KR102266099B1 - ESS operating system and method for small power brokerage transactions - Google Patents

Abstract

The present invention relates to a method for generating an ESS charge/discharge schedule to reduce a peak load and maximize economic benefits, and a system thereof. To this end, the system comprises: a data collection unit receiving and storing power data including price variables for each time period; an environment setting unit setting variables including a control variable, a state variable, and an output variable and functions including an output function and a cost function on the basis of the data stored in the data collection unit; and a predictive control calculation unit generating control variables for each time period by using the variables and functions set in the environment setting unit. The environment setting unit sets a battery charge/discharge amount (BAT(k)) for each time period as a control variable, sets the amount of electricity input to the grid for each time period as a state variable (GRID(k)), sets a saving amount for each time period based on a state variable as an output variable (SAVE(k)), and sets a target function (Rs) for a saving amount and an output function (Y) including the saving amount for each time period. The predictive control calculation unit calculates a difference of the control variable for each time period so that the cost function is minimized on the basis of a cost function (J), which uses a difference between the target function and the output function and a difference of the control variable for each time period, thereby generating a control value for each time period. Accordingly, a problem of time inconsistency between power generation and use in an ESS is systematically solved, thereby providing advantages of reducing a peak load and maximizing benefit against cost.

Description

ESS operating system and method for small power brokerage transactions

The present invention relates to an ESS operating system and method for a small-scale power brokerage transaction, and more specifically, to an ESS operating system and method for a small-scale power brokerage transaction that reduces peak loads and maximizes economic benefits.

Communication and power use are getting smarter, and supplying stable power without power cut is one of the important factors for operators. In this aspect, in terms of power operation, it is required to build a power storage device at a low cost and supply power without power interruption.

Accordingly, it is predicted that ESS devices will be expanded to small businesses and buildings along with the large-scale ESS market, and various functions suitable for the smart environment for monitoring and control that can operate them, as well as interoperability and scalability of energy storage devices There is a demand for an excitation power supply device for

On the other hand, if there is a large difference between the predicted solar power generation amount and the real-time power supply amount on the same day, the LNG generator must be additionally operated or stopped, resulting in unexpected additional costs. Moreover, there is a problem in that it is difficult to predict the amount of power generated on the day of power delivery because renewable energy such as solar power has no obligation to predict the amount of power generation.

As an energy management system of a conventional ESS, referring to Patent Publication No. 10-2019-0112441, the energy management system includes a power demand correction unit and an operation schedule update unit. If the difference exceeds the reference value, the power demand data is corrected in such a way that it increases or decreases by a predetermined value.

In addition, the operation schedule update unit updates the charge/discharge operation schedule in consideration of the power demand correction data and the collected charge/discharge state data extracted by collecting the electric power demand correction data and the battery charge/discharge state corresponding to the same time period from the energy storage device. .

Although it is possible to prepare for electricity demand through the updated charge/discharge operation schedule in this way, it has to be continuously updated in real time, so it takes a lot of load on the server, and it is updated on a timely basis, so there is a problem that the future forecast is not systematic.

In addition, the prior art as described above does not consider a method of maximally reducing the error of the generation amount prediction value only to prepare for the power demand.

The present invention is to solve the above problems, and specifically, to provide an ESS operating system for small-scale power brokerage transactions that can minimize the error between the real-time generation amount and the generation amount predicted value after at least 24 hours and maximize the generation income. it is for

In order to achieve the above object, the present invention provides a data collection unit for receiving and storing power data including a price variable for each time period; an environment setting unit for setting a variable including a control variable, a state variable, an output variable, an output function, and a function including a cost function based on the data stored in the data collection unit; It is configured to include a predictive control calculation unit for generating control variables for each time period using the variables and functions set in the environment setting unit,

The environment setting unit,

Column vector (BAT(k), GRID(k), LOAD(k), PV(k), SELL(k)) ^T as control variable, column vector (aBAT(k), aGRID(k), aLOAD(k) ), aPV(k), aSELL(k)) ^T as the state variables, and the time-saving amount based on the state variable is set as the output variable SAVE(k), and the target function Rs for the savings amount and the time-saving amount Sets an output function Y including, and the predictive control calculation unit is a control variable for each time period that minimizes the cost function based on a cost function J using the difference between the target function and the output function and the difference of the control variable for each time period It provides an ESS charge/discharge schedule generation system that generates control variables for each time period by calculating the difference between

The cost function J is

J = (Rs - Y) ^T (Rs - Y) + ΔU ^T RwΔU,

Preferably, the predictive control calculator calculates the cost change function as ΔU = (Φ ^T Φ+Rw) ^-1 Φ ^T (Rs-FX(k)).

