JP2009245044A - Power price predicting device and power price predicting method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To estimate a bid curve by estimating marginal cost curves, reserves and profits of other transaction participants. <P>SOLUTION: The marginal cost curves, reserves and profits of the other transaction participants are set on the basis of a contracted price of a power exchange and an actual value of an agreement amount, and the marginal cost curves, reserves and profits of the other transaction participants are updated so as to minimize differences among predicted past contracted price, agreement amount and actual values on the basis of bit curves of the other providers and the bid curve of one's company from the set marginal cost curves, reserves and profits to thereby calculate the most accurate estimated value. The estimated value of the marginal cost curves, reserves and profits of the other transaction participants are used to create a bid curve, so that future contracted price and agreement amount can be predicted. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention calculates the marginal cost curve, reserve capacity, and profits of other trading participants from the past power trading price of the power exchange and the actual amount of quantification, its own bidding curve, and other electricity demand forecasts of other trading participants. The present invention relates to an apparatus and a method for estimating and predicting a power price and a quantified amount of a power exchange by using the estimation result, future own bid curve and demand forecast value forecast of other trading participants.

  The Japan Wholesale Electric Power Exchange has disclosed the electricity price and approximate amount for each product to trading members. However, the bid curve information of each trading participant, whose price and approximate amount are determined, has not been released. However, if the bid curve and interconnection capacity of each trading participant is used, the contract processing method is publicly available, so the contract price and contract quantification can be determined. Since the information on the tender curve is not disclosed, it is difficult to predict future contract prices and contract quantification. Electricity prices are more fluctuating than other products in terms of domestic and international conditions. For this reason, for the trading participants, the profit of the trading department is greatly influenced by the power price of the power exchange and the value of the fixed amount.

  In the following Patent Document 1, a tender curve is created based on power supply characteristics, demand forecast, and the like, and a power price and a fixed amount are determined. Here, on the assumption that the power supply characteristics of all trading participants of the company and other companies are known, a power plant operation plan is created so as to satisfy the expected power demand, and a marginal cost curve is obtained. However, in a competitive environment, it is not possible to obtain data on power supply characteristics related to costs such as fuel cost characteristics of other companies. For this reason, fuel cost characteristics cannot be set appropriately.

  There is also a method for predicting future power prices using a neural network that inputs power factors such as the nearest power price that affects power prices into the forecast method used in power demand forecasting. Yes (Non-Patent Document 1). However, for the trading department, there is a gap between the pricing mechanism and the physical characteristics related to power generation costs, so the validity of the forecast results cannot be confirmed.

JP 2007-65954 A Utsumi, Sugimoto et al. “Study on Price Fluctuation Prediction in the Electric Power Market”, IEEJ Power Technology Study Group, PE-05, No.159-170, Page.63-67 (2005)

  As electric power liberalization progresses, electric power companies must promote business in a competitive environment. In particular, in the trading department of an electric power company, it is necessary to increase profits by buying and selling electricity, and it is important for profit improvement to be able to appropriately predict the contract price and contract amount at the power exchange. However, because of the competitive environment, it is not possible to obtain information on other electric utilities, especially past and future bidding curves. For this reason, based on the past contract prices and contracts published by the power exchange and the free capacity of the interconnection lines disclosed by the neutral institutions, the other companies ensure that the errors in the contract prices and contracts are minimized. The marginal cost curve, reserve capacity, and profit are estimated, and the other company's bid curves are created. Future contract prices and contract quantification can be accurately predicted from the other company's tender curves, in-house tender curves and interconnected line free space, by contract processing equivalent to exchanges.

  The present invention has been made in response to the above-mentioned problems, and the purpose of the present invention is to execute a contract processing based on the estimated bid curve of the other person, the bid curve of the company, and the free capacity of the interconnection line. An object of the present invention is to provide an apparatus and a method for accurately predicting future contract prices and contract quantification.

