KR101597993B1 - Energy storage system - Google Patents

Abstract

According to an embodiment of the present invention, an energy storage system allows the number of charging/discharging operations under a condition of a real-time charge system to be restricted and predicts a load and generated amount of new generation energy to decrease a maximum power peak, thereby reducing the charge based on the amount of power. The energy storage system may comprise: a power charging/discharging part; and a charging/discharging control module. The power charging/discharging part charges and discharges power between a battery and a power line. The charging/discharging control module predicts the amount of power required for the next day depending on photovoltaic power generator installation information, weather information and load information to control the power charging/discharging operation of the power charging/discharging part.

Description

[0001] ENERGY STORAGE SYSTEM [0002]

BACKGROUND OF THE INVENTION 1. Field of the Invention [0001] The present invention relates to a technology for storing a renewable energy source such as wind power and sunlight, and more particularly, to an energy storage device for efficiently charging and discharging a renewable energy source stored in an energy storage device .

Generally, the output of a renewable energy source such as wind and sunlight is stored in an energy storage device and then supplied to a grid system through a power conversion device. This is an operation scheduling of an energy storage device for micro grid independent operation, (Time Of Use) rate system, which is linked to wind power and solar power generators.

Considering operational scheduling of the energy storage device, there are many error factors and difficulties, such as a power generation forecast error, a load and power price error of a wind power and a solar power generator, and a state of charge (SOC) Accordingly, conventionally, as described in the following prior art documents, only when renewable energy remains at a specific time, it is operated as a system for charging surplus electric power by using an energy storage device, so that it is suitably applied to a real- There is a difficult problem. In addition, since charging and discharging are performed without limiting the number of charge and discharge cycles of the energy storage device, the life of the energy storage device is rapidly shortened.

Korean Patent Publication No. 10-2013-0003409

According to an embodiment of the present invention, an energy storage device capable of limiting the number of charge / discharge cycles in a real-time charge environment and reducing a maximum power peak by predicting load and renewable energy generation amount to reduce a power charge / RTI >

According to an aspect of the present invention, there is provided an energy storage device including: a power supply for charging / discharging power between a battery and a power line; And a charge / discharge control module for predicting a necessary amount of power of the next day according to the photovoltaic generator installation information, weather information and load information, and controlling the power charge / discharge of the power charge discharge part.

According to an embodiment of the present invention, the charging / discharging control module may include a PV prediction module based on inputted photovoltaic generator installation information and predicting the amount of photovoltaic generation of the next day by stepping the degree of cloudiness among the inputted meteorological information; A load prediction module for predicting an amount of load power of the next day based on inputted load information; And an optimization module for controlling power charging / discharging of the power charging unit according to the prediction information of the PV prediction module and the load prediction module.

According to an embodiment of the present invention, the PV prediction module can estimate the amount of solar power generation according to a value obtained by multiplying a set solar radiation amount by a weighted value according to a stepwise light amount.

According to an embodiment of the present invention, the PV prediction module may include a solar radiation amount prediction module for predicting the solar radiation amount according to input solar generator installation information and a corresponding date and time; A light quantity prediction module for predicting the light quantity according to the input weather information and the predicted solar radiation amount from the solar radiation amount prediction module; And a power generation amount determination module for predicting the solar power generation amount according to the predicted solar radiation amount from the solar radiation amount prediction module and the predicted cloud amount from the lightness prediction module.

According to an embodiment of the present invention, the optimization module may determine a battery output amount according to dynamic programming of prediction information of the PV prediction module and the load prediction module.

According to an embodiment of the present invention, the charge / discharge control module may include a first control module for predicting an optimal generation amount of a next day and providing a charge / discharge control result; A second control module for providing a charge / discharge control result by predicting an optimal generation amount of the same day; And a real-time control module that performs charge / discharge control based on each control result of the first control module and the second control module and a real-time power situation.

According to an embodiment of the present invention, the first control module is based on inputted solar generator installation information, and the next day PV for predicting the solar power generation amount of the next day by stepping the degree of cloudiness among the input weather information Prediction module; A next load prediction module for predicting the load of the next day based on past load data; And a first optimization module for determining an optimal battery output amount of the next day according to the prediction information of the next day PV prediction module and the next day load prediction module.