The relation between the control variable and the state variable is

aGRID(k+1) = aGRID(k) + GRID(k)×CHARGE(k)

aBAT(k+1) = aBAT(k)+BAT(k)

aLOAD(k+1) = aLOAD(k) + LOAD(k)×CHARGE(k)

aPV(k+1) = aPV(k) + PV(k)×CHARGE(k)

aSELL(k+1) = aSELL(k) + SELL(k)×SMPREC(k)

can be set to

The output variable is

SAVE(k) = aBAT(k)×CONV - aGRID(k) + aLOAD(k) + aSELL(k) is set, and the output function is

Y = [SAVE(k+1|k) SAVE(k+2|k) … SAVE(k+Np|k)] ^T is preferably set.

The target function Rs is

(LOAD(k) is the predicted load power for each time period, CHARGE(k) is the electricity rate per kWh for each time period, and Np is the predicted number of samples).

In addition, according to the present invention, the data collecting unit comprising the steps of receiving and storing the power data including the price variable for each time period; setting, by an environment setting unit, a variable including a control variable, a state variable, an output variable, an output function, and a cost function, based on the data stored in the data collection unit; performing the step of generating control variables for each time period using the variables and functions set in the environment setting unit by the predictive control calculation unit,

The environment setting unit,

Column vector (BAT(k), GRID(k), LOAD(k), PV(k), SELL(k)) ^T as control variable, column vector

(aBAT(k), aGRID(k), aLOAD(k), aPV(k), aSELL(k)) Set ^T as the state variable, and the amount of time savings based on the state variable as the output variable SAVE(k), , a target function Rs for the savings amount and an output function Y including the savings amount for each time period, and the predictive control calculation unit is a cost function J using the difference between the target function and the output function and the difference between the control variables for each time period. It provides a method for generating an ESS charge/discharge schedule for generating a control variable for each time period by calculating the difference of the control variable for each time period based on which the cost function is minimized.

According to the present invention, there is an advantage in that the peak load can be reduced and cost benefits can be maximized by systematically solving the problem of time inconsistency between power generation and use in the ESS.

In addition, there is an advantage in that it is possible to reduce the load on the server and perform relatively long-term and systematic control by generating predictive control variables for each time period in units of days.

1 is a block diagram showing the configuration of an ESS charge/discharge schedule generation system according to the present invention;
2 is an exemplary graph showing the predicted load and predicted solar power generation by time for two days;
3 is a predictive control graph showing the amount of power input into the grid and the amount of battery charge/discharge for each time period for two days;
4 is a prediction graph showing a change in a state variable based on FIG. 3;
5 is a prediction graph showing a change in the accumulated battery charge amount based on FIG. 3;
6 is a graph of results showing the target function and output variables for two days.

The configuration and operation of specific embodiments according to the present invention will be described in detail with reference to the drawings.

Referring to FIG. 1 , the ESS operating system for generating the ESS charging/discharging schedule according to the present invention may be connected to the ESS and provided to control the ESS, the data collection unit 100, the environment setting unit 200, the predictive control It is configured to include a calculation unit (300).

In the data collection unit 100, various types of data such as ESS charge/discharge data, predicted load data, and cost data are transmitted and stored. The data stored in the data collection unit may include data transmitted and stored from the outside and data input by a user.

Referring to FIG. 2 , the electricity rate varies by time period, and the present invention proposes a method for maximizing the rate gain by using the electricity rate for each time period.

Referring to FIG. 3 , the predicted load data may be stored for each time period, for example, may be made of data processed from past data. That is, the data may be formed by interpolating an average or trend for each time period based on past data.

It is preferable that the prediction load data consists of data according to a predetermined time interval. For example, the time interval may be 15 minutes, and may consist of 96 pieces of data based on 24 hours a day. In FIG. 2 , the vertical axis indicates kW, and the horizontal axis 1 indicates 15 minutes, showing two days of data.

The environment setting unit 200 and the predictive control calculation unit 300 generate ESS optimal charging/discharging data based on MPC (Model Predictive Control).

Specifically, the environment setting unit 200 includes a control variable setting unit 210 , a state variable setting unit 220 , an output variable setting unit 230 , a target function setting unit 240 , and an output function setting unit 250 . , and a cost function setting unit 260 .

The control variable setting unit 210 consists of a column vector, specifically

(BAT(k), GRID(k), LOAD(k), PV(k), SELL(k)) Set ^T as a control variable. Here, T means transpose.

