JP6472615B2 - Power trading support device, power trading system, control method and control program - Google Patents

Description

  Embodiments described herein relate generally to a power trading support device, a power trading system, a control method, and a control program.

Conventionally, when power generated by renewable means such as solar power generation or wind power generation is sold in the market, it is usually necessary to sell it at the same time as power generation, and it does not always match the market demand.
For this reason, there was a case where it was forced to sell at a cheap price.
Therefore, it is conceivable that electric power can be sold in a time zone where the market demand is high and the price is high by providing power storage equipment (power storage device) such as a storage battery.

In addition, as a means for effectively utilizing the generated power, there is self-consumption in which the seller himself consumes power with his / her equipment without selling the power.
If there is flexibility in the manner of consumption of the generated power in this way, self-consumption is performed during the time when the power selling price is low, and profit is increased by selling power to the system during the time when the power selling price is high be able to.

  Therefore, when there is an option other than selling the generated power on the spot, such as a power storage facility or a facility that consumes power by itself, it is desirable to create a power sale plan that enables more optimal power transactions.

JP 2012-053721 A

  However, renewable energy such as solar power generation and wind power generation greatly depends on weather conditions.

On the other hand, in the market, it is necessary to bid for the amount of power sold at each time zone at the previous day, and the amount of power sold needs to be determined based on some prediction.
By the way, there is uncertainty in weather prediction, and it is desirable that a power sale plan (electric power transaction plan) be created in consideration of this uncertainty.

  The present invention has been made in view of the above, and an electric power transaction support device, an electric power transaction system, and a control method capable of creating a more preferable electric power transaction plan in consideration of uncertainty of weather prediction And to provide a control program.

The power trading support apparatus according to the embodiment has facilities capable of temporal adjustment with respect to storage or use of power, and sets a power sale schedule for a power system of a power generation system that generates power using renewable energy.
Then, the weather prediction unit of the power trading support device performs ensemble prediction of the weather and generates a plurality of ensemble members.
The power selling price prediction unit predicts the spot power selling price in the spot market on the next day based on the ensemble member.
The market electricity sales price prediction unit predicts the market electricity sales price in the market of the next day based on the ensemble members.
The market power purchase price prediction unit predicts the market power purchase price in the market market the next day based on the ensemble members.
The power generation amount prediction unit predicts the power generation amount of the power generation system on the next day based on the ensemble member.
As a result, the electricity sales volume determination unit maximizes the expected profit based on the predicted spot power sale price for the next day, the next day's market power sale price, the next day's market power purchase price, and the next day's power generation system. Create a power sale schedule.
Here, in the power sale amount determination unit, the constraint condition creation unit creates a constraint condition that restricts the power sale amount based on the power generation amount of the power generation system predicted by the power generation amount prediction unit. Based on the electricity price, the assumed power selling price is set, the objective function creation unit creates the objective function for calculating the profit at the time of power sale, and the mathematical plan solution creation unit Solve a mathematical programming problem set by an objective function and constraint conditions applying the power selling price and the assumed power selling price, and create a power selling schedule that maximizes the expected profit.

FIG. 1 is a schematic configuration block diagram of a power system according to an embodiment. FIG. 2 is a block diagram of a schematic configuration of the power sale schedule creation device. FIG. 3 is a schematic configuration diagram of the weather prediction unit. FIG. 4 is an operation flowchart of the power sale schedule creation device. FIG. 5 is an operation flowchart of the power sale amount determination unit. FIG. 6 is a diagram for explaining the variables used in the constraint condition creation unit and their relationship. FIG. 7 is an operation explanatory diagram of the bid creation unit. FIG. 8 is a schematic configuration diagram of a power sale price prediction unit according to the second embodiment. FIG. 9 is a schematic configuration diagram of the power sale price history DB. FIG. 10 is a schematic configuration diagram of the weather condition history DB. FIG. 11 is a schematic configuration diagram of the weather prediction unit of the third embodiment. FIG. 12 is an operation flowchart of the third embodiment.

Next, embodiments will be described in detail with reference to the drawings.
[1] First Embodiment FIG. 1 is a schematic configuration block diagram of a power system according to an embodiment.
The power system 10 is roughly divided into a power generation / consumption system 11 that generates and consumes power, and a power sale schedule creation device 13 that is connected to an external power market MCT via a communication network 12 and creates a power sale schedule. And a power management unit 14 that manages the power generation / consumption system 11 in accordance with the power sale schedule created by the power sale schedule creation device 13.

  The power generation / consumption system 11 includes a power generation module 21 that generates power using renewable energy, such as a solar power generation device, a wind power generation device, a geothermal power generation device, and a wave power generation device, and a power storage unit 22 that includes a storage battery and stores power. A power selling unit 23 for performing an interface operation for selling the power generated by the power generation module 21 or the power stored in the power storage unit 22 to the power system PW, and for purchasing power from the power system PW. A power supply unit 24 that performs an interface operation, power generated by the power generation module 21, power stored in the power storage unit 22, or power purchased from the power system PW via the power purchase unit 24, and a mechanical device (not shown) And a power consuming unit 25 that consumes power by loads such as lighting and air conditioning equipment.

In the above configuration, the power generation module 21 generates electric power from renewable energy such as solar energy and wind energy, such as a solar power generation panel and a wind turbine for wind power generation.
The power storage unit 22 is configured as a device that stores power such as a storage battery and can be taken out when necessary.

FIG. 2 is a block diagram of a schematic configuration of the power sale schedule creation device.
The power sale schedule creation device 13 is roughly divided into a weather prediction unit 31 that performs weather prediction based on the input basic weather information, a power generation amount prediction unit 32 that performs power generation prediction based on the weather prediction, An SOC initial value acquisition unit 33 that captures an initial value of stored electric power stored in the storage unit 22, a price prediction unit 34 that predicts a power selling price, a marketed power purchase price, a marketed power selling price, and a next day price, and a power generation amount A power sale amount determination unit 35 that determines a power sale amount based on the prediction result, the obtained SOC initial value, and the price prediction result.

  The weather prediction unit 31 predicts the next day's weather conditions by a numerical weather simulation. In order to quantify the reliability of the forecast, the weather forecasting unit performs ensemble forecasting that creates multiple weather paths. The weather prediction unit 31 acquires basic weather information such as observation data necessary for weather prediction from, for example, a database of the Japan Meteorological Agency and its external organizations.

