 132, and the reference curve storage unit The weight storage unit 134, the correlation matrix storage unit 135, the power price storage unit 136, and the transaction amount storage unit 137 are realized as part of the memory area provided by the transaction amount determination apparatus 10 and the storage area provided by the storage device 103. It is possible to

  The actual value acquisition unit 111 acquires the actual value of the temperature. In the present embodiment, the actual temperature value is obtained by measuring the temperature of the outside air with a thermometer every predetermined time (one hour in the present embodiment, but may be any time), that is, time It will be the actual value for each band. As described above, in the present embodiment, the time zone is in units of one hour, and is divided into 24 segments from 1 o'clock to 24 o'clock. The actual value acquisition unit 111 can receive a measurement result automatically transmitted from, for example, a measurement device of air temperature, or can access the measurement device to acquire a measurement result. The actual temperature value may be, for example, a value provided by the Japan Meteorological Agency or a meteorological company. In this case, the actual temperature value can be obtained by accessing the computer of the meteorological agency or the meteorological company. The actual value acquisition unit 111 acquires actual temperature values for a plurality of measurement locations. As a measurement place, an area name to which a weather forecast is provided, that is, one of “Hiroshima”, “Okayama”, “Matsue”, “Tottori”, and “Yamaguchi” is set. The actual value acquisition unit 111 registers the acquired actual value of the temperature in the actual temperature value storage unit 131.

  The air temperature actual value storage unit 131 stores the air temperature actual value. FIG. 4 is a diagram showing an example of the configuration of the air temperature result value storage unit 131. As shown in FIG. The actual temperature value storage unit 131 stores the actual value for each measurement location in association with the measurement date (day), the time zone, and the season information. Of course, the measurement location can be set arbitrarily, but it should be the same as (or able to coincide with) the point at which the predicted values of the maximum temperature and the minimum temperature are provided. The season information represents season segments divided arbitrarily, and in the present embodiment, it is determined according to a predetermined rule according to the date and time, and the season information is "winter", "early spring", "spring", "early summer", "Saiyu" "Summer" "Autumn" shall be divided into seven divisions. The seasonal information may be set, for example, every other year for "Suiyu". It is assumed that actual temperature values for the past ten years are stored in advance in the temperature actual temperature value storage unit 131. Note that a measurement result may be acquired from the measurement device, or an actual value registration unit may be provided which receives an input of the measurement result from the operator and registers the result in the air temperature actual value storage unit 131. Also, less than or more than 10 years worth of actual value may be registered.

  The predicted value acquisition unit 112 acquires predicted values of the maximum temperature and the minimum temperature of the prediction target day. The predicted value acquisition unit 112 can, for example, access a computer of the Meteorological Agency or a meteorological company to receive data (meteorological data) including a predicted value announced by the Meteorological Agency or the meteorological company. The predicted value acquisition unit 112 may receive input of predicted values of the maximum temperature and the minimum temperature from the operator.

  In the present embodiment, it is premised that the temperature forecasting after two days is performed to forecast the power demand after two days, and the forecasted value acquiring unit 112 calculates the maximum temperature and the minimum temperature two days after the two days before the forecast target day. Assume that a predicted value is to be obtained. Further, the predicted value acquisition unit 112 acquires predicted values of the maximum temperature and the minimum temperature for each area to be predicted. That is, the predicted value acquisition unit 112 acquires predicted values of the maximum temperature and the minimum temperature as weather forecasts for “Hiroshima”, “Okayama”, “Matsue”, “Tottori” and “Yamaguchi”. In the present embodiment, it is assumed that weather data including weather forecasted by the Japan Meteorological Agency, the probability of precipitation, and predicted values of the maximum temperature and the minimum temperature are acquired.

  The predicted value acquisition unit 112 periodically (for example, every day at a predetermined time) acquires weather data including predicted values of the maximum temperature and the minimum temperature two days later from the Meteorological Agency or a meteorological company. The predicted value acquisition unit 112 registers the acquired predicted value in the predicted air temperature value storage unit 132.

