JP2019046467A - Power transaction support device and power transaction support method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To allocate a bid amount to a market one day before and a market on the day so that the expected value of an electric power procurement fee is minimized according to the prediction accuracy of the previous day and the prediction accuracy of the day. SOLUTION: An electric power transaction support device 1 predicts the previous day forecast amount which is the predicted electric power demand amount at the time of bidding to the market one day before, based on the past electric power demand data, and the previous day forecast amount prediction unit 21 and the past electric power. Based on the demand data, the current day forecast amount prediction unit 22 that predicts the current day forecast amount h, which is the forecast power demand amount at the time of bidding to the market on the day, and the past previous day forecast amount and the corresponding past actual power demand amount. The prediction error of the previous day, which is the prediction error of The configuration is provided. [Selection diagram] Fig. 3

Description

  The present invention relates to a power transaction support device and a power transaction support method for supporting a bid on a wholesale power transaction market.

  At Japan Wholesale Power Exchange (JEPX: Japan Electric Power eXchange), a market one day ago (spot market) and a day market as its adjustment market (time market) are opened as main markets. The market one day before is a market for trading of power to be delivered the next day, and the market on the same day is a market for trading for power to be delivered the day. In the market one day ago and the market on the same day, power is traded every 48 bidding time slots (pieces) that divide the day into 30 minutes.

  The electric power company procures power from two markets, one day ago and the same day market, and supplies the procured power to power suppliers. However, if the amount of electricity procured by the electricity supplier (hereinafter referred to as “procured amount”) falls below the actual demand, the electricity supplier pays a penalty fee, which is a reimbursement according to the shortage. The (imbalance charge) must be paid to the power company (for example, a general transmission and distribution company). The unit price of this penalty charge is generally set higher than the unit price on the market one day ago (whole electricity price) and the unit price on the same day market. On the other hand, if the amount procured by the electric utility exceeds the actual demand, the excess will be provided free of charge to the power company unless the electricity supplier has a storage system. Therefore, in this case, there is a waste due to the free provision.

  As described above, when the procurement amount of the electric power supplier does not match the actual demand amount, a penalty charge and waste occur, so it is required to minimize the mismatch between the procurement amount and the demand amount. In order to do this, it is necessary to forecast the amount of demand at the time of bidding to the market one day ago and the day market, but in general, the forecasting accuracy of the amount of demand at the time of bidding to the day market It tends to be higher than the forecast accuracy of demand at the time of bidding. Also, in general, the unit price of electricity in the current market tends to be higher than the unit price of electricity in the market one day ago. Therefore, the charge required for the electric power company to procure power from the wholesale power transaction market in consideration of the forecasted demand volume at the time of each bidding and the electricity unit price in each market (hereinafter referred to as “power procurement fee” There is a need for a technique to allocate bid volumes to the market one day ago and the same day market so as to minimize the

  This type of technology, for example, is a technology for providing bid support to multiple power markets with different closing times, and in consideration of the difference in prediction accuracy between the previous day and the current day, optimizes the bidding volume to multiple markets. There is known a technology for generating a power sale schedule that can be distributed (see Patent Document 1).

JP, 2016-62191, A

  However, in the prior art described in Patent Document 1 above, the bid amount is optimally allocated to a plurality of markets by considering the difference in prediction accuracy on the previous day and the current day, but the prediction accuracy on the previous day and the current day If at least one of the prediction accuracy is low, of course, it is not possible to allocate bid amounts to the market one day ago and the market today so as to minimize the power supply charge.

  The present invention has been made in view of the problems of the prior art as described above, and according to the prediction accuracy of the previous day and the prediction accuracy of the day, the expected value of the power procurement fee is minimized so as to minimize the market. And the main object of the present invention is to provide a power transaction support device and a power transaction support method that make it possible to allocate a bid amount to the current day market.

  The present invention is a power transaction support apparatus for supporting a bid on a wholesale power transaction market, which predicts a forecasted amount on the previous day which is a forecasted power demand amount at the time of a bid on the market one day ago based on past power demand data. Based on the forecasted amount forecasting unit and the forecasted forecasted amount of electricity required at the time of bidding to the current market based on the forecasted power demand data, the forecasted amount forecasting unit on the current day, and the previous day's forecasted amount in the past and corresponding Prediction for calculating the previous day prediction error, which is a prediction error between the past actual power demand amount and the previous day prediction error, and the corresponding forecast error between the past current day predicted amount and the corresponding past actual power demand amount An error calculation unit corrects the predicted amount on the previous day based on the predicted error on the previous day, and bids the market one day before the market with the bid amount determined based on the predicted amount on the previous day after the correction A market bidding unit that corrects the on-the-day forecast amount based on the on-the-day forecast error and bids the on-the-day market with the bid amount determined based on the on-the-day forecast amount after the correction. And are provided.

  According to the present invention, it is possible to allocate bid amounts to the market one day ago and the market today so as to minimize the expected value of the power procurement fee according to the prediction accuracy of the previous day and the prediction accuracy of the current day.

BRIEF DESCRIPTION OF THE DRAWINGS The schematic diagram of the power transaction system to which the power transaction assistance apparatus which concerns on this invention is applied. Hardware configuration of the power transaction support apparatus according to the present invention Functional block diagram of the power transaction assisting apparatus according to the present invention The figure for demonstrating the process in the same-day market bid part of the power transaction assistance apparatus which concerns on this invention Flow chart showing a flow of increase / decrease parameter determination processing by the power transaction assisting apparatus according to the present invention A flow chart showing the flow of the one-day market bidding process by the power transaction assisting apparatus according to the present invention A flow chart showing the flow of the same-day market bidding process by the power transaction assisting apparatus according to the present invention Graph of expected value of electricity procurement rate Graph of distribution of electricity procurement costs Histogram showing simulation result of power supply charge C (A, B) when parameter (A, B) is (0, 0) Histogram showing simulation results of the power supply charge C (A, B) when the parameter (A, B) is (0.6, -2) Graph of expected value of power supply charge C when standard deviation σ ₁ is changed to 5 Graph of variance of power supply charge C when standard deviation σ ₁ is changed to 5 Graph of expected value of power supply charge C when standard deviation σ ₂ is changed to 0.1 Graph of variance of power supply charge C when standard deviation σ ₂ is changed to 0.1 Graph of the expected value of the power procurement fee C when changing the unit price b of the day to 1.2 Graph of variance of power supply charge C when the unit price b of the day is changed to 1.2 Graph of expected value of power supply charge C when changing the unit price b of the day to 2.8 Graph of variance of the power procurement charge C when the unit price b of the day is changed to 2.8 Graph of expected value of power supply charge C when changing the unit price a of the previous day to 0.5 Graph of variance of power supply charge C when changing the unit price a on the previous day to 0.5 Graph of expected value of power supply charge C when penalty unit price c is changed to 3.5 Graph of variance of power supply charge C when penalty unit price c is changed to 3.5

  A first invention made to solve the above-mentioned problems is a power transaction support device for supporting a bid on a wholesale power transaction market, which is based on past power demand data and predicted power at the time of bidding on the market one day ago. The previous day predicted amount prediction unit predicts the previous day predicted amount which is the demand amount, and the same day predicted amount forecasting unit predicts the current day predicted amount which is the predicted power demand amount at the time of bidding on the day market based on past power demand data A previous day prediction error which is a prediction error between the previous day predicted amount in the past and the corresponding past actual power demand amount, and a prediction between the past said day predicted amount and the corresponding past actual power demand amount A prediction error calculation unit for calculating the prediction error on the current day, which is an error, and the bid amount determined based on the prediction amount on the previous day after correcting the prediction amount on the previous day based on the prediction error on the previous day The day before market bidding section which bids on the day before market, the day predicted amount is corrected based on the day predicted error, and the bid amount determined based on the day predicted amount after the correction It is characterized by having the same day market bidding part which bids on the day market.