  The main feature of the present invention is to estimate the marginal cost curve, reserve power, and profit of other companies so that the error from the past contract price and contract quantification is minimized. A bid curve of other companies is created from these estimation results, combined with the company's bid curve, and considering the free capacity of the interconnection line, the contract price is determined by the power price forecasting device and method that executes the same contract processing as the exchange. And predict about quantification.

  According to the present invention, the marginal cost curve, reserve capacity, and profit of other companies are estimated so as to minimize the error from the past contract price and actual performance, and the other company's bid curve is created using these. By executing the contract with the company's own bidding curve, it is possible to accurately predict the power price and the fixed amount.

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

  FIG. 1 is a calculation processing flow by the power price prediction apparatus to which the present invention is applied, FIG. 2 is a schematic overall configuration diagram of the power price prediction apparatus 1, FIG. 3 is a detailed configuration diagram of the power price prediction apparatus, and FIG. 6 is a condition setting screen, FIG. 7 is an example of parameters of an optimization method (PSO: Particle Swarm Optimization) used for parameter estimation, and FIG. 8 is an explanation of updating variables to optimize the variables of the PSO. FIG. 9 and FIG. 9 are screens that display information on the estimated parameters. FIG. 10 shows a curve in which selling bids and buying bids of all trading participants are aggregated, and is a diagram for explaining the contract price and contract quantification. FIG. 11 to FIG. 14 are diagrams for explaining processing when parameters are estimated. FIG. 15 shows a revenue calculation screen.

  The power price prediction apparatus 1 includes a central processing unit (CPU) 2, a main storage device 3, an input / output device 4, and an external storage device 5.

  FIG. 3 is a detailed functional configuration diagram of the power price prediction apparatus 1 according to the present invention. The power price prediction apparatus 1 includes the above-described input device 6, display device 7, database 9 and reading device 10, and a calculation processing unit 20. The calculation processing unit 20 is realized by the CPU 2 executing a program that is stored in advance in the storage medium of the external storage device 5 and is read into the main storage device 3 via the reading device 10. Is not limited to such a programmed general purpose processor. For example, the calculation processing unit 20 can be configured by a combination with a specific hardware device including a wired logic that executes each processing of the present invention.

  The input / output device 4 includes an input device 6 having a keyboard and a mouse shown in FIG. 3 and a display device 7 as an output device. The input / output device 4 may be provided with an input device such as a pointing device or a touch sensor, or an output device such as a liquid crystal display device, a printer, or a speaker instead of or in combination with these. The external storage device 5 includes a hard disk device, a floppy disk device, a CD-ROM (compact disc-read only memory) device, a DAT (digital videotape) device, a RAM (random access memory) device, a DVD (digital video disc) device, A nonvolatile memory or the like can be used. The external storage device 5 includes a mass storage device for holding the database 9 shown in FIG. 3, a storage medium for holding a processing program, and a reading device 10 for reading information held in the storage medium. Although used, both the database and the processing program can be held in one external storage device. As the storage medium, a floppy disk, CD-ROM, magnetic tape, optical disk, magneto-optical disk, DAT, RAM, DVD, nonvolatile memory, or the like can be used.

  The input device 6 accepts selection of options displayed on the display device 7, data input, and the like, and transmits them to the calculation processing unit 20. The display device 7 displays the data sent from the input device 6. The calculation processing unit 20 uses the data transmitted from the input device 6, the data read from the database 9, and the processing program read from the reading device 10 to support the electric power transaction for the bid curve of the company with the highest profit. Transmit to system 11. In the power trading support system 11, if necessary, the operator corrects the bid curve as necessary, and transmits the bid curve to the power exchange system 12. The power exchange system 12 transmits the contract result to the power trading support system 11.

  The processing result of the calculation processing unit 20 is sent to the display device 7 for display and stored in the database 9.

  The calculation processing unit 20 includes a control unit 21, a condition setting unit 22, a parameter setting unit 23, a contract processing unit 24, a result output unit 25, and a profit calculation unit 26. The calculation processing unit 20 is connected to the power trading support system 11 via the communication line 31, and the power trading support system 11 is connected to the power exchange system 12.

