JP2006050834A - Risk management support system for power supply enterprise - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To rationally predict electric energy to be reserved by procuring from an outside source at prediction point in time by which a power procurement cost becomes minimum or a revenue becomes maximum. <P>SOLUTION: A risk management support system for a power supply enterprise is applied to the power supply plan of a power enterprise. In the system, electricity is generated at in-house facilities or facilities of a contracted company, electric power is procured by purchasing the power procured from the outside source such as an outside industry or the like, and it is supplied to a customer. The system requires at least a data input means, a demand predicting means, a predicting means of probable generated energy, and a calculating means of the generated energy to be procured from the outside source. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a system for estimating business profit risk of a specific scale electric power operator (PPS).

  With the recent liberalization of power retailing, it has become possible to retail electricity without being a traditional power company, and a new specific-scale power operator (PPS) has emerged. The target of retail liberalization has spread to customers over 500 kilowatts since April 2004, and more than 50 kilowatts since April 2005, and many customers such as factories and commercial facilities offer prices and services. You can buy electricity from your favorite company.

  While electricity companies before liberalization supplied power to all customers in each jurisdiction, specific-scale power utilities that emerged as a result of liberalization had a limited total amount of electricity that could be supplied. Sign power contracts with only more customers. Then, from the viewpoint of the expected value of profit and the risk, the current business feasibility is evaluated, and the customer who makes the power transaction contract is determined. Therefore, it is required to more accurately estimate the expected value and variance of revenue when one or more specific customers are customers.

  In the first place, the profit of a specific scale electric power company is obtained by subtracting the power procurement cost, the power consignment cost, the sales cost, etc. from the income obtained by selling power to the consumer. Here, the power procurement cost will be described in detail. In general, PPS procures power mainly from power plants that have a power trading contract or from its own power plant, and from other power companies, wholesale power markets, and other specific power companies. In the text, for convenience, power other than contracted power plants or in-house power plants is called “externally procured power”. However, this is not a general name. Of the externally procured power, the power purchased from the power company is called backup power (hereinafter referred to as BU power), but BU power must be reserved in advance, for example, until 10 am the previous day. Yes. Even if the amount of power from the power plant + BU power amount + other externally procured power amount is less than the customer's "demand amount" and the supply is insufficient, the power company will supply more power than the reserved amount. Can be purchased from, but depending on the amount and time of supply shortage, you have to pay higher than usual. In addition, in the wholesale power market, it is possible to make a reservation purchase of power from the next day. For this reason, the power procurement cost depends on how much of the contracted power plant and externally procured power has been procured and how much of the externally procured power has been reserved by the previous day.

  Here, a description will be given of a method in which a specific-scale power company determines a reserved amount of power other than the contract power plant in actual daily work. First, the demand amount on the next day of the customer and the power generation possible amount on the next day of the company's own power plant or contracted power plant are predicted. However, the prediction here is to obtain a predicted value by some method. For example, the predicted value of the power generation capacity of the contracted power plant may be notified from the power plant. Then, the reserved amount of the externally procured power is determined based on the predicted value of the demand amount and the power generation possible amount.

  When deciding on the reservation amount, instead of reserving only the expected shortage amount, that is, the predicted value of demand-the predicted amount of power generation, the supply amount is insufficient with respect to the demand amount and pays a large amount of money to the power company To avoid this situation, we will make a little more reservations. However, depending on how many reservations are made, the power procurement cost (power procurement cost from the power plant + external procurement power procurement cost) varies greatly. Furthermore, customers who have difficulty in predicting demand must reserve a larger amount of externally procured power for safety. If you do not reserve a large amount of externally procured power, there is a high probability that supply shortages will occur, and you will have to pay a large amount of money to the power company. Therefore, even if the demand is the same, the total cost for procuring electric power varies depending on the ease of demand prediction.

  By the way, Patent Document 1 discloses a simulation system that calculates operating revenue and risk of electric power retail using power usage record data, power plant characteristic data, power plant contract condition data, and power company contract condition data of each consumer. Has been.

JP 2003-70164 A

  However, Patent Document 1 describes the data to be used, but does not describe a specific simulation method. In addition, since the demand prediction means and the externally procured power amount calculation means are not provided, the power procurement cost cannot be estimated accurately, and the operation profit of the power retail cannot be estimated with high accuracy.