The BAT(k) represents a battery charge/discharge amount for each time period. Here, k represents an order according to time. For example, if the number of samples is 96, k is 1,2,3,… ,96, and if data is for 24 hours a day, k may represent data at 15-minute intervals.

The control variable refers to a physical quantity actually controlled by the system, and BAT(k) indicates the amount of charge or discharge of the battery according to time. For example, when BAT(k) is +, it may mean charging of the battery, and when BAT(k) is -, it may mean discharging of the battery.

The GRID(k) represents the amount of grid input power for each time period, and LOAD(k) represents the predicted load power amount for each time period. The LOAD(k) may be transmitted from the data collection unit 100 or input from the user.

The PV(k) means the predicted solar power generation amount for each time period, and SELL(k) means the solar power generation sales volume for each time period.

In the state variable setting unit 220, the column vector is also

(aBAT(k), aGRID(k), aLOAD(k), aPV(k), aSELL(k)) Set ^T as a state variable.

The state variable is a physical quantity that changes from time to time according to the control variable at a specific time. aBAT(k) represents the accumulated battery charge/discharge amount, aGRID(k) represents the accumulated grid incoming electricity charge, and aLOAD(k) is It represents the cumulative predicted load power rate. In addition, aPV(k) means the cumulative predicted solar power generation amount, and aSELL(k) means the cumulative solar power generation sales volume.

In the environment setting section, the relation between the control variable and the state variable is

aGRID(k+1) = aGRID(k) + GRID(k)×CHARGE(k) (1)

aBAT(k+1) = aBAT(k)+BAT(k) (2)

aLOAD(k+1) = aLOAD(k) + LOAD(k)×CHARGE(k) (3)

aPV(k+1) = aPV(k) + PV(k)×CHARGE(k) (4)

aSELL(k+1) = aSELL(k) + SELL(k)×SMPREC(k) (5)

It can be set to be Here, CHARGE(k) is the electricity rate per kWh per time period as shown in FIG. 1 , and SMPREC(k) refers to the sales amount per kWh per time period.

The output variable setting unit 230 sets the saving amount SAVE(k) for each time period as an output variable. The output variable is a variable that changes according to the state variable, and SAVE(k) means the accumulated saving cost up to the k-th time due to the charging and discharging of the ESS battery. In other words,

SAVE(k) = aBAT(k)×CONV - aGRID(k) + aLOAD(k) + aSELL(k) (6)

can be set to Here, CONV may be set as the largest amount of CHARGE(k), ie, max(CHARGE(k)), as a power rate per kWh for the remaining amount of the battery.

The target function setting unit 240 sets ideal data for the saving amount, and if r(k) is the target saving amount at the k-th time, the target function Rs can be expressed as follows.

Rs ^T = [1 1 1 … 1]r(k) (7)

Here, T denotes a transpose matrix, and the number of columns is equal to the number of predicted samples Np. The number of prediction samples is a number indicating how many time intervals are included when predictive control is performed once. For example, if prediction control is performed for 24 hours a day and prediction control is performed every 15 minutes when prediction control is performed once, the number of prediction samples Np is 24 × 60÷15 = 96.

The output function setting unit 250 represents the output function Y in the form of a vector using an output variable according to each time interval. In other words,

Y = [SAVE(k+1|k) SAVE(k+2|k) … SAVE(k+Np|k)] ^T (8)

can be set to Here, SAVE(k+m|k) is an output variable predicted at time k, and means the amount of savings at time k+m.

The environment setting unit 200 configures a charge/discharge control formula based on the above settings. Specifically, the environment setting unit configures the following determinant.

X(k+1) = A(k)X(k) +B(k)U(k) (9)

Y(k) = C(k)X(k) (10)

here,

,

,

,

,

,

,

is composed of

The cost function setting unit 260 configures a cost function J for the difference using the target function and the output function.

J = (Rs - Y) ^T (Rs - Y) (11)

In addition, the cost function setting unit 260 may be set by including the cost change function ΔU directly.

J = (Rs - Y) ^T (Rs - Y) + ΔU ^T RwΔU (12)

where ΔU = [ΔU(k) ΔU(k+1) ΔU(k+2) ... ΔU(k+Nc-1)] ^T ,

ΔU(k+1) = U(k+1) - U(k), Nc is a number indicating how many times control is performed when predictive control is performed once, and Nc≤Np.