According to an embodiment of the present invention, the second control module is based on input real-time solar power generation data, and the degree of cloudiness among the input weather information is stepped up to estimate the amount of solar power generation Real - time PV prediction module; A real time load prediction module for predicting the load on the day based on the real time load data and the past load data; And a second optimization module for determining a battery output amount of the same day according to the prediction information of the real time PV prediction module and the real time load prediction module.

According to an embodiment of the present invention, the real-time control module includes a real-time SOC maintenance module that provides a charge / discharge control signal based on the measured power peak, current power consumption, and real-time charge / discharge scheduling output to date; And a real-time peak suppression module that provides a charge-discharge control signal based on the next-day charge-discharge scheduling output and the real-time charge-discharge scheduling output.

According to an embodiment of the present invention, when the measured power peak is greater than the difference between the current consumption power and the real-time charge / discharge scheduling output, the real-time control module controls the power- And controls the output of the power charging unit according to charge / discharge control of the real time peak suppression module when the measured power peak is smaller than the difference between the current consumed power and the real time charge / discharge scheduling output .

According to an embodiment of the present invention, when the measured power peak is smaller than the difference between the current consumed power and the real-time charge and discharge scheduling output, the real-time control module determines that the real- The next day charge / discharge scheduling output and the real-time charge / discharge scheduling output are controlled based on the maximum output of the next day charge / discharge scheduling output and the real-time charge / discharge scheduling output when the real- The output of the power charging unit can be controlled based on the half of the sum of the outputs.

According to an embodiment of the present invention, it is possible to limit the number of charge / discharge cycles in a real-time charge environment, reduce the maximum power peak by predicting the load and the amount of renewable energy generation, and reduce the power charge.

FIG. 1 is a schematic configuration diagram of a photovoltaic power generation system employing an energy storage device according to an embodiment of the present invention. Referring to FIG.
2 is a schematic block diagram of a charge / discharge control module of an energy storage device according to an embodiment of the present invention.
3 is a schematic block diagram of a charge / discharge control module of an energy storage device according to another embodiment of the present invention.
4 is a table showing a coefficient a according to time and cloud in Equation 2 of the present invention.
FIG. 5 is a graph showing a value according to time when it is clear (1), and FIG. 6 is a graph showing a value of coefficient (a) according to time when cloudiness is 2 (2).
FIG. 7 is a graph showing a cubic coefficient for obtaining the coefficient (a) value according to time for each cloud and Equation (3).
8 is a graph of predicted solar power generation and measured solar power generation.
9 is a table showing a Relative Root Mean Square Error (rRMSE) by the cloud.
10 is a graph showing a value of coefficient (a) according to cloudiness at 13 o'clock, for example.
11 is a graph showing a calculation method of the difference in cloudiness based on the value of the real-time coefficient (a).
12 is a schematic block diagram of a cloud prediction module.
13 is a schematic block diagram of a power generation amount determination module.
14 is a schematic block diagram of a real-time PV prediction module.
15 is a graph of the predicted PV and the real-time predicted PV and the measured PV of the next day.
16 is a graph showing a load pattern for each day of the factory.
17 is a graph showing a seasonal load pattern of a factory.
18 is a graph showing a load pattern according to the factory temperature.
Fig. 19 is a graph showing the temperature-load pattern of the factory by time zone. Fig.
20 is a graph showing the load prediction result.
21 is a diagram showing a method for solving dynamic programming.
22 is a schematic block diagram of the next day prediction module.
23 is a schematic block diagram of a real-time prediction module.
24 is a diagram showing an annual electricity rate.
25 is a diagram showing a daily power peak for one month.
26 is a graph showing a peak forming case according to a prediction error.
27 is a schematic block diagram of a real-time SOC maintenance module.
28 is a schematic block diagram of a real-time peak suppression module.
29 and 30 are graphs showing power curves.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings, in order that those skilled in the art can easily carry out the present invention.

FIG. 1 is a schematic configuration diagram of a photovoltaic power generation system employing an energy storage device according to an embodiment of the present invention. Referring to FIG.

Referring to FIG. 1, a photovoltaic generation system employing an energy storage device 100 according to an embodiment of the present invention transmits solar PV generation amount P _PV (t) to a solar photovoltaic generator A, power line (E) and the power (P _GRID (t)) from the transmission, in the energy storage device 100 in accordance with one embodiment of the present invention outputs the charging electric power to the battery (B) or the battery (B) So that the power can be charged. For example, a load C consuming power such as a factory may be supplied with the required power P _LOAD (t).