FIG. 3 is a schematic configuration diagram of the weather prediction unit.
Then, the weather forecasting unit 31 performs numerical weather calculation based on the initial state creation unit 31A that creates a plurality of initial states of weather based on the acquired basic weather information and the plurality of created initial states IS1 to ISM, respectively. And a plurality of numerical weather calculation units 31B for calculating the weather prediction members WM1 to WMM.

The power generation amount prediction unit 32 calculates the power generation amount of the power generation module 21 from the weather prediction.
For example, in the case of photovoltaic power generation, the amount of power generation can be calculated from the solar radiation intensity and temperature using the performance specifications of the power generation panel.

と表すものとする。 Methods for using weather prediction for photovoltaic power generation are, for example, “Hidetoshi Tamura, Hiromaru Hiraguchi, Atsushi Hashimoto, Keiichi Nishizawa, Koji Wada, 2010,“ Development of a method for predicting solar radiation for photovoltaic power generation (Part 1). ", Chuo Electric Power Research Laboratory Report N10029".
In the case of wind power generation, the power generation amount can be calculated from the wind power using the performance specifications of the windmill. Methods for using weather prediction for wind power generation prediction are, for example, `` Atsushi Hashimoto, Yasuo Hattori, Soichiro Sugimoto, Shinji Kadokura, Koji Wada, Hiromaru Hiraguchi, Nobukazu Tanaka, Toshiya Nanahara, 2008, `` It is described in “Meteorological model applicability evaluation”, Electric Power Research Institute report N08027 ”.
Below, the predicted value of power generation corresponding to the weather forecast member m is

It shall be expressed as

  The SOC initial value acquisition unit 33 acquires the initial state of the power storage unit 22 necessary for creating a power sale schedule from the power management unit 144. This is because the amount of power at the beginning affects the creation of subsequent schedules.

  The price prediction unit 34 is a power sale price prediction unit 34A that predicts a power sale price based on the weather forecast, a market power purchase price prediction unit 34B that predicts a market power purchase price based on the weather forecast, and a weather forecast. In addition, a market power sale price prediction unit 34C that predicts a future power sale price and a next day price prediction unit 34D that predicts a next day price based on weather forecasts are provided.

  In the above configuration, the expected power purchase price prediction unit 34B predicts the expected power purchase price for each time frame. However, depending on the power market MCT, the expected power purchase price may be constant. No prediction is necessary.

  Similarly, the expected power selling price prediction unit 34C predicts the expected power selling price for each time frame. However, depending on the power market MCT, the expected power selling price may be constant. No prediction is necessary.

  Further, the next day price prediction unit 34D predicts the price of power after the next day. In order to determine how much power should be left in the power storage unit 22 at the end of the day, it is necessary to predict the price of the next day, but when considering only the maximum profit on the next day, It is not always necessary to predict the price of power after the next day.

  The power sale amount determination unit 35 assumes a power sale price based on the power sale price prediction of the power sale price prediction unit 34A, calculates the assumed power sale price, and the predicted market power purchase price. An objective function creation unit 35B for creating an objective function based on the predicted marketed power selling price, the predicted next day price, and the calculated assumed power selling price, and the power generation amount prediction and power storage unit 22 of the power generation module 21. Based on the initial SOC value, the constraint condition creation unit 35C that creates a constraint condition that restricts the amount of power sold and the amount of power purchased, and the created objective function are solved under the created constraint condition, and the bid price at each bid time is calculated. It includes a mathematical plan solution creation unit 35D to be set, and a bid creation unit 35E that bids on the electric power market MCT based on the set bid amount.

The electric power market MCT is a market for buying and selling electric power. There are various market mechanisms. In this embodiment, the following market is assumed.
The power market MCT divides the time of the next day into a unit time τ (typically 30 minutes, one hour, etc.). The power generation company bids a set of power sale amount and bid price (bid price per unit power sale amount: unit price) for each time frame. In this way, a market in which a time zone is designated and bids on the previous day is called a spot market.

An object of the present embodiment is to create a power sale schedule in the spot market.
In the following, when simply selling power, it means selling power in this spot market, and is distinguished from market power selling that sells excess power on the spot.

Here, since the amount of electric power sold generally changes according to the price, a plurality of sets are presented for the combination of the amount of electric power sold and the bid price.
For example, the price per unit power sold in the power market MCT, that is, the amount of power desired to be sold when the market price = bid price PR1 is set as the amount of power sold = SP1, and the power sold when the market price = bid price PR2. The desired amount of power is set as the amount of power sold = SP2, and when the market price = the bid price PRN, the amount of power desired to be sold is set as the amount of sold power = SPN. Here, the bid price PR1 <the bid price PR2 <.

  That is, when the market transaction price in the electric power market MCT = PR1, the power sale amount = SP1 is sold, and when the market transaction price = PR2, the power sale amount is sold up to the power sale amount SP2. If = PRN, the bid is for selling power up to the power sale amount SPN.

Here, since the bid price does not always match the market transaction price, in such a case, the power sale amount corresponding to the market transaction price may be set by interpolation.
For example, the market transaction price = PRMCT, the bid price close to the market transaction price PRMCT is the bid price PR3 and the bid price PR4 (PR3 <PRMCT <PR4), and the power sale amount = SP3 when the bid price = PR3, When the amount of power sold when the bid price is PR4 = SP4, the amount of power sold SPX is obtained by the following equation.

SPX = {(SP4-SP3) × (PRMCT-PR3)} / (PR4-PR3)
+ SP3
On the other hand, bidding is also performed on the demand side, and in the power market MCT, the power price at which the supply and demand can be balanced, that is, the actual market price is determined.

  By the way, when the market price is determined, the power distributor is obligated to supply the specified amount of power at the time of bidding. (The amount of electricity sold) must be supplied. However, it is not always possible to accurately generate the amount. In particular, in the case of solar power generation, wind power generation, and the like, the amount of power generation depends on weather conditions and may vary greatly. In this case, generators have the following options.