  The control unit 21 smoothly exchanges data, processing programs, and the like between the input device 6, the display device 7, the printing device 8, the database 9, the reading device 10, the power transaction support system 11, and the processing units 22 to 26. The data is processed and processed, and the exchange is controlled so that the entire process operates normally.

  The condition setting unit 22 is stored in the storage medium of the database 9 and / or the reading device 10 via the control unit 21. The past contract price, the contract amount, the free space on the interconnection line, the meteorological data such as the maximum temperature, the demand The actual values, demand forecast values, power outage plans, the company's tender curve results, the trade participant's area and operator type default conditions, etc. are read and displayed, and these conditions are displayed on the display device 7, Change the condition. Display device for forecast date, data usage period, evaluation function, inter-area connection relationship, initial value of marginal cost curve, upper and lower limit values, trading power, reserve power, selling profit, initial value such as buying profit, upper and lower limit values 7 is displayed and a value is set. The set conditions are stored in the database 9 via the control unit 21.

  The parameter estimation unit 23 reads the data held in the database 9 via the control unit 21, determines the most appropriate parameter, and further creates a bid curve. The determined parameter values and the created bid curve are transmitted to the processing units 24 to 26 via the control unit 21 and stored in the database 9.

  The contract processing unit 24 uses the bid curve created by the parameter estimation unit and the value read by the condition setting unit 22 and the set value to simulate the contract processing of the power exchange system to determine the power price and contract quantification. To do.

  The result output unit 25 causes the display device 7 to display the parameter value estimated by the parameter estimation unit 23, the contract price calculated by the contract processing 24, and the contract quantitative amount via the control unit 21. Alternatively, they are printed on the printing device 8.

  The profit calculation unit 26 displays the bid curve of the trading participant or the marginal cost curve estimated by the parameter estimation unit 23 on the display device 7 through the control unit 21 and corrects the parameters as necessary. Then, using the corrected value, the result of the contract processing by the contract processing unit 24 is displayed on the display device 7 or printed on the printing device 8.

  Next, the operation will be described with reference to the processing flowchart shown in FIG. 1 and FIGS. First, setting of conditions for estimating the marginal cost curve in step S102 will be described.

  When the power price prediction apparatus 1 starts the power price prediction process, a screen for setting calculation conditions or reading from the database 9 is displayed on the display device 7 (not shown). The screen transitions from this screen to the screens of FIGS. 4, 5, and 6. G101 in FIG. 4 is a screen for setting a prediction target date and time, a use period of data used for estimating parameters, a day type used in the period, and a use target time. Numerical values are directly input for the prediction target date and the data period used to estimate the parameters. The day type of the usage target used during the period is a toggle switch, and one of them is selected. In addition, the evaluation function is the sum of errors between the predicted execution price and the execution quantity estimated from the past execution price and execution quantity and the estimated parameters. Here, the smaller the evaluation function, the better. Since there are two types of error, a contract price and a contract amount, the values of the weighting factors k1 and k2 are set. Further, a weighting coefficient cj is set corresponding to the day of the week for error evaluation. For example, in the example of FIG. 4, the weighting factor for the weekday price is 1.0, the weighting factor for the price on Saturday is 0.5, the weighting factor for the fixed amount on weekdays is 0.1, and the weighting factor for the fixed amount on Saturday is It becomes 0.05. Use the corresponding two weighting factors. The smaller the error is, the better the accuracy of the estimated parameter is. Therefore, if the error is small, the evaluation function may be set to be small. Therefore, the absolute value of the error may be used instead of the square of the error shown in FIG. By using the contract price and contract price announced to trading participants on the power exchange, errors other than those shown here can be calculated and incorporated into the evaluation function. For example, at the Japan Wholesale Power Exchange, the contract price and contract price when there is sufficient interconnected line capacity, the contract price of the area group that is the same price when considering complaints and handling the interconnected line capacity. Selling and buying contracts are published. An evaluation function can also be set by multiplying the error between the published actual value and the predicted value corresponding thereto by multiplying each error coefficient.