This point will be described with a specific simple example. At a certain time on a certain day, the possible power generation amount of the power plant = 1100 kW, the demand amount = 1400 kW, the predicted power generation amount predicted on the previous day = 1000 kW, the predicted demand value predicted on the previous day = 1500 kW, the external reservation amount reserved on the previous day = 600 kW, procurement cost of externally procured power = 15 yen / kW, and power procurement cost of power plant = 10 yen / kW. Since the amount of the external reserved amount must be purchased, the power procurement cost at this time is covered by the externally procured power of 600 kW and the power of the 800 kW power plant for the demand of 1400 kW. Therefore, power procurement cost = 600 × 15 + 800 × 10 =
It is 17,000 yen. However, in a simulation system that does not include an externally procured power amount calculation means, it is difficult to know how much the external reservation amount is for a demand of 1400 kW. Thus, for example, suppose that a procurement method that minimizes the power procurement cost, that is, an externally procured power of 300 kW and an electric power from a 1100 kW power plant met the demand of 1400 kW. Then, power procurement cost = 300 x 15 + 1100 x 10 = 15,500 yen,
There is an error of 1,500 yen. In this case, it cannot be said that PPS revenue is estimated with high accuracy. Note that the numbers used as examples are different from the actual ones.

  The problem to be solved by the present invention is to reasonably predict an externally procured power reservation amount at which the power procurement cost is minimized or the profit is maximized at the time of reservation.

Electricity supply business risk management applied to the power supply plan of the electric power business that generates power from in-house facilities or contracted facilities and purchases externally procured power from outside companies, etc., and supplies it to consumers The support system needs the following means.
1. Data input means Demand forecasting means 3. Predictable power generation means 4. Externally procured power calculation method

  According to the present invention, an externally procured power reservation amount that minimizes the power procurement cost or maximizes the profit can be reasonably predicted at the time of reservation.

  Hereinafter, an example of the best mode for carrying out the present invention will be described with reference to the drawings. FIG. 1 shows a functional block diagram of the present invention. Data input means (101), demand prediction means (110), power generation possible quantity prediction means (111), externally procured power amount calculation means (106), power procurement cost calculation means (107), business revenue calculation means (108), It comprises business revenue display means (109). Further, the demand forecasting means (110) comprises a demand quantity expected value forecasting means (102) and a demand forecast error distribution calculating means (103), and the power generation possible quantity prediction means (111) is a power generation possible quantity expected value prediction means (104). And a power generation possible amount prediction error distribution calculating means (105).

  The present invention estimates the expected return and variance when the PPS has been operating for a certain period of time by these means, and displays the result. For example, when estimating expected profit and variance for one year, the profit for each year is calculated based on actual data for a plurality of years, the expected value and variance are calculated, and this is displayed as an estimated value. Alternatively, instead of the actual data for a plurality of years, for example, the profit for each year may be obtained based on 1000 sets of random numbers generated based on the actual data, and the expected value and variance may be calculated.

  From here, each means will be described in detail, taking as an example a case where expected return and variance are estimated and displayed based on 1000 sets of random numbers.

The data input means (101) includes a demand prediction means (110), a power generation possible quantity prediction means (111), an externally procured power amount calculation means (106), a power procurement cost calculation means (107), and a business revenue calculation means (108). The necessary information is input to the system database (hereinafter referred to as DB). In the demand prediction means (110), as shown in FIG. 2, a calendar (day of the week, holiday, etc.), a customer event (regular holiday, sale, opening / closing time change, inventory, etc.), weather (clear, cloudy, For example, a random number such as rain), past demand, maximum temperature and minimum temperature, for example, 1000 sets for one year are required. The power generation amount predicting means (111) requires information on the characteristics of the generator (such as the relationship between the temperature and the maximum output of the generator) and random numbers such as the temperature, for example, 1 year × 1000 pairs. In addition, when a part of the power generated by the power plant is used in-house, the same data as the demand prediction means (110) is also required to predict the demand. In the externally procured power amount calculating means (106) and the electric power procurement cost calculating means (107), information on the price of the externally procured power, for example, information on terms and conditions with the electric power company is necessary when procuring from the electric power company. In the case of an electric power company, the price system is somewhat complicated, for example, the price varies depending on the difference between the actual procurement amount and the reservation amount as of the previous day. The business profit calculation means (108) needs information on the wholesale power price for each customer and the consignment cost of power.