When the power generation amount exceeds the promised power sale amount, for example, there are the following options (a) to (c).
(A) The power storage unit 22 is charged. This is possible if the power storage unit 22 has a margin.
(B) Consumed by the power consumption unit 25. However, there is not always an effective power usage method.
(C) Sell electricity. Ask the power system to buy the extra power. This shall always be possible in any amount. However, the electricity sales price is low.

When the power generation amount falls below the promised power sale amount, there are the following options (d) to (e).
(D) Discharge from the power storage unit to compensate. This is possible if there is power stored in the power storage unit.
(E) Going to buy electricity. Purchase the missing power from the power system side and make up for it. This always assumes that any amount is possible. However, the electricity purchase price is high. There is a penalty for the fact that the promised amount could not be sold.

  The power sale schedule creation device 13 of the embodiment creates a power sale schedule for how much power is sold in the spot market on the next day. The created power sale schedule is bid on the electricity market MCT. The amount of electric power sold and the electric power selling price for each time frame determined as a result of the electric power market MCT are sent to the electric power management unit 14 and used for controlling electric power equipment.

Next, the operation of the power sale schedule creation device 13 will be described.
FIG. 4 is an operation flowchart of the power sale schedule creation device.
First, the weather prediction unit 31 executes a weather prediction calculation, and makes a weather prediction member a power sale price prediction unit 34A, a marketed power purchase price prediction unit 34B, and a market power sale price prediction unit 34C, which will be described later, which constitute the price prediction unit 34. The next day price is output to the daily price prediction unit 34D and the power generation amount prediction unit 32, respectively (step S11).

  More specifically, the initial state creation unit 31A of the weather prediction unit 31 creates an initial state for numerical weather calculation from the acquired basic information. Since the power sale schedule needs to be created in consideration of the uncertainty of prediction based on numerical weather calculation, the weather prediction unit 31 performs prediction called ensemble (aggregate) prediction.

  Since the uncertainty of the weather prediction is largely due to the uncertainty of the initial state, the initial state creation unit 31A uses a plurality of initial states (see FIG. 3) with randomly generated perturbations for ensemble prediction. Then, M). Such initial state creation methods have a proven track record in the field of meteorological science, such as `` Toth, Z., Talagrand, O., Candille, G., & Zhu, Y. (2003). Probability and ensemble forecasts. Forecast Verification: A Practitioner's Guide in Atmospheric Science, 137-163.

  For each initial state thus created, the plurality of numerical weather calculation units 31B take into account the hydrodynamic and thermodynamic processes of the atmosphere, and change the state corresponding to the subsequent time transition. Create a state (path). Each of these is a prediction of the future atmospheric conditions. In the following, each path is referred to as an ensemble member (weather prediction member).

を計算する（ステップＳ１２）。
続いて売電価格予測部３４Ａは、翌日のスポット市場における電力売電価格を予測（計算）する（ステップＳ１３）。 Then, the plurality of numerical weather calculation units 31B include the calculated weather prediction members, a power sale price prediction unit 34A, a market power purchase price prediction unit 34B, a market power sale price prediction unit 34C, and a next day price. Output to the prediction unit 34D and the power generation amount prediction unit 32, respectively.
Thereby, the power generation amount prediction unit 32

Is calculated (step S12).
Subsequently, the power selling price prediction unit 34A predicts (calculates) the power selling price in the spot market on the next day (step S13).

Here, a method of predicting a power selling price based on a relatively simple past history will be described.
Electricity demand is high during the day, and demand decreases at night, especially after midnight. Therefore, the price tends to be high in the daytime and low in the nighttime. By capturing this pattern, the power selling price can be predicted.
As a technique for that purpose, for example, past histories for a predetermined number of days may be averaged. Let Pt (l): the spot market price for the time frame t one day before the power sale price history.

を計算する。 And, as an average value of the price for the past L days, the predicted power sale price for member m of the weather forecast,

Calculate

  Thus, when the predicted value of the power sale price is determined based only on the past history, the predicted value of the power sale price is the same for all members of the weather forecast.

Next, the expected power purchase price prediction unit 34B predicts (calculates) the expected power purchase price in the spot market on the next day (step S14).
The market power purchase price prediction unit 34B also performs prediction in the same procedure as the power sale price prediction unit 34A.

That is, the marketed power purchase price prediction unit 34B stores the marketed power purchase price history,
Qt (l): The expected power purchase price for the time frame t one day ago.

を計算する。 Then, the expected power purchase price prediction unit 34B calculates the expected power purchase price for the member m of the weather forecast as the average value of the price of the next day's expected power purchase price.

Calculate

Next, the expected power selling price prediction unit 34C predicts (calculates) the expected power selling price in the spot market on the next day (step S15).
The expected power selling price prediction unit 34C also performs prediction in the same procedure as the power selling price prediction unit 34A.

That is, the marketed electricity sales price prediction unit 34C stores the marketed electricity sales price history.
Rt (l): The market price of power for the time frame t one day ago.

を計算する。 Then, the expected power sale price prediction unit 34C calculates the expected power sale price for the next day as an average value of prices for the past L days.

Calculate

Next, the next day price prediction unit 34D predicts (calculates) a power sale price in the spot market on the next day (step S16).
The next day price prediction unit 34D has the same configuration as the power sale price prediction unit 34A.
The average of the daily price is calculated from the history of the power sale price, and the average of the past L days is taken as the next day price forecast.

  As described above, all the weather forecast members generated by the weather forecast unit 31 are the power sale price forecast unit 34A, the expected power purchase price forecast unit 34B, the expected power sale price forecast unit 34C, the next day price forecast unit 34D, It is sent to the power generation amount prediction unit 32 and is used for prediction of the spot power selling price, the expected power purchase price, the expected power selling price, the next day price, and the power generation amount according to the weather conditions represented by each weather prediction member. However, there are cases where weather forecasts are not necessarily used, such as in a market where the electricity sales price is fixed.

  Incidentally, the next day's weather forecast information is required by the power sale price prediction unit 34A, the market power purchase price prediction unit 34B, and the market power sale price prediction unit 34C. On the other hand, what is necessary for the next day price forecast unit 34D is weather forecast for the next day. In this respect, the information required by the next day price prediction unit 34D is different from the others.