  The predicted temperature value storage unit 132 stores predicted values of the maximum temperature and the minimum temperature. FIG. 5 is a view showing a configuration example of the predicted air temperature value storage unit 132. As shown in FIG. The predicted temperature value storage unit 132 stores predicted values of the weather, the probability of precipitation, the maximum temperature, and the minimum temperature in association with the name (area name) of the prediction target area and the day of the prediction target. The area name is either "Hiroshima", "Okayama", "Matsue", "Tottori" or "Yamaguchi". In the present embodiment, it is assumed that only one set of the predicted value of the highest temperature and the predicted value of the lowest temperature corresponding to the prediction target day is stored in the predicted temperature value storage unit 132. As described above, the predicted value after two days announced in the past is registered in the predicted temperature value storage unit 132. Further, in the present embodiment, it is assumed that predicted values for the past three years are stored in the predicted temperature value storage unit 132.

  The reference curve creation unit 113 creates a reference curve. Specifically, the reference curve creation unit 113 records, for each date of one year and each region to be predicted, the temperature value of the temperature corresponding to the region of the same month stored in the temperature actual value storage unit 131. , The lowest temperature is "0", and the highest temperature is normalized as "1". More specifically, the reference curve creation unit 113 uses the temperature actual value storage unit 131 for 10 years of the same area on the same day of the same month for each day of the year and each area (the leap year in the past 10 years for February 29 The actual value of the air temperature is read out, the average air temperature value is calculated for each time zone, and the lowest average value (hereinafter referred to as the average lowest air temperature) and the average highest air temperature (hereinafter referred to as the average air temperature) The highest average temperature is selected, and the average minimum temperature is subtracted from the average value of the actual values, and a value obtained by subtracting the average minimum temperature from the average maximum temperature is calculated as a normalized value. It is a standard curve that arranged the normalization value for every time zone in order of time zone. The reference curve creation unit 113 registers the created reference curve in the reference curve storage unit 133.

  The reference curve storage unit 133 stores a reference curve for each prediction target area. FIG. 6 is a view showing a configuration example of the reference curve storage unit 133. As shown in FIG. The reference curve storage unit 133 associates the temperature of each time zone with a numerical value between “0” and “1” for each date of one year in association with the name (area name) of the prediction target area (FIG. 6 In the example, the normalized value (hereinafter referred to as normalized temperature) of the average temperature expressed as a percentage unit from 0% to 100% is stored.

  The weight storage unit 134 stores weights for each area. FIG. 7 is a view showing a configuration example of the weight storage unit 134. As shown in FIG. In the present embodiment, the weight uses the ratio of the power demand in each area. The user may change the weight arbitrarily, or may set the weight regardless of the amount of power demand.

  The temperature prediction unit 114 predicts the temperature using the reference curve. The temperature prediction unit 114 is given the prediction target area and the highest and lowest temperatures, and the temperature prediction unit 114 takes a value obtained by subtracting the given lowest temperature from the given highest temperature as the given prediction target area. The temperature prediction value of the prediction target area is calculated by multiplying the numerical value of the corresponding reference curve and adding the lowest temperature given thereto. Further, the air temperature prediction unit 114 uses the weights stored in the weight storage unit 134 to perform weighted averaging of the air temperature prediction values of the area to be predicted, and calculates the air temperature prediction value of the whole (in this embodiment, the Chugoku region). .

  The correlation matrix creation unit 115 creates a matrix (hereinafter referred to as a correlation matrix) that represents the correlation between prediction errors related to two time zones. The correlation matrix generation unit 115 causes the air temperature prediction unit 114 to calculate the air temperature predicted value of the time zone of the past date for all the time zones, and the air temperature actual value of the time zone of the past date is the air temperature actual value Reading from the storage unit 131, a prediction error that is the difference between the predicted air temperature value and the actual air temperature value is calculated. The correlation matrix generation unit 115 calculates correlations of prediction errors for all pairs of time zones using a general statistical method, and generates a correlation matrix storing the calculation results. The correlation matrix creation unit 115 registers data representing the created correlation matrix in the correlation matrix storage unit 135.

  The correlation matrix generation unit 115 corresponds to the correlation value calculation unit of the present invention. The air temperature prediction unit 114 and the correlation matrix generation unit 115 correspond to the prediction error acquisition unit of the present invention.