  According to the power transaction assisting apparatus of the first invention, the previous day prediction error and the current day prediction error respectively calculated based on the previous day predicted amount and the current day predicted amount and the corresponding past actual power demand amount Based on the previous day predicted amount and the current day predicted amount can be corrected. As a result, according to the prediction accuracy of the previous day and the prediction accuracy of the current day, it becomes possible to allocate the bid amount to the market one day ago and the current day market so as to minimize the expected value of the power supply charge.

  A second invention is characterized in that, in the first invention, the previous day prediction error and the current day prediction error follow a distribution described by a mathematical expression.

  According to the power transaction assisting apparatus in accordance with the second aspect of the present invention, it is possible to easily and accurately obtain the previous day prediction error and the current day prediction error by following the distribution in which the previous day prediction error and the current day prediction error are described using formulas. Become.

  Further, according to a third aspect of the present invention, in the first aspect or the second aspect, the first increase / decrease parameter for increasing / decrementing the predicted amount on the previous day based on the previous day prediction error and the previous day prediction error The increase / decrease parameter determination unit further determines a second increase / decrease parameter for increasing / decreasing the amount, and the increase / decrease parameter determination unit calculates the one calculated based on the previous day predicted amount corrected by the first increase / decrease parameter. Expected value of bid price for day market and expected value of bid price for the current day market calculated based on the predicted day value corrected by the second increase / decrease parameter, and bid for the previous day market and the current day market The first increase / decrease parameter and the first increase / decrease parameter so that the sum with the expected value of the imbalance fee to be paid when the amount falls below the actual demand amount is minimized. And determining the value of the second increase and decrease parameters respectively.

  According to the power transaction assisting apparatus of the third aspect of the invention, it is possible to easily and accurately correct the predicted amount on the previous day and the predicted amount on the same day by using the first increase / decrease parameter and the second increase / decrease parameter. It becomes.

  In a fourth aspect based on the third aspect, the one day earlier market bidding unit bids the sum of the predicted amount on the previous day and the first increase / decrease parameter against the market one day ago as a bid amount If the sum of the predicted amount on the day and the second increase / decrease parameter is equal to or more than the sum of the predicted amount on the previous day and the first increase / decrease parameter, the current day market bidding section takes the difference as the bid amount If a bid is made to the market on the day and the sum of the forecasted quantity on the day and the second increase / decrease parameter is smaller than the sum of the forecasted quantity on the previous day and the first increase / decrease parameter, It is characterized by not bidding.

  According to the power transaction assisting apparatus in accordance with the fourth aspect of the present invention, since the on-the-day market bid unit compares the on-the-day forecast amount after correction and the on-the-day forecast amount after correction on the day. It will be possible to do.

  In the fifth invention according to the third invention, the increase / decrease parameter determination unit determines that the bid amount for the market one day before and the market on the day exceeded the actual demand amount to the expected value of the imbalance fee. The present invention is characterized in that the values of the first increase / decrease parameter and the second increase / decrease parameter are respectively determined by adding the expected value of the power sale income obtained in the case.

  According to the power transaction assistance device pertaining to the fifth aspect of the invention, when surplus electricity is generated in the electric power company, bidding on the market one day and the same day one day ago is appropriately performed, including the operation of selling the surplus power. It will be possible to do.

  The sixth invention is a power transaction support method for supporting a bid on a wholesale power transaction market, wherein a predicted amount of power demand on the previous day which is a predicted power demand amount at the time of a bid on the market one day ago based on past power demand data The previous day predicted amount predicting step for predicting the current day predicted amount, which is the predicted power demand amount at the time of bidding on the current day market, based on the previous day predicted amount predicting step for predicting And the corresponding prediction error of the previous actual power demand amount, and the previous day prediction error of the previous day prediction amount that is the previous day predicted amount on the day and the corresponding past actual power demand amount The prediction error calculation step to calculate each, the previous day prediction amount is corrected based on the previous day prediction error, and the bid amount determined based on the previous day prediction amount after the correction The daily market bid step for bidding on the market, and the forecasted amount on the day is corrected based on the forecast error on the day, and the bid amount determined based on the forecasted amount on the day is corrected on the day It is characterized by having an on-site market bidding step for bidding against.

  According to the power transaction support method of the sixth aspect, the previous day prediction error and the current day prediction error respectively calculated based on the previous day predicted amount and the current day predicted amount and the corresponding past actual power demand amount Based on the previous day predicted amount and the current day predicted amount can be corrected. As a result, according to the prediction accuracy of the previous day and the prediction accuracy of the current day, it becomes possible to allocate the bid amount to the market one day ago and the current day market so as to minimize the expected value of the power supply charge.

  The seventh invention is characterized in that, in the sixth invention, the previous day prediction error and the current day prediction error follow a distribution described by an equation.

  According to the power transaction support method of the seventh invention, it is possible to easily and accurately obtain the previous day prediction error and the current day prediction error by following the distribution in which the previous day prediction error and the current day prediction error are described using formulas. Become.

  In an eighth aspect based on the sixth aspect or the seventh aspect, the first increase / decrease parameter for increasing or decreasing the previous day prediction amount based on the previous day prediction error and the current day prediction error, and the current day prediction The method further includes an increase / decrease parameter determination step of determining a second increase / decrease parameter for increasing / decreasing the amount, wherein the increase / decrease parameter determination step is calculated based on the previous day prediction amount corrected by the first increase / decrease parameter. The expected value of the bid price for the current day market and the expected price of the bid price for the current day market calculated based on the predicted day value corrected by the second increase / decrease parameter, and the expected price for the day before market and the current day market Said first increase so that the sum with the expected value of the imbalance fee to be paid when the bid amount falls below the actual demand amount And determining parameters and the values of the second increase and decrease parameters respectively.

  According to the power transaction support method of the eighth invention, it is possible to easily and accurately correct the previous day predicted amount and the current day predicted amount by using the first increase / decrease parameter and the second increase / decrease parameter. It becomes.

  In a ninth aspect of the invention according to the eighth aspect, in the one day before market bidding step, the sum of the predicted amount on the previous day and the first increase / decrease parameter is used as a bid amount for the one day before market. In the same-day market bidding step, when the sum of the predicted amount on the current day and the second increase / decrease parameter is equal to or more than the sum of the predicted amount on the previous day and the first increase / decrease parameter, the difference is taken as the bid amount If a bid is made to the market on the day and the sum of the forecasted quantity on the day and the second increase / decrease parameter is smaller than the sum of the forecasted quantity on the previous day and the first increase / decrease parameter, It is characterized by not bidding.

  According to the power transaction support method of the ninth invention, since the on-the-day market bidding step compares the on-the-day forecast amount after correction and the on-the-day forecast amount after correction on the day It will be possible to do.

  In a tenth aspect based on the eighth aspect, in the increase / decrease parameter determination step, the bid amount for the market one day before and the market on the day exceeded the actual demand amount to the expected value of the imbalance fee The present invention is characterized in that the values of the first increase / decrease parameter and the second increase / decrease parameter are respectively determined by adding the expected value of the power sale income obtained in the case.

  According to the power transaction support method pertaining to the tenth aspect of the invention, when surplus electricity is generated in the electric power company, bidding on the market one day and the same day one day ago is appropriately performed, including the operation of selling the surplus power. It will be possible to do.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is a schematic view of a power transaction system to which a power transaction assistance device 1 according to the present invention is applied.

  The power transaction support device 1 according to the present invention is for supporting the bid of the electric power company 3 with respect to the wholesale power exchange market established in the wholesale power exchange 2. The power transaction support device 1 is installed at a facility or the like of the electric utility 3 and connected to the wholesale power exchange 2 via a network 4 such as the Internet or another known communication network.

  The wholesale power exchange 2 is the Japan Wholesale Power Exchange (JEPX: Japan Electric Power eXchange), etc., and as a main market, the market one day ago (spot market) 5 and the same day market as its adjustment market (pre-hourly market) 6 and has been established. One day ago market 5 is a market where electricity supplier 3 trades power on the day before the power supply day supplying power to customers (customers), and on the same day market 6 on the day of power supply day Market for trading. In the market 5 one day ago and the market 6 on the day, the power is traded every 48 bidding time slots (pieces) dividing the day into 30 minutes.