  However, in the case of numerical weather prediction, the prediction of the next day can be obtained by continuously executing the calculation performed for the prediction of the next day. Therefore, the weather forecast given to the daily price forecasting unit 34D the next day is an extension of each weather forecasting member of the weather forecast given to the power selling price forecasting unit or the like.

Next, the SOC initial value acquisition unit 33 acquires the SOC initial value (step S17).
That is, if the operation of the power storage unit 22 of the current day is completed at the time when the power sale schedule is created, the SOC initial value acquisition unit 33 uses the current stored power as it is tomorrow's initial value, so Get the amount of power.

In addition, even if the power storage unit 22 is still in operation at the time of creating the power sale schedule, the SOC initial value acquisition unit 33 is tomorrow according to the schedule if the charge / discharge schedule until the start of tomorrow is confirmed. Calculate the initial value of.
On the other hand, when the power sale schedule is not confirmed, the SOC initial value acquisition unit 33 estimates the initial value as follows.

The current time, that is, the time point when the power sale schedule is created is t _0, and the amount of power stored in the power storage unit 22 at this time is c _t0 . The amount of power sales scheduled in the future is given by the following formula.

On the other hand, the amount of power generation predicted in the future is given by the following formula as an ensemble average.

で与えられる。 Therefore, the stored electric energy c _{0 in} the initial state of tomorrow is

Given in.

However, when this value becomes negative or exceeds the upper limit C of storable power, the correction is made as follows.
If c ₀ <0, then c ₀ = 0
If c ₀ > C, then c ₀ = C

をベースとして、この予測価格から市場価格がある程度上下した場合を想定して、そのおのおのの想定に対して売電量を決めなければならない。 As a result, the power sale amount determination unit 35 creates a power sale schedule (step S18).
Incidentally, the power selling price in the power market MCT is not necessarily the price predicted by the price prediction unit 34.

Based on the above, assuming that the market price has risen or lowered to some extent from this predicted price, the amount of power sold must be determined for each assumption.

FIG. 5 is an operation flowchart of the power sale amount determination unit.
First, the power sale price assumption unit 35A, the objective function creation unit 35B, and the constraint condition creation unit 35C capture a predicted value (step S21).

  Specifically, the power sale price assumption unit 35A receives all of the power sale prediction prices corresponding to each of all weather prediction members (in the above example, M weather prediction members) from the power sale price prediction unit 34A. Capture. Further, the objective function creation unit 35B is configured to obtain all the expected power purchase prices corresponding to each of all weather forecast members from the expected power purchase price prediction unit 34B, the expected power sale price prediction unit 34C, and the next day price prediction unit 34D. Captures the electricity sales forecast price and the forecast price the day after next. In addition, the constraint condition creation unit 35C captures the power generation amount prediction value corresponding to each of all weather prediction members.

Further, the constraint condition creation unit 35C takes in the SOC initial value from the SOC initial value acquisition unit 33 (step S22).
Subsequently, the power sale amount determination unit 35 sets the value of the process counter n to an initial value = 1 (step S23).

の確率分布

を作成する。ここでは例として、時刻ｔにおける

が、

の周りにガウス状に分布するというモデル、

と、隣接する時刻間の価格に相関があるとするモデル

と、を組み合わせて

とする。 Next, the power sale price assumption unit 35A creates an assumed power sale price (step S24).
Here, the power sale price assumption unit 35A creates an assumed power sale price by sampling using a random number from a plurality of taken power sale prediction prices.
Specifically, the power sale price assumption unit 35A first sets a value series that the assumed power sale price at each time can take.

Probability distribution

Create Here, as an example, at time t

But,

A model that is distributed in a Gaussian shape around

And a model where there is a correlation between prices between adjacent times

And the combination

And

である。 Where Z is a constant for normalization,

It is.

  From the probability distribution model determined in this way, a desired number of samples can be generated by using the Markov chain Monte Carlo method. For the Markov chain Monte Carlo method, see, for example, “Hirohiro Omori. Recent developments in the Markov chain Monte Carlo method. Journal of the Japan Statistical Society, 2001, 31.3: 305-344.” The price determined by sampling in this way is assumed as the assumed power selling price.

Assume that the assumed power selling price (sample) is a predetermined number N of samples.
In this case, if the number of samples N is too small, a price range in which the profit associated with the price is not sufficiently evaluated is created. Conversely, if the number of samples N is too large, the calculation time will be longer than necessary. It is assumed that an appropriate value of the sample number N is obtained in advance by experiments.

Next, a constraint condition is created by the constraint condition creating unit 35C (step S25).
The constraint condition creation unit 35C sets a constraint condition for the power sale schedule x _t (t = 1,..., T) as follows.

FIG. 6 is a diagram for explaining the variables used in the constraint condition creation unit and their relationship.
As shown in FIG. 6, the point P needs to be balanced, and a condition that satisfies this balance is a constraint condition.
The following conditions can be considered as conditions for optimization when setting the constraint conditions.

Optimization conditions:
C: Capacity of the power storage unit 22 (kWh)
S: Chargeable / dischargeable electric energy (kW)
_cf : target SOC at the end of the day (kWh)
f: Charging efficiency (0 ≦ f ≦ 1)
τ: Time unit for buying and selling electricity in the electric power market MCT T: Number of daily market divisions V: Power consumption value (yen / kWh)
It expresses how much it is worth converting to electricity purchase when it consumes electricity.
U _t : electric energy that can be consumed at time t (kWh)
U _all : Amount of power that can be consumed per day (kWh)

The conditions (prediction amount) (t = 1,..., T) determined according to the prediction ensemble member are as follows.

The variables determined by optimization are as follows.
x _t : Electricity sales at time t

  Furthermore, the variable when the weather prediction member m implement | achieves as a variable determined supplementarily is mentioned.

（ｂ）充電可能な電力は蓄電池の性能で制約（制限）される。

（ｃ）放電可能な電力は蓄電池の性能で制約される。

（ｄ）蓄電量が負にならない。蓄電量は、最大容量を超えない。

（ｅ）各時間帯に自己消費できる電力量には上限がある。

（ｆ）一日に自己消費可能な電力量には上限がある。

（ｇ）電力量のつりあい。

（ｈ）余剰電力が畜電池に蓄えられる。

Based on the variables, the constraint condition creating unit 35C creates the constraint conditions (a) to (h) shown below.
(A) The amount of electricity purchased and the amount of electricity sold are both 0 or more

(B) Rechargeable power is restricted (limited) by the performance of the storage battery.