  The correlation matrix storage unit 135 stores the correlation matrix. FIG. 8 is a diagram showing an example of the correlation matrix stored in the correlation matrix storage unit 135. As shown in FIG. The correlation matrix in this embodiment is a 24 × 24 matrix for 24 time zones.

  The power price storage unit 136 stores the sales price per unit power in the power exchange market for each time zone. FIG. 9 is a diagram showing a configuration example of the power price storage unit 136. As shown in FIG. As shown in the figure, the power price storage unit 136 stores the power price in association with the time zone.

The total transaction volume acquisition unit 116 acquires the sales volume (hereinafter referred to as the total transaction volume) of power in the power exchange market in a predetermined period (one day in the present embodiment). The total transaction amount acquisition unit 116 can receive input of the total transaction amount from the user, for example. Further, for example, the total transaction volume acquisition unit 116 may determine the total transaction volume so as to maximize the transaction balance due to the power transaction. It is assumed that the predicted value of the amount of power demand for each time zone is given as a probability distribution (assuming a normal distribution in this embodiment), and its 50th percentile value (hereinafter referred to as X percentile value of amount of power demand X% significant amount that. the total daily) and ^{N 50%,} accounts, the ^{Y 50%} when the amount of power demand was the amount of 50% demand, the total transaction amount _{w ttl,} the average daily electricity price Assuming that P, the amount of power generation is W, and the power generation cost such as fuel for the generator can be obtained by the function F (W), determine the total transaction amount w _ttl that Y ^50% maximizes by the following equation (2) Can.

The transaction volume determination unit 117 determines the power transaction volume for each time zone. In the present embodiment, the transaction amount determination unit 117 assumes that the total transaction amount per day is the value w _ttl acquired by the total transaction amount acquisition unit 116, and maximizes equation (1), each time zone Determine the amount of electricity traded. That is, the transaction amount determination unit 117 performs a process of allocating the total transaction amount w _ttl to each time zone.

Here risk sigma _P is a return when the _{X 1%} demand _R ⁱ X1%, a return when _{X 2%} demand as ^X2% _R ^i, is defined by the following equation (3). However, it is assumed that X ₁ <X ₂ . In this embodiment, assuming X ₁ = 50 and X ₂ = 90, from the 50% demand amount which is the expected value in the demand amount of normal distribution, the risk is calculated when the 90% demand amount is assumed as the maximum blur width. X ₁ and X ₂ are risk assessments, for example, taking X ₁ = 30 into consideration, taking into account downturn in demand, X ₂ = 70, etc. It can be set arbitrarily.

^{_{Incidentally, [w i (R i X1}} % -R i X2%)] , the amount of trading blur width when demand _{X 2%} demand from _{X 1%} demand calculated for each time period i has above shake It is a 24x1 matrix (vector) which makes a value which multiplied by.

The return _Ri ^x% at the time of X% demand is represented by the following formula (4), where the transaction balance at the time of X% demand is Y ^x% .

The transaction amount determination unit 117 maximizes the following expression (1-1) so that the sum of w _i becomes w _ttl , using expressions (2), (3), and (4) as X = 50. Determine w _i (optimal value of w _i ) (optimal allocation).

Further transaction amount determination unit 117, the optimum allocation of time of demand on blur, when the X> 50 (in this embodiment X = 70 and X = 90), also calculates the optimum value of w _i for. The transaction volume determination unit 117 registers the transaction volume w _i for each determined time zone in the transaction volume storage unit 137.

  The transaction amount storage unit 137 stores the optimum value of the transaction amount for each time period determined by the transaction amount determination unit 117. FIG. 10 is a diagram showing a configuration example of the transaction amount storage unit 137. As shown in FIG. The transaction amount storage unit 137 stores the optimal transaction amount of the time zone at the time of the power demand amount in association with the percentile of the power demand amount and the time period.

  The transaction volume output unit 118 outputs the optimal transaction volume. The transaction amount output unit 118 outputs the transaction amount stored in the transaction amount storage unit 137. The transaction volume output unit 118 may output a transaction volume corresponding to a plurality of power demand volumes in a table format as shown in FIG. 10, or may output in a graph format. Alternatively, the specification of the percentile of the power demand may be received, and only the transaction amount corresponding to the predicted value of the power demand of the designated percentile may be output.