  The electric power company 3 may be a retail electric power company, a general power transmission and distribution business, a power transmission business, a specified power transmission and distribution business, a power generation business, and the like. In this embodiment, the power supplier 3 provides power using the power transmission network of a major power company in each area, and is generally called “new power (PPS)”. It is assumed that retail electric utilities that have entered the market since their electricity retailing has been liberalized. However, the power supplier 3 is not limited to the new power (PPS), and may be another power supplier.

  The power transaction assisting apparatus 1 comprises a computer having a known configuration, and determines the distribution of bid amounts to each market of the market 5 and the market 6 on the day one day, thereby enabling the electric utility companies on the market 5 and the day market 6 one day ago Support 3 bids.

  FIG. 2 is a hardware configuration diagram of the power transaction assistance apparatus 1. As shown in FIG. 2, the power transaction support device 1 has a known hardware configuration, and is a CPU (Central Processing Unit) that comprehensively executes various information processing and control of peripheral devices based on a predetermined control program. , RAM (Random Access Memory) 12 functioning as a work area of processor 11, etc., ROM (Read Only Memory) 13 for storing control programs and data executed by processor 11, and network 4 A network interface (I / F) 14 or the like for executing communication processing is provided.

  In addition, the power transaction support device 1 includes, as a peripheral device, an input device 15 configured of a keyboard, a touch panel, and the like for the user of the power transaction support device 1 to input various commands and data; Storage composed of a monitor 16 that displays various information related to the power transaction support process to be executed by 1 and various data used in the power transaction support process, an HDD (Hard Disk Drive) that stores calculation results of those data, etc. It has 17 mag. The components 11-17 described above are connected to one another via the bus 10.

  At least a part of the various functions of the power transaction assistance apparatus 1 described in detail later with reference to FIG. 3 can be realized by the processor 11 executing a predetermined control program (for example, a program for power transaction assistance). The power transaction assistance apparatus 1 is not limited to a computer, and may be composed of another information device (for example, a server or the like) that performs the same function. In addition, at least a part of the various functions of the power transaction assistance apparatus 1 may be replaced by processing by other known hardware.

  FIG. 3 is a functional block diagram of the power transaction assistance apparatus 1. As shown in FIG. 3, the power transaction support device 1 has a demand amount prediction unit 20 that predicts a predicted power demand amount in a bidding time zone in which the electric utility 3 bids, and the market 5 a day before in a bidding time zone. Unit price forecasting unit 30, which predicts unit price of electricity unit price, electricity unit price in the current market 6, and penalty charge (imbalance charge), and prediction error for calculating the prediction error between the predicted amount of power demand and the actual amount of power demand Based on the calculation unit 40, the increase / decrease parameter determination unit 50 that determines increase / decrease parameters for increasing / decrementing the predicted power demand amount based on the prediction error and the output of the unit price prediction unit 30, For the market 5 and the market 6 on the day with the correction and the bid amount determined based on the corrected predicted power demand amount (here, the corrected predicted power demand amount and the bid amount are equal) And a bid unit 60 to bid respectively. Each unit is controlled by a control unit (not shown).

  The demand amount prediction unit 20 includes a previous day predicted amount prediction unit 21, a first storage unit 22, a current day predicted amount prediction unit 23, and a second storage unit 24.

  The previous day forecasted amount forecasting unit 21 is the day of bidding on the market 5 one day ago, that is, the day before the power supply date, the past power demand data (demand performance data) of the power supply destination, weather information of the power supply date, power supply destination On the basis of the event data and the like of the above, the predicted amount g of the previous day which is the predicted power demand amount of the bidding time zone for which the electric power company 3 bids is predicted. Therefore, the predicted amount g on the previous day is the predicted amount of power demand when the electric utility 3 bids on the market 5 one day ago. The previous day prediction amount g predicted by the previous day prediction amount prediction unit 21 is stored in the first storage unit 22.

  On the day of bidding to the market 6 on the day, that is, the day of the power supply day, the forecasted amount prediction unit 23 on the day is based on the past power demand data of the power supply destination, weather information of the power supply day, event data of the power supply destination, etc. The predicted amount h on the day, which is the predicted power demand amount in the bidding time zone, is predicted. Therefore, the forecasted amount h on the day is the forecasted power demand amount when the electric utility 3 bids on the market 6 on the day. The predicted amount h on the current day predicted by the predicted amount prediction unit 23 on the current day is stored in the second storage unit 24.

  It is assumed that the past power demand data of the power supply destination is acquired from the wholesale power exchange 2 or the power demand data storage unit 42 described later. In addition, weather information (including forecast and predicted information) on the date of power supply shall be acquired from the Meteorological Agency or the like. The weather information may be, for example, temperature, humidity, weather, and the like. As described above, in general, the prediction accuracy on the day of the power supply day (when bidding on the day market 6) is the same as the day before the power supply day (when bidding on the market one day ago). It tends to be higher than the prediction accuracy. Therefore, generally, the previous day predicted amount g and the current day predicted amount h are different from each other.

The previous day predicted amount prediction unit 21 and the current day predicted amount prediction unit 23 predict the previous day predicted amount g and the current day predicted amount h, for example, using any one of the following methods (1) to (4) or a combination thereof. can do.
(1) A method of forecasting using multiple regression analysis based on power demand data and meteorological data (air temperature, humidity etc.) at the same time in the past (2) based on the power demand results of the previous day and meteorological data (air temperature, humidity etc) Forecasting method (3) Forecasting method based on power supply date and weather information (air temperature, humidity, weather, etc.) and power demand data of the day when the day is similar (4) Power demand on the same day at specific time in the past several years Method of forecasting based on the relationship with the maximum demand

  Unit price prediction unit 30 includes transaction data acquisition unit 31, power unit price prediction unit 32, and penalty unit price prediction unit 33.

  The transaction data acquisition unit 31 acquires past power transaction data from the wholesale power exchange 2 or another database via the network 4. In the past electricity transaction data, the past electricity unit price in the market 5 a day ago (hereinafter referred to as "the day before electricity unit price a"), the past electricity unit price in the day market 6 (hereinafter referred to as "the day electricity unit price b" And the unit cost of the past penalty charge (hereinafter referred to as “penalty unit cost c”).

  The power unit price prediction unit 32 predicts the power unit price a on the previous day and the power unit price b on the current day based on the past power transaction data acquired by the transaction data acquisition unit 31. Similarly, the penalty unit price prediction unit 33 predicts the penalty unit price c based on the past power transaction data acquired by the transaction data acquisition unit 31.

  The power unit price prediction unit 32 and the penalty unit price prediction unit 33, for example, from the past power transaction data, data of a day on which conditions such as the power supply date and time, day, weather information (air temperature, humidity, weather, etc.) are similar Each unit price of the previous day power unit price a, the current day power unit price b, and the penalty unit price c can be predicted based on the retrieved data.

  The prediction error calculation unit 40 includes a power consumption acquisition unit 41, a power demand data storage unit 42, and a prediction error distribution determination unit 43.

  The power consumption acquisition unit 41 acquires the power consumption data of the power supply destination via the network 4. Specifically, for example, power consumption data may be acquired from a smart meter installed at a power supply destination. The smart meter is a known power meter having a communication function capable of transmitting power consumption data to a predetermined communication party, and is used by the power consumption acquisition unit 41 at predetermined time intervals via the network 4 Send power data. The power consumption data acquired by the power consumption acquisition unit 41 is stored in the power demand data storage unit 42 as the power demand data of the power supply destination. The smart meter is installed at each power supply destination, and the used power amount acquisition unit 41 acquires the used power amount data of each power supply destination from the smart meter installed at each power supply destination. Do.

The prediction error distribution determination unit 43 determines the past based on the previous day prediction amount g stored in the first storage unit 22 and the past power demand data of the power supply destination stored in the power demand data storage unit 42. The previous day prediction error distribution P _G (x), which is the prediction error distribution of the previous day predicted amount g and the corresponding past actual power demand, is determined. Thus, it is possible to obtain a prediction error for the previous day predicted amount g predicted by the previous day predicted amount prediction unit 21 in the past.