In addition, Rw = rwI (I is Nc x Nc identity matrix), and rw (≥0) is a control parameter that determines the weight of ^{(Rs - Y) T} (Rs - Y) and ΔU ^{T RwΔU terms in the cost function.}

When the cost function is determined by the environment setting unit 200, the predictive control calculation unit 300 calculates a predictive control variable U(k) equal to the number of predicted samples by using the cost function.

In addition, the environment setting unit 200 may set the following expression as a constraint condition.

(13)

(13)

0≤GRID(k)≤PEAK_LIMIT (PEAK_LIMIT is the limiting peak power)

Hereinafter, a detailed prediction control variable calculation process in the prediction control calculation unit 300 will be described.

(9) by the formula

X(k+1|k) = AX(k) + BΔU(k)

X(k+2|k) = A ² X(k) + ABΔU(k) + BΔU(k+1)

…

SAVE(k+1|k) = CAX(k) + CBΔU(k)

SAVE(k+2|k) = CA ² X(k) + CABΔU(k) +CBΔU(k+1)

SAVE(k+3|k) = CA ³ X(k) + CA ² BΔU(k) +CABΔU(k+1) + CBΔU(k+2)

…

become,

ΔU = [ΔU(k) ΔU(k+1) ΔU(k+2) … ΔU(k+Nc-1)] ^T

If indicated as

Y = FX(k) + ΦΔU (14)

organized in the form

here,

F = [CA CA ² CA ³ … CA ^Np ] ^T

to be.

In the predictive control calculation unit 300, the cost function

J = (Rs - Y) ^T (Rs - Y) + ΔU ^T Calculate the optimized value of ΔU by partial differentiation of RwΔU with ΔU. In other words,

If you calculate

ΔU = (Φ ^T Φ+Rw) ^-1 Φ ^T (Rs-FX(k)) (15)

This is calculated

The predictive control calculation unit 300 calculates the optimized ΔU using the above equation, and accordingly, ΔU(k+1) according to one predictive control sampling number is automatically calculated.

4 shows a prediction control graph generated through such a process in the prediction control calculation unit 300 . In the battery graph, if it is less than 0, it indicates discharge, and if it is greater than 0, it indicates the charging schedule.

Here, it can be seen that the GRID(k) value of FIG. 4 does not exceed the peak value in all sections (T1 to T6) in which the load prediction value exceeds the limit set value (2.199kW) in FIG. 3 . This means that the peak load is effectively reduced due to the battery charge/discharge schedule by the predictive control variable according to the present invention.

On the other hand, in this embodiment, the target function is

(16)

(16)

is set to

Meanwhile, in the control variable setting unit, the maximum value Cap of aBAT(k) may be set, and 0 ≤ aBAT(k) ≤ Cap. In addition, in this embodiment, Cap was set to 6 kWh ≤ Cap ≤ 60 kWh.

When Rs is set, the control variable aU(k) is controlled such that the difference between the output variable and the saving amount SAVE(k) with Rs is as small as possible as shown in FIG. 6 . Therefore, according to the present invention, it is possible to reduce the peak load and control the saving amount to the maximum by the predictive control variable.

2 to 6 show a simulation experiment in which an ESS charge/discharge schedule is generated using the ESS charge/discharge schedule generation system according to this embodiment, and together with this, how much operation gain (SAVE) occurs was investigated.

In this experiment, the predicted load power and the predicted solar power generation were set as shown in FIG. 2, and the limit peak power was set to 30.609 kW, which is 90% of the maximum load.

As a result, as shown in FIG. 3 , the predictive control calculation unit generated a predictive control graph for the ESS battery. In the graph, when the battery is less than 0, it indicates a discharge, and when it is greater than 0, the charging schedule is indicated.

4 shows a prediction graph for the resulting state variables, and FIG. 5 shows the accumulated battery capacity. And, FIG. 6 shows a graph showing the difference between the target function (SET-POINT) and the output variable.

As described above, according to this experiment, the predictive control calculation unit generates an ESS charge/discharge schedule that satisfies the constraint (13) and provides an ESS optimal operation technique that maximizes the operating profit.

Although the above has been described with reference to the embodiments of the present invention, those of ordinary skill in the art can variously modify and modify the present invention within the scope without departing from the spirit and scope of the present invention described in the claims below. You will understand that it can be changed.