== processing ==
Hereinafter, the process performed by the transaction amount determination apparatus 10 of this embodiment is demonstrated.

<Pre-processing>
FIG. 11 is a diagram showing a flow of pre-processing performed before the processing of determining the transaction amount.

  The actual value acquisition unit 111 acquires the actual value of the air temperature and registers it in the air temperature actual value storage unit 131 (S201), and the predicted value acquisition unit 112 acquires the weather data and registers it in the air temperature predicted value storage unit 132 ( S202). The reference curve creation unit 113 creates a reference curve from the actual temperature value (S203).

  FIG. 12 is a diagram showing a flow of processing for creating a reference curve. The reference curve creation unit 113 first reads out the actual temperature value of the same region on the same month and the same day from the actual temperature value storage unit 131 for each region and each day of the year (including February 29) (S221) The actual temperature values are averaged for each time zone to calculate an average temperature for each time zone (S222). The reference curve creation unit 113 selects an average maximum temperature and an average minimum temperature from among the average temperatures (S223). The reference curve creation unit 113 subtracts the average lowest temperature from each average temperature (S224), and divides the subtracted average temperature by the value obtained by subtracting the average lowest temperature from the average highest temperature to calculate a normalized value (S225). The reference curve creation unit 113 registers, in the reference curve storage unit 133, a reference curve in which normalized values are arranged in order of time zone (S226).

  Returning to FIG. 11, the temperature prediction unit 114 performs past temperature prediction for each region and date (S204).

  FIG. 13 is a diagram showing a flow of past air temperature prediction processing.

  The air temperature prediction unit 114 performs the following processing for all the prediction target days stored in the air temperature prediction value storage unit 132, that is, each of the prediction target days for the past three years.

  The temperature prediction unit 114 predicts the predicted values of the maximum temperature and the minimum temperature corresponding to the region and the prediction target date for each region of “Hiroshima” “Okayama” “Matsue” “Tottori” “Yamaguchi” temperature predicted value storage unit 132 The reference curve corresponding to the area and date is read out from the reference curve storage unit 133 (S242). The air temperature prediction unit 114 multiplies the normalized value of each time zone included in the read reference curve by a value obtained by subtracting the predicted value of the lowest air temperature read out from the read out predicted value of the highest air temperature, By adding the predicted value of the lowest temperature, the predicted value of the temperature for each time zone is calculated (S243).

  Referring back to FIG. 11, the correlation matrix generation unit 115 performs a process of generating a correlation matrix (S205).

  FIG. 14 is a diagram showing a flow of creation processing of a correlation matrix.

  The correlation matrix generation unit 115 calculates prediction target dates for all prediction target dates stored in the predicted temperature value storage unit 132, that is, for each of the prediction target dates for the past three years, each region, and each time zone, The actual temperature value and the seasonal information corresponding to the area and the time zone are read out from the actual temperature value storage unit 131 (S261), and the read actual temperature value of the temperature and the time zone calculated by the temperature prediction unit 114 The difference (prediction error) between the area and the predicted value of the temperature corresponding to the target date is calculated (S262).

  After performing the above processing, the correlation matrix creating unit 115 performs the above-described prediction error for the first time zone and the second time for each of the two time zone (first time zone and second time zone) pairs. A correlation coefficient with the above prediction error for the band is calculated (S263). A general statistical processing method is used to calculate the correlation coefficient. The correlation matrix creation unit 115 registers the calculated correlation coefficient in the correlation matrix storage unit 135 in association with the first time zone and the second time zone (S 264).

  As described above, the reference curve for each region is created and registered in the reference curve storage unit 133, and the correlation coefficient indicating the correlation of prediction errors in two time zones is used as the correlation matrix in the correlation matrix storage unit 135. be registered.

<Trade volume determination processing>
FIG. 15 is a diagram showing the flow of processing for determining the transaction amount for each time zone.