The energy storage device 100 according to an exemplary embodiment of the present invention may include a power charging unit 110 and a charge and discharge control module 120.

The power charging and discharging unit 110 can charge and discharge electric power between the battery B and the power line and the charge and discharge control module 120 predicts the required amount of power of the next day according to the photovoltaic generator installation information, So that it is possible to control the charge / discharge of the electric power charging / discharging unit 110.

2 is a schematic block diagram of a charge / discharge control module of an energy storage device according to an embodiment of the present invention.

2, the charge / discharge control module 120 of the energy storage device according to an exemplary embodiment of the present invention may include a PV prediction module 121, a load prediction module 122, and an optimization module 123 .

The PV prediction module 121 can predict the solar power generation amount of the next day based on input solar power generator installation information and stepwise the degree of cloudiness among the inputted weather information.

The load prediction module 122 can estimate the load power amount of the next day according to the inputted load information. The load prediction module 122 may receive real-time load data and past load data.

The optimization module 123 may control the power charging / discharging of the power charging unit 110 according to the prediction information of the PV prediction module 121 and the load prediction module 122.

As described above, the PV prediction module 121 can predict the amount of solar power generation on the next day by stepping the degree of cloudiness among the inputted weather information.

In this regard, it is possible to predict the amount of power generation from the amount of solar radiation obtained on a clear day, but the actual amount of power generation is affected not only by the amount of sunshine but also by the daylight hours. Since the sunshine time is negatively correlated with the cloudiness, the higher the cloudiness, the less sunshine time, and the lower the cloudiness, the higher the sunshine time. Even if the amount of sunshine is large, the solar power generation will decrease if the sunshine time is short. Using this, the amount of solar power generation was calculated according to the formula (1) according to the solar radiation (IC) and cloud amount obtained on a clear day.

Equation (1)

In order to determine the amount of electricity generated by the equation (1), a correlation coefficient between the solar power generation amount and the solar radiation amount according to the cloudiness should be obtained. Since the amount of solar radiation is generally known, the detailed description is omitted. In the case of cloudiness, it is possible to acquire information of a certain region of a certain date. For example, the weather agency site acquired the cloudiness of the Cheongju region on the date. At present, the cloudiness of Korea is predicted in four stages of fine (1), cloudy (2), cloudy (3) and cloudy (4) ), Cloudiness (3 ~ 5), cloudiness (6 ~ 8), cloudiness (9 ~ 10). In addition, the rainy day is classified as cloudy because the result is different from the same cloudy day, and the rainy day is classified as cloudy day. Therefore, correlation coefficients between solar power generation and solar radiation, which were classified as total clear, cloudy, cloudy, cloudy, cloudy - rain, cloudy -

Equation (2)

FIG. 5 is a graph showing a value according to time with a fine (1), FIG. 6 is a graph showing a value of a cloud (2) according to time and cloudiness in Equation 2 of the present invention, ), And FIG. 7 is a graph showing a cubic coefficient for obtaining a coefficient a according to time for each cloud and Equation (3).

For example, a value of a according to time in the sixth stage of lightness is found as shown in the table of FIG. When the coefficients (a) are connected to each other at the same light amount, they can be expressed by a cubic function with respect to time as shown in FIG. 5 or FIG. Therefore, even if it is not a specific time like 12 o'clock, if the weather is obtained through the meteorological office and the current time is known, the coefficient (a) can be obtained through the cubic function. The coefficients of the cubic function for each cloud are shown in FIG.

8 is a graph of predicted solar power generation and measured solar power generation.

Prediction results as shown in FIG. 8 can be obtained by estimating the amount of power generation from the solar radiation amount and the coefficient (a). This example is a case where the cloud changes from hour to hour as an example on September 15, 2013. Cloud forecasts were cloudy from 0 to 12, clouds were predicted from 12 to 15 and cloudy after 15. The prediction of solar power generation also predicted that the value will be small until 12 o'clock, Actual results also show that solar power generation prediction accuracy is improved by prediction including cloudiness when power generation is low until 12 o'clock and power generation has increased since then.

9 is a table showing a Relative Root Mean Square Error (rRMSE) by the cloud.

In order to evaluate the accuracy of the PV prediction module 121 of the present invention, a relative root mean square error (rRMSE) was measured. The mean square root relative error is used to determine the accuracy of the new and renewable generators and the method to be obtained is shown in Equation (4).