  Next, the demand amount expected value predicting means (102) will be described. The processing flow of (102) is shown in FIG. Various demand prediction models are stored in the demand prediction model group DB (301). As an example of the prediction model, a regression model is given first. The explanatory variables include quantitative variables such as maximum and minimum temperatures and past demand, customer events (regular holidays, sales, changes in opening and closing hours, inventory, etc.), and calendars (day of week, holidays, etc.) ) And weather (clear, cloudy, rain, etc.). That is, electric power demand = α × highest temperature of the previous day + β, electric power demand = α × highest temperature forecast value + β × highest temperature forecast squared + γ, and electric power demand = α × the same time of the previous day Various regression models are conceivable, such as demand amount + β × demand amount at the same time the previous day + γ. Parameters such as α and β may be obtained by the least square method. These regression models with different combinations of explanatory variables are distinguished as different models and stored in the DB (301). For an introduction to regression models, refer to Chapters 4 to 6 of “Introduction to Multivariate Analysis (Joint Nagata and Masahiko Munechika)” of Science.

  In addition to the regression model, a method that uses the average value of actual values over a certain period in the past as a predicted value, a method that uses actual values for similar days searched based on information such as weather and calendar, etc., and a neural network Methods are also examples of prediction models. As a reference for neural networks, there is “Nonlinear Multivariate Analysis (by Hideki Toyoda)” by Asakura Shoten.

In the past demand, weather information, calendar, and event DB (302), as shown in FIG. 4, each customer's past demand for each unit time (1 hour or 30 minutes), the location of the customer Stores weather information (temperature, weather, etc.), calendar (day of the week, holidays, etc.), and event information for each customer (regular holiday, sale, opening / closing time change, inventory, etc.). This is input by the data input means (101), and 1000 sets of data exist in the case of 1000 simulations.

  In the calculation of the predicted value (303), the predicted values for each unit time of each consumer based on the respective prediction models are calculated using the DBs (301) and (302). For example, when predicting the demand amount of a certain customer at 13:00 on March 1, using all prediction models stored in the DB (301) as shown in (401) of FIG. Calculate the predicted value for 13:00 a day. In addition, in the accuracy evaluation (305) of the prediction model, which will be described later, when the standard deviation of the prediction error in the past month is used as the evaluation method, for example, the predicted value for the past month is also calculated.

  The prediction value DB (304) stores the result (501) calculated in (303).

  In the accuracy evaluation (305) of the prediction model, the accuracy of each model is quantitatively evaluated. As an example of the evaluation method, there is a standard deviation of a prediction error or a mean value of an absolute value of the prediction error when daily prediction is performed for the past one month using each prediction model. There is also a method of evaluation using a contribution rate or AIC (Akaike's Information Criteria). For the contribution rate, see “Introduction to the Multivariate Analysis Method” (by Nagisa Nagata and Masahiko Munechika) P.16, and for AIC, “Nonlinear Time Series Analysis (Ikuo Matsuba)” by Asakura Shoten, P.44-P.47. Please refer to. In (502) shown as an example, the evaluation is based on the standard deviation of prediction errors for the past month.

  In selecting the best prediction model (306), the one that is determined to have the highest prediction accuracy from the result of (305) is also selected. In (502), the model 2 with the smallest standard deviation is selected.

In the prediction value confirmation (307), the prediction value based on the prediction model selected in (306) is searched from the prediction value DB (304), and the value is confirmed as the final prediction value. In the example of FIG.
Since model 2 is selected from (402), 110 kW of the predicted value on March 1 of (501) is set as the final predicted value.

  The demand prediction error distribution calculation means (103) calculates the probability distribution of the prediction error of the demand amount expected value prediction means (102). For example, an error empirical distribution is derived from the difference between the daily demand forecast value and the actual value in a certain period such as the past month, that is, the forecast error. Alternatively, some probability distribution may be assumed. That is, for example, assuming a normal distribution, only the expected value and standard deviation of the prediction error may be estimated.

  The possible power generation amount expected value predicting means (104) will be described with reference to FIG. Based on the generator characteristics DB (601) such as fluctuations in the amount of power that can be generated due to temperature and deterioration over time, weather information such as temperature and weather, and information DB (602) such as power plant events such as periodic inspections, The power generation and bleedable amount is predicted (603), and the predicted value is output (604). The DB (602) is input by the data input means (101).

  The power generation possible amount prediction error distribution calculating means (105) calculates the probability distribution of the prediction error of the power generation possible amount expected value prediction means (104). For example, as shown in FIG. 5, an error empirical distribution such as (502) is obtained from the difference between the daily demand forecast value and the actual value in a certain period such as the past month, that is, the prediction error (501). Alternatively, assuming that the prediction error follows a probability distribution such as a normal distribution, only the average value and standard deviation of the prediction error may be calculated to obtain a distribution such as (503).