(C) The power that can be discharged is limited by the performance of the storage battery.

(D) The amount of stored electricity does not become negative. The amount of electricity stored does not exceed the maximum capacity.

(E) There is an upper limit to the amount of power that can be consumed by each time zone.

(F) There is an upper limit to the amount of power that can be consumed per day.

(G) Balance of electric energy.

(H) Surplus power is stored in the livestock battery.

Next, an objective function is created by the objective function creation unit 35B (step S26).
Here, the objective function is expressed as the sum of the expected value of the entire power sales profit for one day when the power market MCT is held and the expected value of profit that the power stored in the power storage device brings on the following day.
The profit of power sale is expressed as the product of the expected price P _t ^(m) and the amount of power sale x _t .

  The value of power consumption is represented as V. Here, for the sake of simplicity, the value V of power consumption is considered to be a constant, but it can be extended to an ensemble member or time-dependent one as well as other values.

また、目的関数作成部３５Ｂは、各アンサンブルメンバーから得られる利益の平均値Ｉをとる。

The objective function creation unit 35B takes an average value I of profits obtained from each ensemble member.

に対応する売電量（ｘ_１，ｘ_２，…，ｘ_Ｔ）を求める。 Next, the mathematical plan solution creation unit 35D creates a mathematical plan solution (step S27).
The mathematical plan solution creation unit 35D solves the planning problem created in this way and generates a series of estimated power selling prices.

The amount of electric power sold (x ₁ , x ₂ ,..., X _T ) corresponding to.

時刻ｔのスポット価格＝Ｐ_ｔ，ｋの時に電力量ｘ_ｔ，ｋを売る
という、売電を行うことである。電力価格が正確に指定した価格にならない場合は、適宜補正するものとする。売電価格の指定は細かくできるので、これは特に問題とはならない。
ここで、売電価格は、予想価格のアンサンブル平均の周りに振れるものとして、

である。 Here, since the constraint condition and the objective function are already given, the problem for obtaining the optimal solution is formulated as a linear programming problem. The solution of the linear programming problem is described in, for example, “Hiroshi Konno,“ Linear Programming ”, Nikka Giren.”
Next, the power sale amount determination unit 35 adds 1 to the count n (step S28), and determines whether the count n is less than N (step S29).
If it is determined in step S29 that the count n is less than N (step S29; Yes), the process proceeds to step S24 again.
If the count n is N in the determination in step S29 (step S29; No), the bid creation unit 35E creates a bid for the power market MCT and outputs it as a power sale schedule (step S30). The output power sale schedule is described as follows.

This is to sell power by selling the electric energy x _{t, k} when the spot price at time t = P _{t, k} . If the electricity price is not exactly the specified price, it shall be corrected as appropriate. This is not particularly a problem because the power selling price can be specified in detail.
Here, the power selling price is assumed to swing around the ensemble average of the expected price,

It is.

Here, ΔP is the step size of the bid. A method in which the standard deviation is set to ΔP with reference to the past price history is conceivable. Or an estimate of the standard deviation of the expected price of the ensemble member

Attention is paid to a specific time t (1 ≦ t ≦ T). For each sample of the power sale assumed power selling price Price assuming unit 35A creates, calculated as ensemble average price / P _t.

Since each sample is created using random numbers as described above, the average ensemble price varies from sample to sample.
FIG. 7 is an operation explanatory diagram of the bid creation unit.
As shown in FIG. 7, for values of P _t, plotting the amount of power sold x _t at time t at that time. By smoothing the plot, it is possible to obtain a power sale amount x _t when the price is P _t .
A bid can be created by reading the value of the smoothed function P _{t, k} (k = −K,..., K).
As described above, according to the first embodiment, it is possible to create a more suitable power transaction plan while taking into account the uncertainty of weather prediction, and the expected profit can be maximized. .

[2] Second Embodiment In the first embodiment, the power sale price prediction unit predicts the power sale price in the spot market based only on the past power sale history.
However, the behavior of power consumers changes according to the weather conditions, and as a result, the power demand changes. For example, the usage amount of air conditioning varies depending on the degree of heat and cold. In addition, if the introduction of renewable energy whose power generation changes according to the weather conditions, the supply will also depend on the weather conditions. Therefore, the electricity price determined by the balance between supply and demand also changes according to the weather conditions. For this reason, the predicted value of the power selling price also varies depending on the ensemble member.
Therefore, in the second embodiment, a case will be described in which a power sale price prediction unit that takes into consideration the weather prediction of the ensemble member is provided.

FIG. 8 is a schematic configuration diagram of a power sale price prediction unit according to the second embodiment.
The power sale price prediction unit 34AX of the second embodiment is roughly classified into a power sale price history DB 41, a weather condition history DB 42, an explanatory variable extraction unit 43, a prediction formula generation unit 44, a price calculation unit 45, It has.

FIG. 9 is a schematic configuration diagram of the power sale price history DB.
The power sale price history DB 41 records a history of power sale prices in the spot market in the past.
More specifically, as shown in FIG. 9, time data 51 storing the date and time when the power selling price history DB 41 acquired the power selling price, and the power selling price acquired at the time corresponding to the time data 51. Power selling price data 52 stored therein.

FIG. 10 is a schematic configuration diagram of the weather condition history DB.
The weather condition history DB 42 records the history of weather prediction in the past.
More specifically, as shown in FIG. 10, the weather condition history DB 42 includes time data 61 storing the date and time when the weather condition is acquired, and a plurality of weather conditions acquired at the time corresponding to the time data 61. A plurality of weather prediction member data 62-1, 62-2,..., 62-M storing the prediction members.
The weather forecast member data includes, for example, temperature data 63 in which the temperatures constituting the weather forecast member are stored, solar radiation data 64 in which the amount of power generation per unit time (Wh) based on the amount of solar radiation, and rainfall (mm). And rain data 65 stored therein.

  The purpose of the power sale price prediction unit 34AX is to predict the price by utilizing the relationship between the weather condition and the power sale price, so that it is preferable if a history of actual weather conditions is available. However, the weather conditions may limit the available data depending on the location and time.