The total transaction amount acquisition unit 116 acquires a total transaction amount w _ttl (S301). The transaction amount determination unit 117 receives an input of the risk avoidance degree λ (S302). The transaction amount determination unit 117 determines the transaction amount w _i maximizing Equation (1-1) such that the sum of w _i becomes W _ttl , where X is 50, 70, and 90, respectively (S 303), in association with the X, it registers the determined trading volume _{w i} in the transaction amount storage unit 137 (S304). After performing the above processing for each percentile of 50, 70, and 90, the transaction amount output unit 118 outputs the transaction amount registered in the transaction amount storage unit 137 (S305). For example, the transaction volume output unit 118 can output a transaction volume for each of the optimal time zones for 50% demand, 70% demand and 90% demand in the form of a table as shown in FIG.

== effect ==
As explained above, according to the transaction volume determination device 10 of the present embodiment, not only the return R but also the risk σ _P is taken into consideration to maximize the profit (ΣY _i ^X% = R _i × w _i ) An optimal combination of trading volumes can be determined. Therefore, it is possible to use the portfolio theory to determine the transaction volume for each time zone using a combination of transactions in each time zone as a portfolio, and if efficient transaction volumes on the frontier are selected as the optimal combination of transaction volumes Since it is good, it can determine the amount of power transactions efficiently and effectively. Therefore, the expected value of the profit concerning the power transaction can be improved.

Also, the transaction amount determination device 10 of the present embodiment, the transaction from the return R _i ^50% at 50% demand, to a value demand minus the return R _i ^90% at 90% demand assuming a case where the above blurring A vector in which the amount of money multiplied by the quantity w _i (R _i ^50% −R _i ^90% ) × w _i is a row for each time zone, the correlation matrix stored in the correlation matrix storage unit 135, and the transposed vector multiplying the column vector obtained by, it is calculated this root as a risk (50% deviation between the time and the time of 90% demand demand) sigma _P.

The above amount (R _i ^50% -R _i ^90% ) × w _i is ^90% of the electricity demand (from 50% demand) because the return is expected to decrease if the electricity demand is upset. The amount of revenue loss due to the upside-down) can be assessed as a risk. While the optimization process is performed to maximize the revenue when determining the transaction volume, according to the transaction volume determination device 10 of the present embodiment, the risk of the profit decline due to the upswing of the power demand is incorporated and the optimization is performed It can be carried out.

  Furthermore, in the above calculation of risk, a correlation matrix is used to incorporate the amount of loss into risk. Therefore, when the correlation of the prediction error is larger in the time zone where the correlation is large, the risk can be also greatly evaluated. Therefore, it is possible to determine the optimal trading volume so as to avoid the risk of conducting power trading by distributing to the time zone in which the error of the temperature prediction (that is, the error of the demand prediction) similarly occurs.

  Further, according to the transaction amount determination apparatus 10 of the present embodiment, it is possible to weight the risk in response to the input of the risk avoidance degree λ. Thus, it is possible to select one of the combinations on the efficient frontier according to the degree of risk avoidance.

== Modifications ==
In the present embodiment, the transaction amount determination device 10 is a single computer, but the present invention is not limited to this and may be realized by a plurality of computers. In this case, one virtual computer can be realized by a plurality of computers, and the transaction amount determination device 10 can be mounted on this. Further, the functional unit and storage unit of the transaction amount determination device 10 may be divided into a plurality of computers and provided. For example, each storage unit can be implemented in a database server so that the transaction amount determination apparatus 10 can access the database server. In addition, for example, a functional unit that performs prediction of air temperature (reference curve creating unit 113, air temperature predicting unit 114), a functional unit that performs analysis processing in advance (correlation matrix creating unit 115), and a process of determining future transaction volume The functional units (total transaction volume acquisition unit 116, transaction volume determination unit 117) may be implemented on separate computers.

  In the present embodiment, although the input from the user is directly input to the transaction amount determination apparatus 10, the present invention is not limited to this. For example, a user device operated by the user and the transaction amount determination apparatus 10 are connected by a communication network Alternatively, the transaction amount determination device 10 may be accessed from the user device via a communication network.