In addition, the prediction error distribution determination unit 43 predicts the current day on the basis of the past predicted amount h stored in the second storage unit 24 and the past power demand data stored in the power demand data storage unit 42. A prediction error distribution P _H (x), which is a prediction error distribution of the quantity h and the corresponding past actual power demand, is determined. As a result, it is possible to obtain a prediction error for the current day predicted amount h predicted in the past by the current day predicted amount prediction unit 23.

It is assumed that each prediction error distribution of the previous day prediction error distribution P _G (x) and the current day prediction error distribution P _H (x) follows a normal distribution of the mean μ and the variance σ ² . By making each prediction error distribution P _G (x) and P _H (x) follow the normal distribution, it is possible to more accurately obtain each prediction error of the previous day predicted amount g and the current day predicted amount h. It should be noted that the assumption that each prediction error distribution follows a normal distribution is the central limit theorem (which discusses the error between the mean of the sample and the mean of the sample, and the distribution of the population is any The error between the mean and the true mean is approximately based on a normal distribution when the sample size (number) is increased.

The prediction error distribution determination unit 43 determines the previous day prediction error distribution P _G (x) and the current day prediction error distribution P _H (x), for example, using any of the following methods (1) to (3). be able to.
(1) A method of representing the probability density distribution using parametric models having a relatively small number of parameters, fitting the model to data, and estimating the parameter that best fits the data (2) specific functional type Method that uses non-parametric models to determine the shape of distribution depending on data without assuming (3) intermediate of the above two methods to express complex distribution A semi-parametric method that expresses a more general functional type than a parametric model by systematically increasing the number of parameters

The increase / decrease parameter determination unit 50 determines the unit price of the previous day power unit price a, the current unit price b, and the penalty unit price c predicted by the unit price forecast unit 30, and the previous day prediction error distribution P _G calculated by the prediction error calculation unit 40. x) and the first increase or decrease for correcting the previous day predicted amount g and the current day predicted amount h predicted by the demand amount prediction unit 20 based on each prediction error distribution of the prediction error distribution P _H (x) on the day The parameter A and the second increase / decrease parameter B are respectively determined.

  Specifically, the increase / decrease parameter determination unit 50 uses a computer software such as Masematica (Mathematica) to expect an expected bid price for the market 5 a day ago, an expected bid price for the day market 6, and the electricity business Sum with the expected value of the penalty charge to be paid when the bidding volume (ie procurement volume) bid by the third party to the market 5 and the day market 6 one day earlier falls below the actual electricity demand volume (hereinafter referred to as The values of the first increase / decrease parameter A and the second increase / decrease parameter B are determined such that the expected value of From a statistical point of view, the smaller the expected value of the power procurement fee, the lower the cost of the electricity supplier 3 is to be able to procure electricity, and the first one is to minimize the expected value of the power procurement fee By selecting the increase / decrease parameter A and the second increase / decrease parameter B, the profit of the electric utility 3 improves. Here, the expected value of the power supply charge can be expressed by the following Equation 1.

In Equation 1 above, E [C ₁ + C ₂ + C ₃ ] is the expected value of the power procurement fee, E [C ₁ ] is the expected value of the bid amount for the market 5 a day ago, and E [C ₂ ] is E [C ₃ ] is the expected value of the penalty charge, which is the expected value of the bid amount for the market 6 of the day. The method of deriving Equation 1 will be described in detail later.

  The bidding unit 60 includes a one day earlier market bidding unit 61 and a same day market bidding unit 62.

  The one day ago market bidding unit 61 first adds the first increase / decrease parameter A obtained by the increase / decrease parameter determination unit 50 to the previous day prediction amount g obtained by the previous day prediction amount prediction unit 21 of the demand amount prediction unit 20, The previous day predicted amount g is increased or decreased. Then, it bids on the market one day before with the predicted amount g on the previous day after correction, that is, “g + A” which is the sum of the predicted amount g on the previous day and the first increase / decrease parameter A as a bid amount.

  First-day market bidding portion 62 adds the second increase / decrease parameter B obtained by increase / decrease parameter determination portion 50 to the current-day predicted amount h obtained by current-day predicted amount prediction portion 23 of demand amount prediction portion 20 first Increase or decrease the prediction amount h. Subsequently, the corrected predicted amount h on the day, that is, “h + B” which is the sum of the predicted amount h on the day and the second increase / decrease parameter B, is calculated by the predicted amount g the day before Compare with certain "g + A". As a result of this comparison, as shown in FIG. 4A, when “h + B” is “g + A” or more, the difference “h + B− (g + A)” is bid on the current market 6 as a bid amount. On the other hand, as shown in FIG. 4B, when “h + B” is smaller than “g + A”, no bid is made to the market 6 on the day. This is because the bid amount (g + A) bid on the market 5 a day ago has already exceeded the forecasted amount (h + B) on the day after the correction, that is, the forecasted demand amount, and therefore the bid on the day market 6 is wasted.

  Next, the flow of each process of the increase / decrease parameter determination process, the one-day market bid process, and the on-the-day market bid process, which is the power transaction support process by the power transaction assistance apparatus 1, is referred to each flowchart in FIG. To explain.

  FIG. 5 is a flowchart showing the flow of increase / decrease parameter determination processing. This increase / decrease parameter determination process is a process for determining the first increase / decrease parameter A and the second increase / decrease parameter B, and is performed one day before the bidding day to the market 5, that is, the day before the power supply date.

  First, the power unit price prediction unit 32 of the unit price prediction unit 30 predicts the power unit price a on the previous day and the power unit price b on the current day based on the past power transaction data (step ST101).

  Subsequently, the penalty unit price prediction unit 33 of the unit price prediction unit 30 predicts the penalty unit price c based on the past power transaction data (step ST102).

Next, in the prediction error distribution determination unit 43 of the prediction error calculation unit 40, the previous day prediction amount g and the previous day prediction amount g stored in the first storage unit 22 and the second storage unit 24 and the power demand data Based on the past actual power demand stored in the storage unit 42, the previous day which is the prediction error distribution between the previous day predicted quantity g and the previous day predicted quantity h and the corresponding past actual power demand The prediction error distribution P _G (x) and the current day prediction error distribution P _H (x) are determined (step ST103).

The increase / decrease parameter determination unit 50 then determines the previous day power unit price a, the current unit price b, the penalty unit price c, the previous day prediction error distribution P _G (x), and the current day prediction error distribution P _H (x) (more specifically, The first increase / decrease parameter A and the second increase / decrease parameter B are obtained based on the variation of the prediction error, that is, the prediction accuracy) (step ST104 (increase / decrease parameter determination step)). As described above, the first increase / decrease parameter A and the second increase / decrease parameter B are the expected value of the power supply charge represented by Equation 1 above (the expected value of the bid amount for the market 5 a day ago, Is determined so as to minimize the sum of the expected value of the bid price and the expected value of the imbalance fee.

  The first change parameter A and the second change parameter B, which have been determined, are used in the following one-day-a-day market bid processing and on-the-day market bid processing. Remember.

  FIG. 6 is a flow chart showing the flow of the market bid processing one day earlier. This one day ago market bidding process is a process for bidding on the market 5 one day ago, and is implemented the day before the day of bidding on the market 5, ie, the day before the power supply day.

  First, the previous day predicted amount prediction unit 21 of the demand amount prediction unit 20 predicts the previous day predicted amount g, which is the predicted power demand amount at the time of bidding on the market 5 one day ago, based on the past power demand data (step ST201) .

  Next, one day before market bidding portion 61 adds the first increase / decrease parameter A determined in the above-mentioned increase / decrease parameter determination processing to the previous day predicted amount g to increase / decrease the previous day predicted amount g (step ST202) .

  Then, the one-day before market bidding part 61 bids on the one-day-before market 5 with "g + A" which is the predicted amount g the day after the correction as a bid amount (step ST203).