  The externally procured power amount calculation means (106) includes a demand amount expected value prediction means (102), a demand prediction error distribution calculation means (103), a power generation amount expected value prediction means (104), and a power generation amount prediction error distribution calculation means. Using the result of (105), the amount of electric power procured from outside the contracted power plant such as an electric power company is calculated. The method will be described in detail with reference to FIG. FIG. 7 is a probability distribution of an insufficient supply amount of power. The expected value is given by demand predicted value-power generation possible amount predicted value, which is demand amount expected value predicting means (102), demand prediction error distribution calculating means (103), and power generation possible amount expected value predicting means (104). Ask more. The shape of the probability distribution is obtained from the demand prediction error distribution calculation means (103) and the power generation possible amount prediction error distribution calculation means (105). This may also be an empirical distribution, or a probability distribution such as a normal distribution may be assumed. The probability that the electric power of a certain amount b (kW / h) or more will be insufficiently supplied is given by the area of the hatched portion in FIG. Procurement of power from the power plant and externally procured power is set as a constraint condition, with the probability of a shortage of more than a certain amount in each unit time, or a probability of shortage of a certain amount over a certain amount of time. Calculate the amount of externally procured power that minimizes the sum of costs.

  The power procurement cost calculation means (107) calculates the total procurement cost of the power from the power plant and the externally procured power in the past fixed period based on the contract contents with the power plant and the power company.

  The business profit calculating means (108) calculates the profit for a certain period in the past from the power procurement cost, the power consignment cost, the labor cost, etc. calculated in (107) and the income obtained by selling the power.

  The business profit display means (109) outputs the result of the business profit calculation means (108) to a display device such as a CRT or TFT.

  Thus, the Example of this invention consists of the following means.

  The data input means includes a random number generated based on the past actual demand value or distribution of actual demand value, and a random number generated based on the past actual power generation amount actual value or distribution of actual power generation amount actual value. , Information necessary for demand forecasting (such as temperature and weather at the customer's location, past demand actual values and events, calendar information such as days of the week and holidays), and information necessary for forecasting power generation capacity ( The temperature of the power plant location, information on events such as periodic inspections of the generator, etc.) are input or retrieved from a database (hereinafter referred to as DB).

The demand prediction means includes demand quantity expected value prediction means and demand prediction error distribution calculation means. The demand amount expected value prediction means predicts the demand for electric power and steam using the information necessary for the above demand prediction,
Store in DB. The demand prediction error distribution calculation unit calculates the distribution of the prediction error of the demand amount expected value prediction unit and stores it in the DB.

  The possible power generation amount predicting means includes a possible power generation amount expected value prediction means and a possible power generation amount prediction error distribution calculation means. Here, the amount of power that can be generated refers to the maximum amount of power that the power plant can transmit to the PPS. That is, when a part of the power generated by the power plant is used in-house and the rest is transmitted to the PPS, the value is obtained by subtracting only the in-house use from the full output of the generator.

  The expected power generation amount prediction means can extract the amount of power generation possible from power plants and / or in-house power plants that have power or steam sales contracts using the information required for the above power generation forecasts. The steam amount is predicted and stored in the DB. The possible power generation amount prediction error distribution calculating means calculates the prediction error distribution of the possible power generation amount expected value prediction means and stores it in the DB.

  The externally procured power amount calculation means takes out the demand prediction value and its prediction error distribution and the power generation possible amount prediction value and its prediction error distribution from the DB, calculates the externally procured power amount based on them, and stores them in the DB. . The power procurement cost calculation means calculates the externally procured power cost by taking out the amount of externally procured power from the DB, and adds the cost required to procure from the power plant that has a business contract or its own power plant to the DB To store.

  The business profit calculating means takes out the result of the power procurement cost calculating means from the DB, calculates the expected profit and variance for a certain period, and stores them in the DB. The business profit display means takes out the result calculated by the business profit calculation means from the DB and displays it on the display device.

  As described above, according to the present invention, the externally procured power reservation amount that minimizes the power procurement cost or maximizes the profit is reasonably predicted at the time of reservation (ie, the demand amount on the next day is not known). be able to. In addition, it is possible to more accurately estimate the operating revenue by more accurately estimating the power procurement cost when a specific scale electric power company makes a power transaction contract with one or more specific customers.