Further, in the present embodiment, as shown in the equation (3), the risk σ _P is evaluated based on the 50% demand amount, and 90% as the power demand amount deviates by more than the 50% demand amount. Although the demand amount is used, the standard may be not the 50% demand amount but an arbitrary percentile value of the predicted value of the power demand amount. Further, the demand amount in the case of upward blurring may also be an arbitrary percentile value instead of the 90% demand amount as long as the value is higher than the reference percentile value.

  Further, in the present embodiment, the predicted value of the power demand amount is assumed to be a normal distribution, but the present invention is not limited to this and may be any distribution. Also, the predicted value of the power demand may be given as a scalar value. In this case, assuming that the forecasted value of the power demand amount given is the 50% demand amount, 70% demand amount and 90% demand amount may be calculated by subtracting an arbitrary value from the forecasted value of the power demand amount .

Further, in the present embodiment, the utility function is a linear expression, but may be a quadratic expression. For example, with U = R−λσ _P ² as a utility function, a portfolio of power trading volume at the point where the curve of this quadratic function is in contact with an efficient frontier may be determined as an optimal combination of trading volumes.

  Further, in the present embodiment, assuming that the power demand is predicted two days later, the temperature prediction two days later is performed. However, the present invention is not limited to this, and after an arbitrary time such as one day, one hour, etc. Power demand forecasting may be performed. In this case, the reference curve may be created and the predicted temperature value may be calculated using the forecast value after an arbitrary time such as one day, one hour, etc., instead of the forecast value after two days.

  Further, in the present embodiment, correlations of prediction errors are calculated for all pairs in a time zone, but not all pairs, for example, limited to a time zone from 9 o'clock to 17 o'clock, etc. The correlation matrix may be created only for the pair. In this case, the same time zone as a row or a column is used for the vectors multiplied by the correlation matrix.

Further, in the present embodiment, although the time zone is one hour, it may be, for example, 30 minutes of arbitrary length. In this case, as shown in FIG. 16, it is assumed that the predicted value of the air temperature in each time zone is the predicted value of the intermediate time of each time zone, and this is divided by the length of the time zone to any length. The predicted value of the time zone of can be obtained. For example, assuming that the predicted value of air temperature in a certain time zone X in one hour unit is T _X , the predicted temperature T _XS of air temperature at the start time Xs of time zone X and the air temperature forecast at the start time (X + 1) S of time zone X + 1 The value T _{(X + 1) S} is obtained by the following equation.

Then, assuming that the time zone of the first half 30 minutes constituting the time zone X is XA and the time zone of the second half 30 minutes is XB, the predicted air temperature values T _XA and T _XB of the time zones XA and XB are determined by the following equations.

The correlation matrix may also be created as a 48 × 48 matrix in accordance with pairs of time zones every 30 minutes.

  Moreover, in this embodiment, it is set as Formula (1) objective function. That is, the central value of cash flow

The objective function is expressed by equation (5) when

And Here, the average value of cash flow

Then, the variance σ _P ² can be transformed as follows.

Furthermore, the equation can be modified as in the following equation (6).

Here, the variance (σ _p ) is the cash flow variation as a risk measure. The variance is based on the assumption that the distribution is symmetrical (normal distribution). However, the revenue curve is often non-linear. Also, the magnitude of the change in earnings when the temperature of one unit fluctuates often differs between the case of 50% temperature and the case of 70% temperature. Therefore, as corresponding to the standard deviation sigma _i, employing the following equation (7).

The equation (7) is applied to σ _i and σ _j in the equation (6) to make a matrix expression, and the root of this is calculated, which is the equation (3) described above. Note that _ij is expressed as a correlation matrix.

However, as the risk measure σ _P , one other than the above equation (3) may be adopted. For example, the lower part below a certain value in the distribution may be used as the risk measure. In this case, the risk measure σ _P is given by the following equation, and the objective function is given by the following equation (8).

Here, g (R _i ) is a probability density function to be R _i, and in the case of k = 1, it is an absolute value of the deviation which is less than a certain value, and in the case of k = 2, it is the square of the deviation. That is, k is determined by the degree of risk appetite.

  Here, instead of solving equation (8), the following equation (8 ′) may be calculated using m scenarios.