  The bid amount (g + A) which has been bid on the market 5 days ago is stored in the storage 17 (see FIG. 2) or the like of the power transaction support apparatus 1 for use in the on-day market bid process.

  FIG. 7 is a flow chart showing the flow of the current day market bidding process. This day market bid process is a process for bidding on the day market 6, and is performed on the day of bidding on the day market 6, that is, the day of the power supply date.

  First, the current day predicted amount prediction unit 23 of the demand amount prediction unit 20 predicts the current day predicted amount h which is the predicted power demand amount at the time of bidding on the current day market 6 based on the past power demand data (step ST301).

  Subsequently, the second-order increase / decrease parameter B determined in the above-mentioned increase / decrease parameter determination processing is added to the current-day predicted amount h by the current-day market bid portion 62 of the bid portion 60 to increase / decrease the current day predicted amount h Step ST302).

  Next, on the current day market bidding portion 62, it is determined whether “h + B”, which is the predicted amount on the current day after correction, is equal to or larger than “g + A”, which is the predicted amount g on the previous day after correction (step ST303). When it is determined that “h + B” is “g + A” or more (step ST303: Yes), the process proceeds to step ST304. On the other hand, when it is determined that “h + B” is smaller than “g + A” (step ST303: No), the process is ended without bidding on the market 6 on the day.

  In step ST304, the bid on the day market 6 is made with "h + B- (g + A)" which is the difference between "h + B" and "g + A" as a bid amount.

  As described above, according to the present invention, the predicted amount g and the predicted amount h on the previous day are calculated by the first increase / decrease parameter A and the second increase / decrease parameter B determined so as to minimize the expected value of the power supply rate. Each can be corrected. Therefore, according to the prediction accuracy on the previous day (that is, the predicted amount on the previous day g) and the prediction accuracy on the current day (that is, the predicted amount h on the day) It becomes possible to allocate a bid amount.

  Next, a method of deriving the above equation 1 representing the expected value of the power supply rate will be described. As described above, in the market 5 one day ago and the market 6 on the day, the transaction is performed in every 48-frame bidding time period, which divides the day into units of 30 minutes. Below, the case where the electric power company 3 bids for the t-th frame of the 48-piece bidding time zone will be described.

(Power supply fee)
First, a power supply charge, which is a charge required for the electric power company 3 to procure power from the market 5 and the market 6 on the day, is obtained. Electricity procurement fee, as the sum of the bid C ₁ to the day before the market 5, the bid C ₂ for the day market 6, the penalty fee C ₃ procurement amount of electric utilities 3 pay When the lower demand Desired.

One day ago The power unit price on the day before the t th frame of the market 5 is a. Then, for the market 5 one day ago, a bid is made with a bid amount g + A obtained by adding a first increase / decrease parameter A (hereinafter simply referred to as “parameter A”) to the predicted amount g the day before. Therefore, the bid amount C ₁ for the market 5 one day ago is expressed by the following equation 2.

The power unit price of the t th frame of market 6 on the day is b. Then, for the market 6 of the day, the sum of the bid amount g + A for the market 5 one day ago and the bid amount for the market 6 of the day is the second increase / decrease parameter B in the forecast amount h on the day (hereinafter referred to simply as “parameter B Bid) so as to be h + B added with That is, in the case of g + A ≦ h + B, a bid is made on the day market 6 with a bid amount of h + B− (g + A). On the other hand, in the case of g + A> h + B, no bid is made to the market 6 on the day. Therefore, bid C ₂ for the day market 6 is expressed by Equation 3 below. However, δ in Equation 3 below is a function that takes 1 if the conditional expression in the parentheses is satisfied, and takes 0 otherwise.

Let f be the predicted power demand for the tth frame and c be the penalty unit price. As described above, in the case of g + A ≦ h + B ≦ f, or in the case of h + B <g + A ≦ f, the procurement amount of the electric utility 3 falls below the demand amount and a penalty charge occurs. Accordingly, penalty C ₃ is represented by Equation 4 below. However, δ in Equation 4 below is a function that takes 1 if the conditional expression in parentheses is satisfied, and takes 0 otherwise.

The power procurement fee C is C ₁ + C ₂ + C _{3 obtained} by adding the bid amount C ₁ to the market 5 a day, the bid amount C ₂ to the day market 6 and the penalty fee C ₃ obtained as described above.

(Expected value of electricity supply rate)
Next, the expected value E [C ₁ + C ₂ + C ₃ ] of the power supply rate is obtained. Let G = f−g and H = f−h be random variables, and let the probability density functions (prediction error distribution) of G and H be P _G (x) and P _H (y), respectively. Further, the predicted power demand amount f, the power unit price a on the previous day, the power unit price b on the day, and the penalty unit price c are each independent.

The expected value E [C ₁ ] of the bid amount C ₁ for the market 5 a day ago is expressed by the following Equation 5.

Using G and H, the bid amount C ₂ for the current day market 6 is expressed by the following equation 6.

Therefore, the expected value E [C ₂ ] of the bid amount C ₂ for the current day market 6 can be expressed by the following equation 7. However, Equation 7 is an equation that does not specifically define the prediction error distributions P _G (x) and P _H (y) (the same applies to Equations 9 and 10 described later).

Penalty C ₃ is the use of G and H, it is expressed by Equation 8 below.

Therefore, the expected value E [C ₃ ] of the penalty charge C ₃ can be expressed by the following equation 9.

The expected value E [C ₁ ] of the bid amount C ₁ for the market 5 a day ago, the expected value E [C ₂ ] of the bid amount C ₂ for the day market 6 and the expectation of the penalty fee C _{3 obtained} as described above _E which is the sum of values _{E [C 3] [C 1} ] + E [C 2] + E [C 3] becomes the expected value _E of the power sourcing fee _{[C 1 + C 2 + C} 3]. Therefore, the expected value E [C ₁ + C ₂ + C ₃ ] of the power supply charge can be expressed by Equation 10 below. Equation 10 is identical to Equation 1 above. In this way, the above equation 1 can be derived. As described above, Formula 7, Formula 9, and Formula 10 do not specify the prediction error distributions P _G (x) and P _H (y). In the following, examples in which the prediction errors follow a normal distribution will be described in detail including specific numerical calculations, but as apparent from the mathematical description of Equations 7, 9, and 10, the present invention has normal distribution of prediction errors. It is needless to say that even in the case where it does not conform to the above, it can be effectively applied as long as it conforms to the distribution described by the mathematical expression.

  When considering the embodiment of the present invention, the patterns of power demand differ between individual facilities such as buildings and a group of clusters composed of a plurality of households. That is, in general, the distribution of prediction errors differs depending on the scale to which the electric utility 3 supplies power. Therefore, since it is considered that the distribution of the error of the demand forecast may be affected by the characteristics of the target to which the power is supplied, the power transaction support device 1 (described above) can apply various distributions as the prediction error distribution. It is practically meaningful to configure the power transaction support program operating in the power transaction support device 1).

(Expected value when prediction error follows normal distribution)
Next, assuming that the prediction error of the predicted power demand amount f with respect to the actual power demand amount follows a normal distribution with an average of 0, the expected value of the power procurement charge E [C ₁ + C ₂ + C ₃ ] (hereinafter simply referred to The “expected value E [C]” or “expected value E” will be described.

The prediction error G of the previous day prediction amount g for the actual power demand amount is a normal distribution with an average of 0 and variance σ ₁ (t) ² = σ ₁ ² and the prediction error H of the day prediction amount h for the actual demand amount is It is assumed to follow a normal distribution with mean 0 and variance σ ₂ (t) ² = σ ₂ ² respectively. In this case, σ ² = σ ^(t) if ^{_{2 = σ 1 (t) 2}} + σ 2 (t) 2 and, in the expected value E _{[C 2]} The formula 11 below the _{C 2,} the expected value of the _{C 3} E [C ₃ ] can be represented by the following Equation 12, respectively.

Therefore, the expected value E [C] of the t-th frame can be expressed by the following Equation 13.