  G102 in FIG. 5 is a screen for setting the connection relationship of the bid area. Inter-area connections indicate that there are interconnected lines between the areas. Since these interconnected lines have the capacity to physically conduct electricity, the trading participants can Depending on whether you buy or sell, the price of electricity and the fixed amount will change. Since the upper right and lower left of the inter-area connection table C121 are symmetrical, it is only necessary to input the lower left part. 0 in the table indicates that the two areas are not connected, and 1 indicates that they are connected. In order to confirm the set result, the connection relationship between the area numbers is displayed by pressing a confirmation button. Since the actual inter-area connection does not change every day, it can be read from the database 9. A screen G103 in FIG. 6 is a screen for setting initial values, upper and lower limit values, and the like of parameters. C131 to C133 are values set to specify the business operator. C131 is an area number, C132 is an operator number, and C133 is an operator type. The business type includes “0” representing the company, “1” representing the general electric power company, “2” representing the PPS business, “3” representing the power business, “4”, and the like. C134 is a table for setting reserve power, selling profit, and buying profit, which are one of the parameters of trading conditions. In this example, reserve capacity is set as a percentage of the estimated demand value set separately, and a large supply capacity is set so that the power supply capacity does not become insufficient due to a generator accident or a forecast error in predicted demand. It is for securing. The reserve power may be specified by an absolute value MW. Here, the initial value is set so that it can be used as the initial value of the first iterative calculation when the parameter is estimated. Moreover, the upper limit and the lower limit indicate a range to be searched when estimating the reserve capacity. A numerical value of “variable / fixed” indicates a flag, and “1” indicates variable, indicating that the parameter is searched for within the upper and lower limits. On the other hand, “0” indicates that the initial value is the estimated value of the parameter and is not a search target. The selling profit is the amount added to the cost curve when selling electricity with respect to the marginal cost curve, and is the profit from the power sale. In the example of FIG. 6, the unit is an amount per 1 kWh, but it may be a ratio (%) to the marginal cost curve. The lower limit of this value is set to be equal to or greater than the transaction fee at the power exchange. Similarly, the buying profit is the amount that is deducted from the cost curve when buying electricity with respect to the marginal cost curve, which is the cost reduction due to power purchase. In other words, if it is cheaper than the unit price of power generated by the owned generator and can be purchased from an exchange, the power generation cost can be reduced by suppressing the output of the generator and purchasing from the exchange. In the example of FIG. 6, the unit is an amount per 1 kWh, but it may be a ratio (%) to the marginal cost curve. The lower limit of this value is set to be equal to or greater than the transaction fee at the power exchange. The marginal cost curve setting C135 is for expressing the marginal cost curve in a piecewise linear manner. The initial value of the cost is set so that it can be used as the initial value of the repeated calculation, as in the case of the trading conditions. Parameters are also set for the upper and lower cost limits. This is a range to search for the cost of each electric energy in the marginal cost curve. The details of the power sources of other companies are unknown, but since the capacity of each power source and the fuel type to be used are known, the upper and lower limits can be set. The amount of power is set in consideration of the characteristics of the generator. The amount of power increases from the left to the right of the table, and the marginal cost increases at the same time. This is because the marginal cost increases as the amount of electric power increases in order to operate sequentially from a power source with a lower unit price of power generation. When the initial value of the marginal cost and the electric energy are set in C135 and then the graph is confirmed, the graph is displayed in C136 and can be confirmed by pressing the C137 button. Here, since the operator type is “0”, that is, it is not necessary to search for the company, the initial value of C134 is set to the value used by the company, and the “variable / fixed” flag is set to “0”. ”Can be fixed. Regarding the cost of the marginal cost curve, the parameters can be fixed by setting the initial value, the upper limit value, and the lower limit value of the cost for each power amount to the same value. In C135, the amount of electric power is not a parameter to be estimated, but the electric energy can be searched by setting a search range. These parameters are set for each company. Although the demand forecast value can be set, the demand forecast value of the company can be estimated by using the actual demand and weather data, for example, by obtaining the regression coefficient by regression analysis and inputting the weather forecast data. Yes, this expected value can be used. Alternatively, when predicted values predicted in the past are stored in the database 9, they can be read from the database 9 and set. When forecasting the demand of other operators, for example, predicting the demand of others by expanding and contracting a certain constant on the value obtained by putting the weather forecast data of the other company's area into the company's demand forecasting model Can do. Alternatively, when the predicted value is read from the database 9, the demand of the other person can be predicted by expanding and contracting by applying a certain constant. Here, as a certain constant, the ratio of the installed capacity between the company and other companies, or the ratio of the maximum demand that has been announced can be used.