100: data collection unit 200: environment setting unit
210: control variable setting unit 220: state variable setting unit
230: output variable setting unit 240: target function setting unit
250: output function setting unit 260: cost function setting unit
300: predictive control calculation unit

Claims (5)

a data collection unit for receiving and storing power data including price variables for each time period;
an environment setting unit for setting a variable including a control variable, a state variable, an output variable, an output function, and a function including a cost function based on the data stored in the data collection unit;
It is configured to include a predictive control calculation unit for generating control variables for each time period using the variables and functions set in the environment setting unit,
The environment setting unit,
BAT(k) is the amount of battery charge/discharge by time period, GRID(k) is the amount of electricity input to the grid by time period, LOAD(k) is the predicted load energy by time period, aBAT(k) is the accumulated battery charge/discharge amount, and aGRID(k) is the accumulated system Incoming electricity rate, aLOAD(k) is the cumulative predicted load electricity rate, k is a variable representing the sampling sequence according to time, PV(k) is the predicted solar power generation amount by time period, SELL(k) is the solar power generation sales volume by time period, aPV(k) is the cumulative predicted solar power generation, and aSELL(k) is the cumulative solar power generation sales.
Column vectors (BAT(k), GRID(k), LOAD(k), PV(k), SELL(k)) ^T is a control variable that is actually controlled by the system, and column vectors (aBAT(k), aGRID(k) ), aLOAD(k), aPV(k), aSELL(k)) ^T are state variables that are physical quantities that change according to the control variable, and the amount of time saved based on the state variable is set as the output variable SAVE(k) and,
Setting a target function Rs for the savings amount and an output function Y including the savings amount for each time period,
The predictive control calculation unit calculates the difference between the control variables for each time period so that the cost function is minimized based on the cost function J using the difference between the target function and the output function and the difference between the control variables for each time period, and calculates the control variable for each time period. ESS operating system to create.
According to claim 1,
The environment setting unit,
The relation between the control variable and the state variable
aGRID(k+1) = aGRID(k) + GRID(k)×CHARGE(k)
aBAT(k+1) = aBAT(k)+BAT(k)
aLOAD(k+1) = aLOAD(k) + LOAD(k)×CHARGE(k)
aPV(k+1) = aPV(k) + PV(k)×CHARGE(k)
aSELL(k+1) = aSELL(k) + SELL(k)×SMPREC(k)
(CHARGE(k) is the electricity rate per kWh per hour, SMPREC(k) is the sales amount per kWh per hour).
According to claim 1,
The output variable is
SAVE(k) = aBAT(k)×CONV - aGRID(k) + aLOAD(k) + aSELL(k)
is set, and the output function is
Y = [SAVE(k+1|k) SAVE(k+2|k) … SAVE(k+Np|k)] ^T
(SAVE(k+1|k) is the amount of saving in k+1 hours predicted in k hours, Np is the number of predicted samples per one time when predictive control is performed, and CONV is the electricity charge per kWh for the remaining amount of the battery. ) ESS operating system, characterized in that set to.
According to claim 1,
The target function Rs is

(LOAD(k) is the predicted load power by time period, CHARGE(k) is the electricity rate per kWh for each time period, Np is the predicted number of samples).
receiving and storing power data including price variables for each time from the data collection unit;
setting a variable including a control variable, a state variable, an output variable, an output function, and a function including a cost function based on the data stored in the data collection unit by the environment setting unit;
performing the step of generating control variables for each time period using the variables and functions set in the environment setting unit by the predictive control calculation unit,
The environment setting unit,
BAT(k) is the amount of battery charge/discharge by time period, GRID(k) is the amount of electricity input to the grid by time period, LOAD(k) is the predicted load energy by time period, aBAT(k) is the accumulated battery charge/discharge amount, and aGRID(k) is the accumulated system Incoming electricity rate, aLOAD(k) is the cumulative predicted load electricity rate, k is a variable representing the sampling sequence according to time, PV(k) is the predicted solar power generation amount by time period, SELL(k) is the solar power generation sales volume by time period, aPV(k) is the cumulative predicted solar power generation, and aSELL(k) is the cumulative solar power generation sales.
Column vector (BAT(k), GRID(k), LOAD(k), PV(k), SELL(k)) ^T is a control variable that the system actually controls.
(aBAT(k), aGRID(k), aLOAD(k), aPV(k), aSELL(k)) ^T is a state variable that is a physical quantity that changes according to the control variable. is set as the output variable SAVE(k),
Setting a target function Rs for the savings amount and an output function Y including the savings amount for each time period,
The predictive control calculation unit calculates the difference between the control variables for each time period so that the cost function is minimized based on the cost function J using the difference between the target function and the output function and the difference between the control variables for each time period, and calculates the control variable for each time period. ESS operation method to create.

Images