(Numerical calculation of expected value)
In the case where the procurement conditions for the electric power company 3 to procure power from the market 5 and the market 6 on the day before are the following, the expected value E [C] of the t th frame is obtained by numerical calculation.
f = 100, σ ₁ = √3, σ ₂ = √2, a = 1, b = 2, c = 3

Regarding the above procurement conditions, the standard deviation σ ₁ = √3 takes into consideration that the target level of the prediction accuracy of the forecast amount g of the electric utility 3 on the previous day is within ± 3% of the actual power demand amount. It is. Further, as described above, since the predicted amount g on the previous day is higher in prediction accuracy than the predicted amount h on the day, the standard deviation σ ₂ = √2. In addition, the penalty unit price c = 3 takes into consideration that the penalty charge may be three or more times the power unit price a on the previous day. The current electricity unit price b was an average of the electricity unit price a on the previous day and the penalty unit price c.

  When parameter A and parameter B are moved in the range of −1.9 ≦ A ≦ 3 and −4.9 ≦ B ≦ 0, points (A, B, E [C]) draw the graph of FIG. However, it is assumed that the parameter A moves an axis directed in the direction of 10 o'clock, and the parameter B moves an axis directed in the direction of 1 o'clock in 0.1 steps. Moreover, the graphs of FIG. 9 and FIG. 12 to FIG. 23 described later shall also conform to this.

  As shown in FIG. 8, when the values of the parameters A and B exceed a certain upper limit, the expected value E [C] monotonously increases as the values of the parameters A and B increase. Also, when the values of the parameters A and B fall below a certain lower limit, the expected value E [C] monotonously increases as the values of the parameters A and B decrease. Therefore, if the increase / decrease tendency of the expected value E [C] around the parameters (A, B) = (0, 0) is examined, the parameter (A, B) which minimizes the expected value E [C] can be obtained. it can.

  The parameters (A, B) that minimize the expected value E [C] are determined by numerical calculation to be (0.6, −2), and the expected value E [C] at this time is E [C] ̃101. It will be 835. On the other hand, when the previous day predicted amount g and the current day predicted amount h are not corrected by the parameters A and B, that is, the expected value E [C] in the case of A = B = 0 is E [C] ̃102.329.

  Therefore, under the above procurement conditions, the market 5 a day ago bids slightly more than the forecast amount g on the previous day (parameter A = 0.6), and the market 6 on the day is somewhat smaller than the forecast amount h on the day (parameter It was found that the expected value E [C] can be reduced by bidding on B = −2) as compared with the case where the previous day predicted amount g and the current day predicted amount h are not corrected by the parameters A and B.

(Stability of expected value)
Even if there is a method to reduce the expected value E [C] as described above, it can not be used in actual power procurement if the variance of the power procurement fee C is large and the convergence of the expected value E [C] is slow. is there. This is because financial risks arise. Therefore, the relationship between the parameters (A, B) for minimizing the expected value E [C] and the dispersion of the power supply charge C was examined. In conclusion, it was found that the parameters (A, B) that minimize the expected value E [C] reduce the variance of the power supply charge C.

The power supply simulation using the Monte Carlo method was performed one million times for each parameter (A, B). In this simulation, the following procurement conditions (hereinafter referred to as "reference procurement conditions") were used.
f = 100, σ ₁ = √3, σ ₂ = √2, a = 1, b = 2, c = 3

Further, it is assumed that the predicted amount g on the previous day and the predicted amount h on the present day follow a normal distribution of g to N (f, σ ₁ ² ), h to N (f, σ ₂ ² ) and an average f.

  When parameter A and parameter B are moved in the range of −1.9 ≦ A ≦ 3 and −4.9 ≦ B ≦ 0, points (A, B, E [C]) draw the graph of FIG. Since the power procurement fee C depends on the parameters (A, B), the power procurement fee C is hereinafter referred to as the power procurement fee C (A, B) in order to make this explicit.

  As a result of this simulation, the parameters (A, B) for minimizing the variance of the power procurement fee C become (1, -1.4), and the variance V [C (1, -1. 4)] became 1.693098. Also, the variance V [C (0, 0)] at the parameter (A, B) = (0, 0) is 2.879739, the variance V at the parameter (A, B) = (0.6, −2) [C (0.6, -2)] became 1.821432.

  Therefore, the variance V [C (0, 0)] of the power supply charge C when the previous day predicted amount g and the current day predicted amount h are not corrected by the parameters A and B is the previous day predicted amount g and the current day predicted amount h as the parameter A , B (1, -1.4), and it is about 1.7 times the variance V [C (1, -1.4)] of the power supply fee C, and the predicted amount g on the previous day and the predicted amount h on the same day Variance V [C (0.6, -2)] of the power supply charge C when the parameter A is corrected with parameters B (0.6, -2) is variance V [C (1, -1.4) ] Was found to be about 1.07 times that of

  10 and 11 are histograms respectively showing simulation results of the power supply charge C (A, B) when the parameters (A, B) are (0, 0) and (0.6, -2). is there.

  Under the above procurement conditions, when the previous day predicted amount g and the current day predicted amount h are corrected with the parameters A and B, compared to the case where the previous day predicted amount g and the current day predicted amount h are not corrected with the parameters A and B, It has been found that the dispersion of the charge C is reduced, which enables more stable procurement of power. This is considered to be because the increase of the bid amount to the market 5 a day earlier suppressed the occurrence of the penalty charge. Further, as can be seen from the graph of FIG. 9, when the value of the parameter A is large, the parameter B hardly affects the dispersion of the power supply charge C. It is considered that this is because, in most cases, it is possible to secure the necessary procurement amount in the market 5 one day ago, because the difference between the variance of the predicted quantity g the day before and the variance of the predicted quantity h on the day is small.

(Simulation when changing procurement conditions)
Next, it was examined how the parameters (A, B) for minimizing the expected value E tended to change when the above standard procurement conditions were changed. Specifically, a part of the standard procurement conditions is changed, power supply simulation using the Monte Carlo method is performed one million times for the changed procurement conditions, and the expected value E and variance V obtained by this simulation are I drew a graph. The parameter A = 0.6 obtained under the reference procurement conditions is referred to as a reference value of the previous day, and the parameter B = -2 is referred to as a reference value of the present day.

(When the prediction accuracy of the previous day prediction amount g is lowered)
Under the above standard procurement conditions, the value of the standard deviation σ ₁ was √3. The value of this standard deviation σ ₁ was changed to 5. This change means that the prediction accuracy of the previous day predicted amount g is lowered with respect to the reference procurement condition. FIGS. 12 and 13 are graphs respectively showing the expected value E and the variance V obtained by the simulation for the changed procurement conditions. However, the range of the parameters A and B is set to −1.9 ≦ A ≦ 3 and −4.9 ≦ B ≦ 0.

  The parameters (A, B) for minimizing the expected value E obtained by this simulation are (0.8, −1), and the parameters (A, B) for minimizing the variance V of the power supply charge C are: (1, -0.4). Then, the expected value E and the variance V when the parameters (A, B) take these values and the value of (0, 0) are as follows.

  From the simulation results, when the prediction accuracy of the previous day predicted amount g is lowered with respect to the reference procurement condition, parameter A and parameter B for parameters (A, B) that minimize the expected value E and the variance V Were found to tend to increase more than the previous day reference value (parameter A = 0.6) and the current day reference value (parameter B = −2).

  When the prediction accuracy of the previous day predicted amount g is increased relative to the standard procurement condition by performing the same simulation, parameters (A, B) that minimize the expected value E and the variance V are parameters. It can be easily derived that A and parameter B tend to decrease below the previous day's reference value and the current day's reference value, respectively.

(When the prediction accuracy of the predicted amount h on the day is increased)
Under the standard procurement conditions, the value of the standard deviation σ ₂ was √2. The value of this standard deviation σ ₂ was changed to 0.1. This change means that the prediction accuracy of the predicted amount h on the day has been made higher than that of the standard procurement condition. FIG. 14 and FIG. 15 are graphs respectively showing the expected value E and the variance V obtained by the simulation for this changed procurement condition. However, the ranges of the parameters A and B are set to −1.9 ≦ A ≦ 3 and −1.9 ≦ B ≦ 3.