  In step S102, the parameter is estimated using the data set or input in step S101. Here, an example using PSO (Particle Swarm Optimization) will be described as an optimization method for estimating a parameter that is a real variable. Since it is a non-linear problem, other methods such as EP (Evolutionary Programming) that optimizes real variables, or tab search methods and genetic algorithms that can be applied to combinatorial optimization problems with discrete parameters Can do. First, the principle of PSO will be briefly described.

  PSO is mainly composed of two concepts. One is a general idea of artificial life, a school of birds, a school of fish, and the other is a concept based on the evolution of genetic algorithms (GA). The PSO that combines these two concepts is a very simple algorithm and can solve the nonlinear optimization problem of continuous variables at high speed, even though only basic arithmetic operations are used.

  The agent corresponding to the bird or fish constituting the flock is called an agent, and each agent is a trading participant parameter yi shown in FIG. 7 (a piecewise linear point representing a trading participant's marginal cost curve, reserve power, selling profit) , A vector consisting of buying profit) is connected as a vector for the trading participants as shown in equation (1), and has information on the position (state quantity x) and speed V for determining this.

(Equation 1)
x = (y1, y2, y3,...)

  The next position of each agent is updated using the position and speed of each agent and the position and speed of the agent with the highest evaluation among the group. Each agent acts so as to optimize the set evaluation function for minimizing the variable for the entire group. Since the evaluation function f (x) is uniquely determined by the state quantity x, each agent can calculate the evaluation function at the current position. Each agent uses the current speed, the best value f (pbest) of the evaluation function in the group of current agents, its position pbest, and the optimum value f (gbest) and the position gbest searched so far, The speed is corrected according to equation (2) so as to approach pbest and gbest. As a result, the current position of each agent is updated by equation (3).

(Equation 2)
V _i ^{k + 1} = W × V _i ^k + C ₁ × rand × (pbest _i −x _i ^k ) + C ₂ × rand × (gbest−x _i ^k )

(Equation 3)
x _i ^{k + 1} = x _i ^k + V _i ^k

Where i is the agent number, x _i is the position of each agent, V _i is the speed of each agent, rand is a uniform random number between 0 and 1, k is the number of position updates, and W, C ₁ and C ₂ are constants. Indicates.
The concept of the PSO search method is shown in FIG.
The execution procedure of PSO is as follows.

Step 1: Initialize agent position x _i and velocity V _i with random numbers. The initial value is set using a random number so as to be a value between the upper limit value and the lower limit value of G103.
Step 2: Evaluate the evaluation function f (x _i ) to be minimized.
Step 3: Compare with the agent's previous best value f (pbest _i ). If current f (x _i ) <f (pbest _i ), then f (pbest _i ) = current f (x _i ), and pbest _i = current position x _i in the search space.
Step 4: Compare with the best value f (gbest) of the entire group. If current f (x _i ) <f (gbest), then gbest = x _i .
Step 5: Update the speed using equation (2) and update the position using equation (3).
Step 6: Return to Step 2 and repeat until the convergence condition (maximum repetition count) is satisfied.

  Through the above processing, it is possible to estimate a parameter that can minimize the evaluation function and reduce the error in the contract price and the contract quantification.