  For this reason, in this 2nd Embodiment, the weather prediction member which is the result of a weather prediction is used as a substitute of actual weather conditions. In this case, it is not always necessary that all weather prediction members of the ensemble prediction are recorded in the weather condition history DB 42. In that case, since the ensemble average (average of all members) can be expected to be the closest value to the actual weather condition, this is calculated and stored in the weather condition history DB 42.

The explanatory variable extracting unit 43 extracts a weather variable that is considered to be effective for the power sale price prediction from the information accumulated in the weather condition history DB 42. Here, the meteorological variable corresponds to an explanatory variable as a general term of regression analysis used in the following procedure.
In general, weather variables that are considered to be related to power demand (power consumption) are temperature near the ground, wind near the ground, solar radiation, rainfall, and the like that are considered to affect human behavior. For example, it is considered that the number of people who use air conditioning equipment increases when the temperature is higher or lower than the temperature that is considered comfortable. When wind, solar radiation, and precipitation are strong, people affect behavior patterns such as refraining from going out. In addition, the impact on the use of lighting, the impact on the cooling effect of the building, etc. can also be considered. Therefore, these meteorological variables (meteorological conditions) affect the power demand.

  On the other hand, the weather variable (meteorological conditions) considered to be related to the power supply amount includes the amount of solar radiation that affects the amount of photovoltaic power generation. Moreover, since the output of a photovoltaic power generation panel also depends on the temperature of the panel, the temperature near the ground is also cited as a weather variable that affects power supply. In the case of wind power generation, the wind speed is mainly affected as a weather variable. Furthermore, when the direction of the windmill is controlled to the windward, if the change in the wind direction is abrupt, it cannot follow efficiently and power generation efficiency decreases. Therefore, the wind direction can also be cited as a weather variable that affects the power supply.

Moreover, from a regional perspective, the electricity sales price can be considered to be mainly influenced by the weather variables (meteorological conditions) in the region where consumers and suppliers enter the market. Hereinafter, this area is referred to as a target area.
The explanatory variable extraction unit 43 extracts only the weather variables related to the target area from the weather condition history DB 42. In this case, the target weather variable may be obtained with a fine resolution in a grid pattern when viewed geographically.

  For example, image data obtained by satellite observation, and grid data output as a result of integrating various data. This is also the case when numerical weather calculation prediction results are used as weather variables (meteorological conditions). In this case, the prediction result (meteorological condition data) of the numerical weather calculation is given a value with a spatial resolution (several Km) far finer than the scale (approximately several hundred Km) of the target area of the electric power market MCT.

By the way, in general, if there are excessive explanatory variables in the regression analysis, a state called overlearning occurs, and the prediction performance deteriorates.
Therefore, when the weather variables having the fineness as described above are used as the explanatory variables, the prediction performance is deteriorated. Therefore, it is necessary to limit the weather variables (explanatory variables) to an appropriate number. As a technique for this, for example, the weather variables (explanatory variables) used for the regression analysis can be narrowed down by averaging the weather variables in the target area.

  However, if the average of the entire target area is summarized into one weather variable, for example, solar power generation facilities are concentrated in a specific area in the target area, or a large city is located in a specific area in the target area. There is a situation where power consumption is concentrated and cannot be reflected.

  Therefore, as one method for avoiding this problem, a method is possible in which the target area is divided into intermediate grids (several tens of kilometers) and the average is calculated within each intermediate grid.

The prediction formula generation unit 44 creates a formula that gives the relationship between the weather conditions and the power selling price.
Here, a method of constructing a prediction formula using linear regression analysis will be described. In the following, time (ie, day (d) and time zone (t)) are represented by a set of (d, t). Now, let P _{d, t} be the power selling price on the date d and time zone t described in the power selling price history DB 41. In addition, the number of weather variables extracted from the weather prediction history DB by the explanatory variable extraction unit 43 is H, and the set is (v ¹ , v ² ,..., V ^H ).

Since electricity fluctuates in demand throughout the day, the electricity sales price also varies greatly throughout the day.
On the other hand, the change pattern of the electricity sales price is often the same even if the day changes. Considering these things, the prediction formula of the power selling price will be made a separate prediction formula for each time zone during the day. At this time

の最小化条件

Residual error squared

Minimization condition

ここに述べた売電価格予測部の予測方法と同様な方法により、気象予測を考慮した成り行き売電価格予測部、成り行き買電価格予測部を作ることができる。
以上の説明のように、本第２実施形態によれば、気象予測に際し、気象変数（説明変数）を適切な個数に限定しつつ、高精度で気象予測を行い、ひいては、適切な入札を行うことが可能となる。

By using a method similar to the prediction method of the power sale price prediction unit described here, it is possible to create a market power sale price prediction unit and a market power purchase price prediction unit that take weather forecast into consideration.
As described above, according to the second embodiment, in weather prediction, the weather variables (explanatory variables) are limited to an appropriate number, the weather prediction is performed with high accuracy, and appropriate bidding is performed. It becomes possible.

[3] Third Embodiment In the first embodiment described above, all ensemble members created by weather prediction are treated as having the same realization probability as prediction.
In the third embodiment, it is also conceivable to create an ensemble having a high realization probability and a low realization probability.

FIG. 11 is a schematic configuration diagram of the weather prediction unit of the third embodiment.
The weather prediction unit 31X is roughly classified into a reference value estimation unit 71, a reference value history DB 72, a weather condition history DB 73, a variance estimation unit 74, a sample creation unit 75, a weather numerical value calculation unit 76, and a weight calculation. Part 77.

In a weather numerical calculation, a typical method for creating an ensemble is a method in which perturbation is added to the initial value of the numerical calculation.
FIG. 12 is an operation flowchart of the third embodiment.
Let X be one of the weather variables. For example, temperature, pressure or wind direction component at a certain point.
The reference value estimating section 71 estimates the reference value X ₀ is a most probable value of the meteorological variable X from the observation data (step S41).

  In the following explanation, it is assumed that the weather variable X is a univariate variable, that is, a scalar that takes a single value. However, in the case of actual weather calculation, the target range is spatially widened and a plurality of grid points are It is different in that it handles multivariables by handling values, handling multiple types of physical variables such as temperature, pressure, and wind speed.