  The parameters (A, B) for minimizing the expected value E determined by this simulation are (0.1, -0.1), and the parameters (A, B) for minimizing the variance V of the power supply charge C ) Was (0.1, 0). Then, the expected value E and the variance V when the parameters (A, B) take these values and the value of (0, 0) are as follows.

  From the simulation results, when the prediction accuracy of the predicted amount h on the day is increased relative to the standard procurement condition, the parameter A is the previous day standard for the parameters (A, B) that minimize the expected value E and the variance V. It was found that parameter B tends to increase above the day's reference value, being lower than the value.

  When the prediction accuracy of the predicted amount h on the day is lowered relative to the reference procurement condition by performing the same simulation, parameters (A, B) for minimizing the expected value E and the variance V are parameters. It can be easily derived that A tends to increase above the previous day's standard value and parameter B tends to decrease below the day's standard value.

(When lowering the unit price b of the day)
Under the standard procurement conditions, the value of electricity unit price b on the day was 2. The power unit price b was changed to 1.2 on this day. This change means that the power unit price b on the current day is lowered with respect to the standard procurement condition, that is, the power unit price b on the current day approaches the power unit price a on the previous day. FIGS. 16 and 17 are graphs respectively showing the expected value E and the variance V obtained by the simulation for the changed procurement conditions. However, the range of the parameters A and B is −1.9 ≦ A ≦ 3 and −2.9 ≦ B ≦ 2.

  The parameters (A, B) for minimizing the expected value E obtained by this simulation are (−0.1, −0.5), and the parameters (A, B) for minimizing the variance V of the power supply charge C. B) was (-0.1, 0). Then, the expected value E and the variance V when the parameters (A, B) take these values and the value of (0, 0) are as follows.

  From the simulation results, when the current unit price b is lowered with respect to the standard procurement condition, that is, when the current unit price b approaches the previous day power value, the parameter that minimizes the expected value E and the variance V (A , B), it was found that the parameter A tends to decrease below the previous day's reference value, and the parameter B tends to increase above the day's reference value.

(When raising the unit price b of the day)
Under the standard procurement conditions, the value of electricity unit price b on the day was 2. The power unit price b was changed to 2.8 on this day. This change means that the power unit price b on the day has been raised with respect to the standard procurement condition, that is, the power unit price b on the day is brought closer to the penalty unit price c. FIG. 18 and FIG. 19 are graphs respectively showing the expected value E and the variance V obtained by the simulation for this changed procurement condition. However, the range of the parameters A and B is set to −1.9 ≦ A ≦ 3 and −4.9 ≦ B ≦ 0.

  The parameters (A, B) for minimizing the expected value E determined by this simulation are (0.7, -3.8), and the parameters (A, B) for minimizing the variance V of the power supply charge C. ) Was (1.2, -2.2). Then, the expected value E and the variance V when the parameters (A, B) take these values and the value of (0, 0) are as follows.

  From this simulation result, when the current unit price b is increased with respect to the standard procurement condition, that is, when the current unit price b approaches the penalty unit cost c, the parameter (A, As for B), it was found that the parameter A tends to increase above the previous day's reference value, and the parameter B tends to decrease below the day's reference value.

(If the previous day's electricity unit price a was lowered)
Under the standard procurement conditions, the previous day electricity unit price a value was 1. The power unit price a was changed to 0.5 the day before. This change means that the power unit price a on the previous day was lowered with respect to the standard procurement conditions. FIGS. 20 and 21 are graphs respectively showing the expected value E and the variance V obtained by the simulation for the changed procurement conditions. However, the range of the parameters A and B is −0.9 ≦ A ≦ 3 and −4.9 ≦ B ≦ 0.

  The parameters (A, B) for minimizing the expected value E determined by this simulation are (1.6, −2.5), and the parameters (A, B) for minimizing the variance V of the power supply charge C ) Was (3.1, -1.7). Then, the expected value E and the variance V when the parameters (A, B) take these values and the value of (0, 0) are as follows.

  From the simulation results, when the power unit price a on the previous day is lowered with respect to the standard procurement condition, the parameter A is lower than the standard value on the previous day for parameters (A, B) that minimize the expected value E and the variance V. It was found that the parameter B tends to decrease below the reference value on the day.

  If the same day electricity unit price a is raised with respect to the standard procurement condition by performing the same simulation, that is, if the previous day electricity unit price a is brought closer to the same day electricity unit price b, the expected value E and the variance V are minimized. As for the parameters (A, B), it can be easily derived that the parameter A tends to decrease below the previous day's reference value and the parameter B tends to increase above the day's reference value.

(If you increase the penalty rate c)
Under the standard procurement conditions, the penalty unit price c was 3. This penalty unit price c was changed to 3.5. This change means that the penalty unit price c has been raised with respect to the standard procurement condition. FIGS. 22 and 23 are graphs respectively showing the expected value E and the variance V obtained by the simulation for the changed procurement conditions. However, the range of the parameters A and B is set to −1.9 ≦ A ≦ 3 and −4.9 ≦ B ≦ 0.

  The parameters (A, B) for minimizing the expected value E obtained by this simulation are (0.8, -1.6), and the parameters (A, B) for minimizing the variance V of the power supply charge C ) Was (1.2, -1). Then, the expected value E and the variance V when the parameters (A, B) take these values and the value of (0, 0) are as follows.

  From the results of this simulation, when the penalty unit price c is raised with respect to the standard procurement condition, the parameters A and B are the same as the previous day for parameters (A, B) that minimize the expected value E and the variance V. It was found that the values tend to increase more than the value (parameter A = 0.6) and the reference value on the day (parameter B = −2).

  Also, by performing the same simulation, the expected value E and the variance V are minimized when the penalty unit price c is lowered with respect to the standard procurement condition, that is, when the penalty unit price c is brought closer to the power unit price b on the day. As for the parameters (A, B), it was found that the parameters A and B tend to be lower than the previous day reference value (parameter A = 0.6) and the current day reference value (parameter B = −2).

The results of the above simulations are summarized in Table 1. Table 1 shows an increase or decrease with respect to the previous day reference value and the current day reference value of the parameter (A, B) that minimizes the expected value E and the variance V when a part of the procurement conditions is changed. Note that in the above, only one case is given to change the elements of the procurement conditions (standard deviation σ ₁ , standard deviation σ ₂ , power unit price a on the previous day, power unit price b on the day, penalty unit cost c) However, it is mathematically obvious that the increase and decrease tendency of the parameters A and B accompanying the change of each element described above is universal.

Hereafter, the modification of this invention is demonstrated. Although penalty C ₃ in the above described embodiment are given by Equation 4, given by equation 20 below in the modification.

The third term and the fourth term are added to the right side of the equation 20 in comparison with the equation 4. In the third and fourth terms, d represents the selling price. That is, when the actual power procurement amount is lower than the predicted power demand amount f, Equation 20 generates a penalty charge based on the penalty unit price c, while the actual power procurement amount exceeds the predicted power demand amount f (ie, (When surplus power occurs) means that power sales revenue based on the power sales unit price d is generated (this power sales revenue is expressed as a negative value on the equation 20). Therefore, cases penalty C ₃ as a total is a negative value is present in the modified example. In addition, when the electric power company 3 owns energy storage means, such as a storage battery, the electric power company 3 temporarily stores surplus electric power (ie,-(f-h-B) or-(f-g-A)). It is also possible to charge a storage battery or the like and to discharge (sell) when the power demand is large.

  In the modified example, the increase / decrease parameter determination unit 50 (see FIG. 3) performs “expected value of bid amount for the market 5 one day before” and “bid amount for the day market 6” in the increase / decrease parameter determination step (see FIG. 5, ST104). Of the first increase / decrease parameter A and the second increase / decrease so that the sum of “expected value of” and “expected value of value obtained by adding power sales revenue (here, a negative value) to Determine the value of parameter B respectively. With regard to this “expected value of the penalty fee plus the power sales revenue”, the “selling price obtained when the bid amount for the market one day ago and the day's market exceeds the actual demand amount” It may be paraphrased as "the value which added the expectation value (negative value) of income".