In addition, an area having a certain value or less in the distribution may be used as the risk measure σ _P. In this case, the objective function is

Furthermore, in the present embodiment, σ _P is evaluated as a risk in the equation (5), but the square of σ _P may be used as a risk measure as in the following equation (5 ′).

Also, the following equation may be used as the objective function.

However, in this case, the following equation is used as a constraint condition.

Here CF _E is the target cash flow.

  As mentioned above, although this embodiment was described, the said embodiment is for making an understanding of this invention easy, and is not for limiting and interpreting this invention. The present invention can be modified or improved without departing from the gist thereof, and the present invention also includes the equivalents thereof.

10 Trading Volume Determination Unit 111 Actual Value Acquisition Unit 112 Predicted Value Acquisition Unit 113 Reference Curve Creation Unit 114 Air Temperature Forecasting Unit 115 Correlation Matrix Creation Unit 116 Total Transaction Volume Acquisition Unit 117 Transaction Volume Determination Unit 118 Transaction Volume Output Unit 131 Air Temperature Actual Value Storage Part 132 Temperature predicted value storage part 133 Reference curve storage part 134 Weight storage part 135 Correlation matrix storage part 136 Power price storage part 137 Transaction amount storage part

Claims (7)

What is claimed is: 1. A system for determining the amount of power traded for each time zone sold in power trade, comprising:
A prediction error acquisition unit for acquiring a prediction error of the past temperature for each of the time zones;
A correlation value calculator configured to calculate a correlation value representing the correlation of the prediction error for each of the plurality of pairs of time zones;
A trading volume determining unit that determines the trading volume for each time zone such that a value obtained by subtracting the risk according to the correlation value from the sum of the revenues from the power transactions for each time zone is maximized;
A power transaction amount determination system comprising: The power transaction amount determination system according to claim 1, wherein
The transaction amount determination unit is a difference obtained by subtracting a second revenue corresponding to a second power demand greater than the first power requirement from a first revenue corresponding to a first power demand. Calculating the risk according to the correlation value,
Power trading volume determination system characterized by. The power transaction amount determination system according to claim 2,
The transaction amount determination unit may be configured to associate the first vector obtained by arranging the difference obtained by subtracting the second revenue from the first revenue into one row for each of the time zones, and the correlation corresponding to the pair of the time zones. Calculating a root of a value obtained by multiplying a correlation matrix, which is a matrix of values, and a second vector obtained by transposing the first vector as the risk;
Power trading volume determination system characterized by. The power transaction amount determination system according to claim 2 or 3, wherein
Power demand is given as a probability distribution,
The trade amount determination unit sets the 50th percentile value of the probability distribution as the first power demand amount, and sets the percentile value larger than 50 of the probability distribution as the second power demand amount. To calculate the
Power trading volume determination system characterized by. The power transaction amount determination system according to any one of claims 2 to 4, wherein
The power generation cost is determined by a function corresponding to the amount of power generation,
For each of the first and second power demand amounts determined by the function, the transaction amount determination unit adds a first power generation amount obtained by adding the trading amount to the first or second power demand amount. The additional cost obtained by subtracting the second power generation cost corresponding to the first or second power demand amount from the first power generation cost corresponding to the first power generation cost is subtracted from the revenue amount obtained by multiplying the sales price Calculating 1 and 2nd revenue,
Power trading volume determination system characterized by. A method of determining sales volume for each time zone to be sold in power trading, comprising:
The computer is
Acquiring a prediction error of the past temperature for each of the time zones;
Calculating a correlation value representing the correlation of the prediction error for each of the plurality of pairs of time zones;
Determining the transaction volume for each time zone such that the value obtained by subtracting the risk according to the correlation value from the sum of the revenues from the power transactions for each time zone is maximized;
A method of determining the amount of power traded to perform. A program for determining the amount of power traded for each time zone sold in power trade, comprising:
On the computer
Acquiring a prediction error of the past temperature for each of the time zones;
Calculating a correlation value representing the correlation of the prediction error for each of the plurality of pairs of time zones;
Determining the transaction volume for each time zone such that the value obtained by subtracting the risk according to the correlation value from the sum of the revenues from the power transactions for each time zone is maximized;
A program to run a program.