Equation (4)

For example, the relative square root mean square error between 9 and 18 o'clock of the solar power generation data from 2013.01.01. ~ 2013.10.31. The relative mean square root error was 37.35%. No. 1, No. 1, pp. 183-186, in German literature on photovoltaics (Elke Lorena, Johannes Hurka Detlev Heinemann, and Hans Georg Beyer, Irradiance Forecasting for the POwer Prediction of Grid-Connected Photovoltaic Systems, IEEE Applied Earth Observations and Remote Sensing, Vol. March, 2009), the average square root relative error of predicted solar power generation in an independent solar power generation site averages 36%, which is equivalent to solar power prediction accuracy.

On the rainy day, we can see that the mean square root relative error increases considerably because the smaller the measured quantity, the larger the mean square root relative error value for small differences. For example, if the prediction is 101 kW and the power generation amount is 100 kW, the mean square root relative error value is 1%, but if the prediction is 2 kW and the generation amount is 1 kW, the mean square root relative error value is 100%. Therefore, on the day of rain, the relative amount of power generation is small, so that the relative error of the mean square root is increased.

For example, FIG. 10 is a graph showing a coefficient a according to clouding at 13 o'clock, and Fig. 11 is a graph showing a calculation method of a clouding difference using a real-time coefficient a.

The above-mentioned method has a disadvantage in that it can not take a value between stages because the prediction results are divided stepwise according to the cloudiness. For example, even if the cloudiness value is equal to 2, the actual cloudiness may vary to 1.8 or 2.2, but this can not be reflected.

In addition, forecasts may differ because weather forecasts by the Korea Meteorological Administration predict forecasts for representative points, rather than forecasts of weather forecasts in areas where solar panels are installed. To compensate for this, real-time power generation was measured to compensate for the cloudiness.

The real-time coefficient (a) can be obtained by dividing the real-time power generation amount by the following equation (5) and the solar radiation amount in a clear day.

Equation (5)

The real-time power generation means the generation energy of the photovoltaic generator developed for 15 minutes. Since the peak power is measured every 15 minutes, it is measured every 15 minutes to change the predicted value every 15 minutes. Further, the coefficient (a) value for the cloudiness at that time can be obtained by using FIG. In this case, when the coefficient (a) is the maximum, the cloudiness is selected as 0, and when the coefficient (a) is 0, the cloudiness is selected as 5 and the coefficient (a) according to the cloudiness at the same time can be obtained as shown in FIG. The graph of FIG. 10 can be approximated by a fifth-order equation because it has six points (cloud 0 to 5) and can determine a value for cloudiness between 0 and 5. FIG.

In the case of the real-time coefficient (a), the fifth order equation can be solved to obtain the cloud quantity as shown in FIG. In this case, there is a difference between the predicted and real-time cloudiness, and the next stage cloud difference (cloud difference t) is calculated using the previous stage cloud difference (cloud difference t-2) and the current stage cloud difference (cloud difference t- (6), the predicted cloudiness is corrected as shown in Equation (7), and the value of the coefficient (a) is changed at the remaining time to predict a new solar power generation amount.

Equation (6)

Equation (7)

FIG. 12 is a schematic block diagram of a cloud prediction module, FIG. 13 is a schematic block diagram of a power generation amount determination module, and FIG. 14 is a schematic block diagram of a real time PV prediction module.

The PV prediction module shown in FIG. 14 may include the cloud prediction module shown in FIG. 12 and the power generation amount determination module shown in FIG.

The input of the PV prediction module may be meteorological information such as weather and precipitation received from the meteorological office, date and time, photovoltaic generator installation data (PV installation data), and real time photovoltaic (PV) The amount of solar power can be predicted.

The rRMSE value of the method of predicting the generation amount in real time is 34.92%, which is better than the rRMSE value of 37.35% of the predicted next day generation amount. FIG. 15 is a graph showing predicted solar power generation amount on the next day, real-time predicted solar power generation amount, and measured solar power generation amount, and is a prediction result based on solar power generation data on April 30, It can be seen that it corrects the real - time cloudiness and predicts the result which is quite similar to the measured value after 10:00 pm, and it becomes more accurate than the next day forecast. As shown in FIG. 14, it can be seen that the module for correcting the cloudiness in real time like the real-time PV prediction module is more accurate than the module for predicting the solar power generation amount in the next day, and finally, The amount of photovoltaic generation can be predicted.