Therefore, in the following description, the normal distribution is used. However, when actual weather calculation is performed, a change such as using a multiple normal distribution instead of the normal distribution is necessary.
However, since the basic idea does not change, in order to simplify the description, the following description will be made using a weather variable X that is a scalar.

The reference value history DB 72 stores the reference value X ₀ estimated by the reference value estimation unit 51.
The weather condition history DB 53 stores past weather variable X values.
Incidentally, the reference value X ₀ to the reference value estimating section 51 has estimated there is an error.
Therefore, the variance estimation unit 54 estimates the standard deviation σ _X of the reference value X ₀ by matching the reference value history DB 52 and the weather condition history DB 53 (step S42).

Assuming normal distribution, the probability distribution of the weather variable X is given by the following normal distribution formula.

The sample creation unit 55 generates M samples (X ₁ , X ₂ ,..., X _M ) from this normal distribution (step S43). Extraction of samples from the normal distribution is described in, for example, “Haruhiko Okumura“ Latest Algorithm Encyclopedia in C ”, Technical Review, 1991. ISBN 4-87408-414-1.
M samples (X ₁ , X ₂ ,..., X _M ) are obtained by adding a perturbation having a magnitude of about the standard deviation σ _X to the original reference value X ₀ .

The weather numerical calculation unit 56 obtains each weather prediction member of the ensemble weather prediction by setting _M samples (X ₁ , X ₂ ,..., X _M ) as initial values of the numerical weather calculation.
The weight calculator 57 calculates the weights of the samples X _{1 to} X _M (Step S44).

  In this case, it is necessary to give individual weights to individual samples, for example, when there is another observation value Y that is not considered at the time of initial value estimation, and the reference value X, observation value Y, This is a case where there is a dependency relationship. The dependency is assumed to be given by the conditional statistical distribution p (X | Y) of the observation location Y when the reference value X is given.

より、参照値Ｘの確率分布は、サンプル作成に用いた統計分布ｐ（Ｘ）そのものではなく、それに統計分布ｐ（Ｙ｜Ｘ）を乗じたものになっていることに注意する必要がある。 At this time, the weight calculator 57

Therefore, it should be noted that the probability distribution of the reference value X is not the statistical distribution p (X) itself used for sample creation, but is multiplied by the statistical distribution p (Y | X).

で与えられる。ここで

となるように規格化されている。 Note that the statistical distribution p (Y) in the above equation does not depend on the reference value X and does not affect the relative weight of the sample. It is assumed that the high realization probability of each of the M members is represented by a weighted set (w ₁ , w ₂ ,..., W _M ). Therefore, the weight w _m given to each member is

Given in. here

It is standardized to become.

At this time, the objective function creation unit 35B generates an objective function represented by the following equation (step S45).

  As a result, according to the third embodiment, since the weather prediction can be performed by appropriately weighting each weather prediction member, it is possible to predict the weather condition after the next day with more certainty, and thus more reliably the profit. A bid can be obtained.

[4] Modified Example of Embodiment A power transaction support device (power sale schedule creation device) of the present embodiment includes a control device such as a CPU, a storage device such as a ROM (Read Only Memory) and a RAM, an HDD, and a CD drive. The apparatus includes an external storage device such as a device, a display device such as a display device, and an input device such as a keyboard and a mouse, and has a hardware configuration using a normal computer.

  The control program executed by the power transaction support apparatus of the present embodiment is an installable or executable file, such as a CD-ROM, a flexible disk (FD), a CD-R, a DVD (Digital Versatile Disk), etc. The program is provided by being recorded on a computer-readable recording medium.

  Moreover, you may comprise so that the control program run with the power transaction support apparatus of this embodiment may be provided by storing on a computer connected to networks, such as the internet, and downloading via a network. Moreover, you may comprise so that the control program performed with the electric power transaction assistance apparatus of this embodiment may be provided or distributed via networks, such as the internet.

Moreover, you may comprise so that the control program of the electric power transaction assistance apparatus of this embodiment may be provided by previously incorporating in ROM etc.
The control program executed by the power trading support device according to the present embodiment includes the above-described units (power sales price prediction unit, market power sales price prediction unit, market power purchase price prediction unit, power generation amount prediction unit, power sales amount determination unit). As the actual hardware, a CPU (processor) reads the control program from the storage medium and executes it to load each of the above units on the main storage device, A market power price prediction unit, a market power purchase price prediction unit, a power generation amount prediction unit, and a power sale amount determination unit are generated on the main storage device.

  Although several embodiments of the present invention have been described, these embodiments are presented by way of example and are not intended to limit the scope of the invention. These novel embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the scope of the invention. These embodiments and modifications thereof are included in the scope and gist of the invention, and are included in the invention described in the claims and the equivalents thereof.

DESCRIPTION OF SYMBOLS 10 Electric power system 11 Electric power generation / consumption system 12 Communication network 13 Electric power sale schedule preparation apparatus 14 Electric power management part 21 Electric power generation module 22 Electric power storage part 23 Electric power sale part 24 Electric power purchase part 25 Electric power consumption part 31 Weather prediction part 32 Electric power generation amount prediction part 33 SOC Initial Value Acquisition Unit 34 Price Prediction Unit 35 Electricity Sales Determination Unit 41 Electricity Sales Price History DB
42 Weather Condition History DB
43 explanatory variable extraction unit 44 prediction formula generation unit 45 price calculation unit

Claims (9)