  Next, a method for creating a bid curve from the marginal cost curve, reserve power, selling profit and buying profit will be described in order with reference to FIGS. Here, it demonstrates using the example of a general electric utility with the generator and the demand which should be supplied. FIG. 9 shows a cost limit curve of a trading participant who is a general electric utility. This curve is, for example, the initial value set in C135 of FIG. The horizontal axis is power demand (power consumption), and the vertical axis is marginal cost. The past power demand forecast value set in the condition setting in step S101 is the amount of power to be supplied as the company's demand. It is necessary to secure the supply capacity by adding reserve power to this electricity demand forecast value as the company's own. In other words, it is the portion that exceeds the reserve capacity that can sell power to the exchange. Here, when there is a planned power supply stop in the condition setting of the process 101, it is assumed that the demand demand value has virtually increased by that amount as necessary. Next, as shown in FIG. 10, a selling bid curve can be created at a price obtained by adding the selling profit to the marginal cost of the power sellable amount portion exceeding the reserve capacity. Since it is assumed that the demand below the demand forecast value should be supplied, the power generation cost can be reduced if power can be purchased at the exchange at a price lower than the marginal cost up to the forecast value. Therefore, the buying bid curve is a curve obtained by subtracting the selling profit from the marginal cost curve and is symmetric with respect to the expected demand line. Finally, the bidding curve of FIG. 11 can be created by translating the starting point of the selling bid curve and the buying bidding curve of FIG. 10 to the origin. As described above, if the marginal cost curve, reserve power, selling profit, buying profit, and power outage plan are determined, the selling bid curve and buying bid curve of the trading participant shown in FIG. 11 can be uniquely determined. . A process for creating a bid curve from this marginal cost curve is determined only for other companies other than the company. For the company's own portion, a bid is actually made in the past, and the curve is read from the database 9 in step S101 and used as it is. As described above, it is possible to create a bid curve for all trading participants.

  Next, in order to estimate the parameters, the contract price and contract quantification are calculated by contract processing, the error from the actual value is calculated, and the evaluation function is calculated. This contract processing simulates the contract processing of the power exchange system. Here, for the sake of simplicity, description will be made with an example in which there is sufficient interconnection line free space. That is, this is an example when the power price is the same in all areas. The selling bid curve and buying bid curve in FIG. 12 are obtained by integrating the selling and buying bid curves of all trading participants. In the power exchange system, the intersection of the two curves is the point where the social surplus of the trading participant is maximized, and this point is the contract price and the contract amount. It is possible to calculate the value of the evaluation function of Expression (1) by calculating respective errors of the contract price and contract price calculated by the contract processing from the estimated bid curve and the actual contract price and contract price.

  By calculating the evaluation function for each agent, it is possible to determine the agent state quantity x to be used in the next iterative calculation. If the agent's state quantity value exceeds the upper and lower limits, it is corrected to the upper limit or lower limit. This is repeated as many times as the maximum number of repetitions, and the state quantity x obtained when the evaluation function is the smallest is set as the optimum parameter estimation value. FIG. 13 is a screen G104 showing the result of estimating the optimum parameters of a certain trading participant. C141 displays an area number, C142 displays a company number, and C143 displays a company type. C144 shows the estimation result of the buying and selling conditions, and is the value of the parameter estimated by the row of estimated values. For the upper limit value, lower limit value, and variable / fixed row value, the values set in C134 of the screen G103 in FIG. 6 are displayed as they are. C145 is a marginal cost curve, and the estimated cost line displays the result estimated by the parameter estimation process. For the upper limit, lower limit, and electric energy of the cost, the values set in C135 of the screen G103 in FIG. 6 are displayed as they are. C146 is a graph displaying the relationship between the estimated cost and the electric energy. A screen G105 in FIG. 14 shows the best agent evaluation function, price prediction error, and about quantitative prediction error obtained at the number of repetitions of writing when the estimation of optimal parameters is repeated using PSO. ing.

  The Japan Wholesale Electric Power Exchange announces the contract price for each product group, each area group that collects areas where there is no congestion. In addition, the sell contract amount and the buy contract amount of the area group that collects the areas where the connection line congestion does not occur are disclosed. It also announces the contract price and contract amount when there is enough free space on the interconnection. Therefore, as the error target used in the evaluation function, it is possible to use the contract price and contract amount when there is sufficient interconnection line capacity, the contract price and sell contract quantity for each area group, and buy contract quantity. For this reason, when forecasting the contract price and contract quantity on the Japan Wholesale Power Exchange, these results can be used, so these are included as errors used in the evaluation function and are shown in G101 of FIG. Further, it is possible to display a screen for setting the respective weighting factors in the setting items, and to use the set data.