18 is a graph showing a load pattern according to a factory temperature, and FIG. 19 is a graph showing a load pattern according to a time-series temperature-load This is a graph representing a pattern.

For example, the day-by-day load pattern of the HHI voice factory from March to May 2013 can be shown in FIG. The difference on the day of the week is small, but the load at dawn on Monday is low compared to other days. However, the purpose of the scheduling is to reduce the electricity bill, and this time period is independent of the peak, and charging is done at dawn. Also, the pattern of lunch time is the same regardless of day of the week. This pattern is a load pattern predicted by removing two bad data (data with the largest error from the load pattern) for each day of the week.

As shown in Fig. 17, the seasons were divided according to the seasons classified by the TOU (Time of Use) plan. (Spring: March to May, summer: June to August, autumn: September to October, winter: November to February) The load pattern shown is a seasonal load pattern. (Load pattern in spring, summer, autumn and winter in the clockwise direction from left to right).

Referring to FIG. 18, it can be seen that there is a large correlation between the temperature and the load. Referring to FIG. 18, it can be seen that as the temperature decreases from a position lower than 16 degrees from about 16 degrees, the load increases with increasing temperature at a position higher than 16 degrees. As shown in FIG. 19, when the temperature is plotted on the x-axis and the y-axis is plotted on a graph basis for each time unit, the correlation of the temperature load varies with time. There is little temperature and load correlation at dawn, and the correlation is clear at 9:30 am just after the work time. If you can see the temperature forecasts over time, you can use the graph below to predict the load.

20 is a graph showing the load prediction result.

Referring to FIG. 20, a load prediction result for a day before scheduling is shown at 24:00 for each date. The error rate is estimated to be within 10%. However, if the vacation season or unexpected situation occurs, the error rate may be worse. There is an advantage in that there is a limit to the prediction of the load before the day because the correlation between the temperature and the load is not used, but the error rate is small in prediction in real time.

As described above, the optimization module for controlling the charge / discharge of the battery based on the predictions of the PV prediction module and the load prediction module will be described.

The optimization module determines the battery charge discharge in the direction of minimizing the electricity charge in the TOU plan. In the TOU plan, the electricity charge is given by the following equation (8), so that the objective function can be expressed by equation (9). Where P _Grid (t) is determined via equation (10). (P _ESSS (t)) of the energy storage device in order to find the PGrid value satisfying the equation (31).

Equation (8)

Equation (9)

Equation (10)

  The constraints on the optimization problem are as follows. Since the output of the battery is limited, a constraint such as Equation (11) below is required. Since the discharge of the battery is positive, the max value will be related to the maximum C-rate when discharging and the min value will be related to the maximum C-rate when charging. For stable operation of the battery, the state of charge (SOC) of the battery must be maintained within a certain range. The change in the state of charge due to the charge is expressed by Equation (12) ), The constraint condition expressed by Equation (14) must be satisfied.

Therefore, the battery output is limited by the battery maximum, minimum output and charge state. The data about the charge state of the previous day is first charged and the final charge state after scheduling should be kept at a specific value so that it can be used as data about the state of charge the next day.

Equation (11)

Equation (12)

Equation (13)

Equation (14)

Also, since the battery output is transmitted through the inverter of the power charging unit 110, the output of the actual energy storage device changes according to the inverter efficiency of the charging unit 110. Expression of this is expressed by Equation (15) when discharging and Equation (16) when charging. Because of battery efficiency, even if the same amount is discharged and discharged, more power is used.

In the case of the factory load of the current target system, since the power consumption is less than the surrounding peak time between 12 and 13 hours, the charge can be additionally performed one more time. However, since the power in this time period is not significantly different from the peak time, Is bad.

Therefore, the number of charge / discharge cycles of the battery was limited to one. In addition, increasing the number of charge / discharge cycles shortens the battery life, so it is the most economical when the battery life is considered.

Equation (15)

Equation (16)

As a result, the optimization module minimizes Equation (9) and is a module that satisfies Equations (11) and (14) and determines the amount of battery power with one charge / discharge cycle. Dynamic programming (DP) was used to solve this optimization problem.

Dynamic programming is one of the techniques for solving optimization problems. It is the method that solves the problem by checking the number of possible cases at the minimum number of times. Therefore, the result is always the optimal result satisfying the objective function.