  In the embodiment of the present invention, there is provided means for storing data received by the input means in a DB, which is hardware, and the calculation is performed by each means based on the data stored in the DB. In addition, it has a storage means for storing the calculation results of each means in the DB, and the calculation is again performed by each means based on the stored DB or displayed on the display device. That is, the present invention is achieved by cooperation of hardware and software.

The functional block diagram of one Example of this invention. Explanatory drawing of the data input means of one Example of this invention. The processing flow figure of the demand amount expected value prediction means of one Example of this invention. Explanatory drawing of the demand amount expected value prediction means of one Example of this invention. Explanatory drawing of the external procurement electric energy calculation means of one Example of this invention. Explanatory drawing of the electric power generation possible amount prediction error distribution calculation means of one Example of this invention. An example of a probability distribution of an insufficient amount of power supply.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 101 ... Data input means 102 ... Demand amount expected value prediction means 103 ... Demand prediction error distribution calculation means 104 ... Power generation expected amount prediction value prediction means 105 ... Power generation possible amount prediction error distribution calculation means 106 ... Externally procured electric power Quantitative calculation means 107, power procurement cost calculation means 108, business profit calculation means 109, business profit display means 110, demand prediction means 111, power generation possible quantity prediction means 201, demand quantity expected value prediction means DB explanatory diagram, 701... Probability density function of supply shortage.

Claims (10)

A power supply business risk management support system applied to a power supply plan of a power business that generates power by own equipment or contracted equipment, procures power by purchasing external power, and supplies it to consumers,
Data input means for receiving calendar, consumer events, weather, past demand or temperature data, and information on generator characteristics;
Storage means for storing data received by the data input means in a database;
Demand forecasting means for forecasting demand for electric power and steam from the calendar, consumer events, weather, past demand, or temperature data stored in the database;
Storage means for storing the demand prediction result predicted by the demand prediction means in a database;
Information relating to the characteristics of the generator stored in the database, and data on the temperature, power generation potential prediction means for predicting the power generation possible amount by the company equipment or contracted equipment,
Storage means for storing a power generation possible amount prediction result predicted by the power generation possible amount prediction means in a database;
An externally procured power amount calculating means for calculating an externally procured power amount to be procured based on the demand prediction result and the power generation potential prediction result stored in the database;
Power supply business risk management support system characterized by comprising: In claim 1,
Storage means for storing the externally procured power amount calculation result calculated by the externally procured power amount calculating means in a database;
Business revenue calculation means for calculating an expected value and risk of business revenue of a power supply plan based on the demand forecast result, the power generation potential forecast result and the externally procured power amount computation result stored in a database This is a power supply business risk management support system. In claim 1,
A power supply business risk management support system comprising: a demand amount expectation value prediction means for predicting an expected value of a demand amount; and a demand prediction error distribution calculation means for calculating a demand prediction error as the demand prediction means. In claim 3,
The demand forecast error calculated by the demand forecast error distribution calculation means is given as a probability distribution,
A power supply business risk management support system characterized by calculating an empirical probability distribution obtained from prediction history data or a probability distribution in which parameters of a given probability distribution function are determined using the prediction history data. In claim 3,
A power supply business risk management system for optimizing the amount of externally procured power under the probability distribution of demand in the planned time interval calculated from the probability distribution of demand prediction error by the demand prediction error distribution calculating means. In claim 1,
As the possible power generation amount predicting means, comprising: an expected power generation amount predicting value calculating means for calculating an expected value of the possible power generation amount; and a possible power generation amount prediction error calculating means for calculating a prediction error distribution of the possible power generation amount, Power supply business risk management support system. In claim 6,
The generation potential prediction error calculated by the power generation amount prediction error calculation means is given as a probability distribution, and the empirical probability distribution obtained from the prediction history data or the parameters of a given probability distribution function using the prediction history data A power supply business risk management support system characterized by being calculated as a determined probability distribution. In claim 6,
A power supply characterized by optimizing an externally procured power amount under a probability distribution of a possible power generation amount in a planned time interval calculated from a probability distribution of a possible power generation amount prediction error by the possible power generation amount prediction error calculating means Business risk management system. In any one of Claims 1 to 8,
A power supply business risk management support system, characterized in that the externally procured power amount calculation means calculates an externally procured power amount that maximizes the expected value of business revenue. In any one of Claims 1 to 8,
The power supply business risk management support system, wherein the externally procured power amount calculating means calculates an externally procured power amount that minimizes an expected value of the externally procured power cost.

Images