  As described above, the most accurate parameters of the marginal cost curve, reserve power, selling profit, buying profit, and selling and buying bid curves of other companies can be estimated.

  In process S103, the marginal cost curve estimated using past performance data, reserve power, selling profit, buying profit and demand forecast value, bid curve and interconnection free capacity of each trading participant created using the power outage plan By using the company's own bid curve, the contract price and contract amount of each area are predicted.

  Finally, in process S105, the profit when the bid curve is changed is calculated. FIG. 15 shows a screen for setting conditions for revenue calculation and displaying the result of revenue calculation. C161 indicates an area number, C162 indicates an operator number, and C163 indicates an operator type. When these items are set and the operator confirmation button of C164 is pressed, the marginal cost curve of C165 and the current bid curve of C166 are displayed, and the fixed amount of the operator of C167, the contract price and the profit or purchase from the power sale Displays the reduction costs due to electricity. As the default corrected bid curve, the same value as the current bid curve is displayed in C169. In this example, since C163 is “0”, the company's marginal cost curve, current bid curve, profit, etc. are shown. In the case of the company, the actual marginal cost curve and the current bid curve of the company read from the database 9 are displayed instead of the estimated parameters. In the case of other companies, the estimated marginal cost curve and the current tender curve are displayed. Next, when the corrected bid curve of C169 is corrected and C168 is calculated, the fixed amount, contract price, profit and profit changed with respect to the current bid curve are displayed in C170. In the bid curve evaluation, when the bid curve is changed by changing the bid price on both sides of the contract price, the profit, the contract price, or the change in the contract price when the bid curve is changed by changing the bid volume on both sides of the contract price Is calculated and displayed on C174 and C175. C174 and C175 in this example are the results when the profit calculation button of C171 is pressed, and C174 displays the change in profit when the bid price is changed and the bid price when the profit is maximized. . Similarly, C175 displays the change in profit when the electric energy is changed and the electric energy when the profit becomes maximum. In this example, the contract price of C167 is 11.50 yen, and the bid points on both sides of the current bid curve are (11.00 yen, 700 MWh) and (12.00 yen, 800 MWh), Two curves are displayed in C174 when the former bid price is changed to (10.00 to 12.00 yen, 700 MWh) and the latter bid price is changed to (11.00 to 13.00 yen, 800 MWh). Yes. As the bid curve increases from the left to the right of the table, both the bid price and the amount of power increase, so the range in which the price is changed is as described above. Similarly, the approximate amount of C167 is 750 MWh, and the bid points on both sides are (11.00 yen, 700 MWh) and (12.00 yen, 800 MWh), so the former electric energy is (11.00). Circles, 600 to 800 MWh), and (12 circles, 700 to 900 MWh), two curves are displayed. Since the bidding curve increases from the left to the right of the table, both the bid price and the amount of power increase, so the range in which the amount of power is changed is as described above. Also, by changing the area number, company number, and company type, the profits of other companies or the contract price and contract quantity of other companies when other companies change the bid curve are calculated. Can also be displayed. Although not shown in the figure, it is possible to calculate the company's profit, contract price, and contract quantity when another company changes the bid curve, so these results can also be displayed. The person in charge of the trading department confirms these evaluation results displayed on the display device 7 and sends a bid curve that maximizes the profit of the company to the power trading support system 11. Here, the person in charge of the trading department corrects the bid curve as necessary and bids on the power exchange system.

  According to the above embodiment of the present invention, the marginal cost curve, reserve power, selling profit, and buying profit parameters of other companies can be accurately estimated from the past contract price, contract quantification, and own bid curve. The contract price and contract quantity can be predicted from the other company's bid curve and the company's bid curve.