Also in the modified example, if G = f−g and H = f−h are random variables, the penalty charge C ₃ when the power sale unit price d is included is expressed by Equation 21 below using G and H. expressed.

Also in the modified example, assuming that the probability density functions (prediction error distribution) of G and H are P _G (x) and P _H (y), respectively, the expected value E [C _{3 of} the penalty charge C ₃ taking into account the power sale income. ] Can be expressed by Equation 22 below.

Therefore, the expected value E [C ₁ + C ₂ + C ₃ ] of the power charge in the t th frame on the day in the modified example can be expressed by the following equation 23 by using equation 22, equation 5 and equation 7. .

  As described above, according to the modification, even if it is possible for the electric utility 3 to sell the surplus power, the bid amount for the market one day and the day one market is set so as to minimize the expected value of the power supply charge. It becomes possible to distribute.

  Although the present invention has been described above based on the specific embodiments, these embodiments are merely examples, and the present invention is not limited by these embodiments. The components of the power transaction support apparatus and the power transaction support method according to the present invention described in the above embodiments are not necessarily all essential, and can be selected appropriately without departing from at least the scope of the present disclosure. It is possible.

  The power transaction support apparatus and the power transaction support method according to the present invention are configured to reduce the amount of bids for the day-ahead market and the current day market so as to minimize the expected value of the power supply charge according to the predicted accuracy of the previous day and the predicted accuracy of the current day. It is useful as a power transaction support device and a power transaction support method that can be distributed.

DESCRIPTION OF SYMBOLS 1 electric power transaction support apparatus 2 wholesale power exchange 3 electric power company 4 network 5 one day ago market 6 present day market 20 demand amount forecasting part 21 previous day predicted quantity prediction part 22 1st storage part 23 present day prediction amount forecasting part 24 2nd storage part 30 Unit Price Forecasting Unit 31 Transaction Data Acquisition Unit 32 Power Unit Price Prediction Unit 33 Penalty Unit Price Prediction Unit 40 Prediction Error Calculation Unit 41 Power Consumption Acquisition Unit 42 Power Demand Data Storage Unit 43 Prediction Error Distribution Determination Unit 50 Increase / Decrease Parameter Determination Unit 60 Bid Unit 61 days ago market bidding section 62 day market bidding section

Claims (10)

A power transaction support device for supporting a bid on a wholesale power transaction market, comprising:
The previous day predicted amount prediction unit that predicts the previous day predicted amount, which is the predicted power demand amount at the time of a bid to the market one day ago, based on past power demand data;
On the day, a forecasting amount forecasting unit that predicts a forecasting amount on the day, which is a forecasting power demand amount at the time of bidding on the day market, based on past power demand data;
The previous day prediction error, which is a prediction error between the previous day predicted amount in the past and the corresponding past actual power demand amount, and the prediction error between the past predicted day amount in the past and the corresponding past actual power demand amount A prediction error calculation unit that calculates the prediction error of the day,
A previous-day market bidding unit that corrects the previous day predicted amount based on the previous day prediction error and bids the previous day market with the bid amount determined based on the corrected previous day predicted amount;
The current market bid unit corrects the predicted amount on the current day based on the predicted error on the current day, and bids the market on the current day with a bid amount determined on the current predicted amount on the current day. A power transaction support device characterized by   The power transaction assistance device according to claim 1, wherein the previous day prediction error and the current day prediction error follow a distribution described by an equation. The increase / decrease parameter determination unit further determines a first increase / decrease parameter that increases / decreases the previous day prediction amount based on the previous day prediction error and the current day prediction error, and a second increase / decrease parameter that increases / decreases the current day prediction amount. Equipped
The increase / decrease parameter determination unit is configured to calculate an expected value of a bid price for the one-day-ahead market calculated based on the previous day predicted amount corrected by the first increase / decrease parameter, and the day corrected by the second increase / decrease parameter. The sum of the expected value of the bid amount for the current day market calculated based on the predicted amount and the expected value of the imbalance fee to be paid when the bid amount for the day before market and the current day market falls below the actual demand amount. The power transaction assistance device according to claim 1 or 2, wherein the values of the first increase / decrease parameter and the second increase / decrease parameter are respectively determined so as to be minimum. The one-day-ahead market bidding unit makes a bid on the one-day-ahead market with the sum of the predicted amount on the previous day and the first increase / decrease parameter as a bid amount,
When the sum of the predicted amount on the day and the second increase / decrease parameter is equal to or more than the sum of the predicted amount on the previous day and the first increase / decrease parameter, the market bid unit on the current day sets the difference as the bid amount. If a bid is made to the market on the day and the sum of the forecasted amount on the day and the second increase / decrease parameter is smaller than the sum of the forecasted amount on the previous day and the first increase / decrease parameter, The power transaction assistance device according to claim 3, characterized in that it does not bid.   The increase / decrease parameter determination unit adds the expected value of power sale income obtained when the bid amount for the market one day before and the current day market exceeds the actual demand amount to the expected value of the imbalance fee, The power transaction assistance device according to claim 3, wherein values of a first increase / decrease parameter and a value of the second increase / decrease parameter are respectively determined. A power transaction support method for supporting a bid for a wholesale power transaction market, comprising:
The previous day predicted amount prediction step of predicting the previous day predicted amount which is the predicted power demand amount at the time of bidding to the market one day ago based on the past power demand data;
On-the-day forecasting amount forecasting step of predicting a forecasting amount on the day, which is a forecasting power demand amount at the time of bidding on the day's market, based on past power demand data;
The previous day prediction error, which is a prediction error between the previous day predicted amount in the past and the corresponding past actual power demand amount, and the prediction error between the past predicted day amount in the past and the corresponding past actual power demand amount Prediction error calculation step of calculating the prediction error of the day,
The previous day market bidding step of correcting the previous day predicted amount based on the previous day prediction error and bidding on the previous day market with the bid amount determined based on the corrected previous day predicted amount;
Correcting the forecasted amount on the day based on the forecasted error on the day, and having a market biding step for bidding on the market on the day with the bid amount determined based on the forecasted amount on the day after the correction. Characteristic power transaction support method.   The power trading support method according to claim 6, wherein the previous day prediction error and the current day prediction error follow a distribution described by an equation. The first increase / decrease parameter for increasing / decreasing the previous day prediction amount based on the previous day prediction error and the current day prediction error, and the second increase / decrease parameter determination step for determining the second increase / decrease parameter for increasing / decreasing the previous day prediction amount Have
In the increase / decrease parameter determination step, the expected value of the bid price for the market one day ahead calculated based on the predicted amount of the previous day corrected by the first increase / decrease parameter, and the current day corrected by the second increase / decrease parameter The sum of the expected value of the bid amount for the current day market calculated based on the predicted amount and the expected value of the imbalance fee to be paid when the bid amount for the day before market and the current day market falls below the actual demand amount. The power transaction assistance method according to claim 6 or 7, wherein values of the first increase / decrease parameter and the second increase / decrease parameter are respectively determined to be minimum. In the one-day-ahead market bidding step, a bid of the sum of the previous day's predicted amount and the first increase / decrease parameter is made to the one-day-ahead market as a bid amount,
In the same-day market bidding step, when the sum of the predicted amount on the current day and the second increase / decrease parameter is greater than the sum of the predicted amount on the previous day and the first increase / decrease parameter, the difference is used as the bid amount. If a bid is made to the market on the day and the sum of the forecasted amount on the day and the second increase / decrease parameter is smaller than the sum of the forecasted amount on the previous day and the first increase / decrease parameter, The method according to claim 8, wherein the bid is not made.   In the increase / decrease parameter determination step, the expected value of the sale fee is added to the expected value of the imbalance fee, the expected value of the sale income obtained when the bid amount for the market one day before and the day market exceeds the actual demand amount. The power transaction assistance method according to claim 8, wherein values of a first increase / decrease parameter and a value of the second increase / decrease parameter are respectively determined.