A power trading support device for setting a power sale schedule for a power system of a power generation system that has facilities capable of temporal adjustment with respect to storage or use of power and that generates power using renewable energy,
A weather prediction unit that performs ensemble prediction of the weather and generates a plurality of ensemble members;
A power sale price prediction unit for predicting a spot power sale price in the spot market of the next day based on the ensemble member;
A market electricity price prediction unit that predicts a market electricity sales price in the market of the next day based on the ensemble members;
An expected power purchase price predicting unit that predicts an expected power purchase price in the market of the next day based on the ensemble member;
A power generation amount prediction unit that predicts the power generation amount of the power generation system on the next day based on the ensemble member;
Based on the predicted next-day spot power selling price, next-day expected power selling price, next-day expected power selling price, and power generation amount of the power generation system on the next day, a power selling schedule that maximizes expected profit is sold. A coulometric determination unit,
The power sale amount determination unit creates a constraint condition that creates a constraint condition that restricts the power sale amount based on the power generation amount of the power generation system predicted by the power generation amount prediction unit;
Based on the spot power selling price, a power selling price assumption unit for setting an assumed power selling price;
An objective function creation unit for creating an objective function for calculating profits at the time of power sale;
By solving the mathematical programming problem that is set by the consequences power purchase price, the objective function and the constraint condition is applied the consequences power selling price and the assumed power selling price, to maximize the expected profit, before Symbol power selling A mathematical plan solution creation section for creating a schedule;
A power trading support device comprising: A next-day price prediction unit that predicts a next-day power sale price that is a power sale price after the next day based on the ensemble member,
The power sale amount determination unit creates a power sale schedule based on the next day power sale price.
The power trading support device according to claim 1. The power generation system includes a power storage unit that stores generated power,
An initial value acquisition unit that acquires an initial value of the stored power amount at the market start time of the next day of the power storage unit,
The constraint condition creating unit creates the constraint condition in consideration of an initial value of the stored energy.
The power trading support device according to claim 1. Based on the assumed power selling price and the power selling schedule, a bid creating unit that creates a bid for the next day power selling market,
The power trading support device according to any one of claims 1 to 3. A weather condition history DB storing a history of weather conditions;
A power sale price history DB storing a history of power sale prices;
A variable extraction unit for extracting a weather variable from the weather condition history DB;
Based on the history of the power selling price, a prediction formula generating unit that generates a prediction formula for predicting the power selling price using the weather variable using a regression analysis method;
A power sale price calculation unit that predicts a power sale price based on the plurality of ensemble members and the prediction formula;
The power trading support device according to any one of claims 1 to 4, further comprising: A weighting unit for weighting the ensemble member;
The power sale amount determination unit creates the power sale schedule in consideration of the weighting.
The power trading support device according to any one of claims 1 to 5. A power generation module that generates power using renewable energy, a power storage unit that stores power generated by the power generation module, and a power consumption unit that consumes power generated by the power generation module or power stored by the power storage unit. Power generation system,
A weather prediction unit that performs ensemble prediction of the weather and generates a plurality of ensemble members, a power sale price prediction unit that predicts a spot power sale price in the spot market on the next day based on the ensemble members, and a ensemble member To the ensemble member, a market power price prediction unit for predicting a market power sale price prediction in the market market of the next day, a market power purchase price prediction unit for predicting a market power purchase price in the market market of the next day based on the ensemble member, and the ensemble member A power generation amount prediction unit that predicts the power generation amount of the power generation system on the next day based on the predicted spot power sale price on the next day, the next day's market power sale price, the next day's market power purchase price, and the next day's Based on the amount of power generation, a power sale schedule that maximizes expected profits Comprising a power sale amount determining section for forming a power transaction support apparatus having a, a,
The power sale amount determination unit creates a constraint condition that creates a constraint condition that restricts the power sale amount based on the power generation amount of the power generation system predicted by the power generation amount prediction unit;
Based on the spot power selling price, a power selling price assumption unit for setting an assumed power selling price;
An objective function creation unit for creating an objective function for calculating profits at the time of power sale;
By solving the mathematical programming problem that is set by the consequences power purchase price, the objective function and the constraint condition is applied the consequences power selling price and the assumed power selling price, to maximize the expected profit, before Symbol power selling A mathematical plan solution creation section for creating a schedule;
Power trading system with A control method executed by a power trading support device that has a facility capable of temporal adjustment with respect to storage or use of power and sets a power sale schedule for a power system of a power generation system that generates power using renewable energy. There,
A weather prediction process that performs ensemble prediction of the weather and generates multiple ensemble members;
A power selling price prediction process for predicting a spot power selling price in the spot market of the next day based on the ensemble member;
A predicted power selling price prediction process for predicting a marketed power selling price in the market of the next day based on the ensemble member;
A process of predicting the electricity purchase price for predicting the electricity purchase price in the market of the next day based on the ensemble member;
A power generation amount prediction process for predicting the power generation amount of the power generation system on the next day based on the ensemble member;
Based on the predicted next-day spot power selling price, next-day expected power selling price, next-day expected power selling price, and power generation amount of the power generation system on the next day, a power selling schedule that maximizes expected profit is sold. A coulometric determination process,
The power sale amount determination process, and constraint creation process of creating a constraint for constraining the coulometric sale based on the power generation amount of the power generation system is predicted in the power generation amount prediction process,
A power selling price assumption process for setting an assumed power selling price based on the spot power selling price;
Objective function creation process for creating an objective function for calculating profits when selling power,
By solving the mathematical programming problem that is set by the consequences power purchase price, the objective function and the constraint condition is applied the consequences power selling price and the assumed power selling price, to maximize the expected profit, before Symbol power selling A control method comprising a mathematical programming solution creation process for creating a schedule. A computer for controlling a power trading support device for setting a power sale schedule for a power system of a power generation system having a facility capable of temporal adjustment with respect to storage or use of power and generating power by renewable energy A control program,
The computer,
A means for predicting ensembles of weather and generating multiple ensemble members;
Means for predicting a spot power selling price in the spot market of the next day based on the ensemble member;
Means for predicting the marketed electricity sales price in the market of the next day based on the ensemble members;
Means for predicting the market price of electricity in the market of the next day based on the ensemble members;
Means for predicting the power generation amount of the power generation system on the next day based on the ensemble member;
Means for creating a power sale schedule that maximizes expected profit based on the predicted spot power sale price of the next day, the next day's market power price, the next day's market power price, and the power generation amount of the power generation system on the next day And function as
Means for creating a power sale schedule that maximizes the expected profit, means for creating a constraint that restricts the amount of power sold based on the power generation amount of the power generation system predicted by the means for predicting the power generation amount,
Means for setting an assumed power selling price based on the spot power selling price;
A means of creating an objective function for calculating profits from power sales;
The power selling schedule for maximizing the expected profit by solving the mathematical programming problem set by the objective function and the constraint conditions applying the marketed power selling price, the marketed power selling price and the estimated power selling price. Function as a means to create,
Control program.

Images