Electricity price forecast processing flow. Overall configuration of the electricity price prediction device. Detailed configuration of power price prediction device. An example of a condition setting screen. An example of an area connection setting screen. Example of parameter setting screen. Example of operator variables with agent state variables. How to update the state variable of Particle Swarm Optimization is explained. Example of marginal cost curve. The figure explaining the determination method of a bid curve. Bidding curve for trading participants. A curve that aggregates the bid curves of trading participants. An example of an estimation result display screen. The example of the display screen of evaluation function transition. Example of revenue calculation condition setting and revenue calculation result display screen.

Explanation of symbols

1 Price prediction device 2 Central processing unit CPU
3 Main storage device 4 Input / output device 5 External storage device 6 Input device 7 Display device 8 Printing device 9 Database 10 Reading device 11 Power transaction support system 12 Power exchange system 20 Price prediction processing unit 21 Control unit 22 Condition setting unit 23 Parameters Estimator 24 Contract processor 25 Result output unit 26 Revenue calculator

Claims (6)

Bid curve of past trading participants, actual price of electricity price and about fixed amount, power outage plan, interconnection capacity and connection between areas, predicted power demand, evaluation function and future power outage plan, Means for setting a free line capacity and a bid curve of the trading participant;
Using the trading participant's bidding curve, past power price, actual value of about fixed quantity, power outage plan, interconnection capacity and inter-area connection relationship, marginal cost curve of other trading participants, reserve power, Means to estimate selling profit, buying profit,
Execution processing means;
A power price predicting device comprising means for outputting an estimated marginal cost curve, reserve power, selling profit, and buying profit.   2. The power price prediction apparatus according to claim 1, wherein the setting means includes means for setting a marginal cost curve and a revised bid curve, and comprises a profit calculation means of the company or another company.   In the means for setting the power demand prediction value of the power price prediction device according to claim 1, the power demand prediction value predicted based on the past demand record and weather record of the transaction participant is taken in and set, and the power A power price prediction apparatus, characterized in that it is a means for fetching and setting a prediction value from a means for predicting other trading participants based on a demand prediction value.   In the means for estimating marginal cost curve, reserve power, selling profit, buying profit of trading participants of the electricity price prediction apparatus according to claim 1, an optimization method, a method for optimizing a real variable, a tab search method, a genetic algorithm Or, using the neighborhood search method, the estimated value and actual value of the power price when there is sufficient interconnection capacity, the estimated value and actual value of about fixed amount, or the predicted value and actual value of the power price for each area, actual value Quantitative sales of the area group with the same price and sales quantification of the area group, purchase quantification of the area group with the same price in the actual results, purchase quantification of the area group, and interconnections changed by power trading An evaluation function that includes at least one error of the expected current calculated from the actual value of the line flow and the contracted quantification, weighted to the error by day of the week, time, contract price, and contracted quantification. Electricity price prediction apparatus according to claim Rukoto. Bidding curves of past trading participants, actual values of power prices and fixed amounts, power outage plans, interconnection capacity and connection between areas, predicted power demand, evaluation function and future power outage plans, Means for setting a free line capacity and a bid curve of the trading participant;
The transaction participant's marginal cost curve, reserve power, and selling profit using the tender curve of the trading participant, the past power price and the actual value of the fixed amount, the power outage plan, the interconnection capacity, and the connection relationship between the areas. , Means to estimate buying profits;
Execution processing means;
A power price predicting device comprising means for outputting an estimated marginal cost curve, reserve power, selling profit, and buying profit.   By using the past power outage plan, interconnection free space capacity, connection relations between areas, and forecasted demand for electricity, and a bid curve of a certain trading participant, The marginal cost curve, reserve power, selling profit and buying profit of other trading participants that minimize the evaluation function using the error from the actual value are determined, and the marginal cost curve and reserve of the other trading participants thus determined are determined. Using the trading participant's bid curve and the certain trading participant's bid curve, demand forecast value of each future trading participant, interconnection free space, and power outage plan created from power, selling profit, buying profit A method for predicting electricity prices, characterized by predicting future electricity prices and their quantification.

Images