Dynamic programming is the process of solving a problem by dividing the state at each stage and then relate the state to the state. In this problem, the stage is time and there are 96 stages divided by 15 minutes for a total of 24 hours. The reason for dividing it by 15 minutes is because the power consumption of the present peak power is calculated in 15 minutes. That is, the power of each stage can be the peak power.

21 is a diagram showing a method for solving dynamic programming.

As shown in Fig. 21, the state in each stage can be a charged state. The reason for moving from a higher stage to a lower stage is to use backward dynamic programming. The power of the energy storage device 100 between the i-th state and the (i-1) -th stage can be calculated by Equation (17) for discharging and Equation (18) Can be obtained.

Equation (17)

Equation (18)

Dynamic programming is impossible to obtain instantaneous max (PGrid) because the stage moves step by step.

Therefore, the equation of Equation (9) must be changed to solve the peak power problem using dynamic programming. If the sum of the power values is the same in the other two times as in the equation (19), the equation (20) can be obtained by squaring both sides. That is, the sum of power values is the same, the closer the value between the two powers is (P2 (t) and P3 (t)), the smaller the sum of squares. Therefore, even if the output of the same amount of energy storage device is charged and discharged, discharging at a high region and charging at a low region are minimized to minimize the sum of squares. Since the result is the same as the minimization of the peak, the objective function can be changed as shown in equation (21). The values of C1 and C2 were selected to minimize the electricity cost when optimizing and C1 = 1 and C2 = 5 were selected.

Equation (19)

Equation (20)

Equation (21)

There are two main ways to determine the energy storage output (P _ESS ) through the optimization module. One is a method (method a) for determining the charge amount of a battery using the next day load prediction and the next day solar power generation prediction as shown in FIG. 22 and the other is a method using real time load prediction and real time solar power generation prediction And a method for determining the charge-discharge amount of the battery (method b).

22, the next day PV predicting module 221a predicts the solar power generation amount of the next day based on the installation data (PV installation data) of the PV generator and the next day weather information, and the next day load prediction module 221b The first optimization module 221c can estimate the load of the next day based on the past load data and the output of the energy storage device ESS according to the prediction results of the next day PV prediction module 221a and the next day prediction module 221b. Can be determined.

Referring to FIG. 23, the real-time PV prediction module 222a can predict the amount of solar power generation every 15 minutes based on the real-time PV data and the weather information on the day, The second optimization module 222c can predict the load of the energy storage device ESS according to the prediction results of the real time PV prediction module 222a and the real time load prediction module 222b, Can be determined.

To determine which of the two methods is dominant, we estimate the load and solar power generation based on measurement data from November 01, 2012 to December 31, 2012, and obtain the output of the energy storage device through the optimization module As shown in Table 24, the electricity bill for one year was calculated.

The base rate used the peak of the month, and the optimal result is the result of applying the optimization module as the measurement result, not the prediction. Changing the prediction in real time makes the prediction result better, so the output of the energy storage device can be changed appropriately according to the situation.

Fig. 24 is a diagram showing annual electricity charges, and Fig. 25 is a diagram showing power peaks per day of a month.

Referring to FIG. 24, the method a yields a profit of 21,721,000 won per year rather than without an energy storage device. However, the method b yields a profit of 22,356,000 won per year, . Because the predictions differed from the measurements, both methods failed to yield optimal results.

As shown in FIG. 25, when the method b is compared with the method a for one year and the method b is always superior to the method a, in reality, when the load pattern or the photovoltaic pattern is wrong, the prediction error is amplified in real- There was also a day when method b was worse. 25, the peak power of method b was lower in most cases as a result of the date by May 2013, but in case of method 2, 3, 8, 15, 24 and 27, . If such a prediction is missed, the next day optimization result is sometimes better.

In the method of determining the output of the two energy storage devices obtained by the optimization module, the method b was relatively superior, but in some cases, the method a was superior in some cases. This is because there is a difference between the predicted pattern and the actual pattern.

As shown in Fig. 26, there are cases 1 where the peaks are different from the predictions in the front side and case 2 where the peaks are generated in the rear side unlike the prediction. case 2 can solve the peak problem by leaving the battery charge amount at the front peak as much as possible and case 1 can be solved by comparing the output results of method a and method b and outputting more output . It is important to determine the output considering these two points because there are many conflicting parts.

27 is a schematic block diagram of a real-time SOC maintenance module.

First, when the sum of the output of the energy storage device through the method b and the load power measured in real time and the photovoltaic power generated by the method b is smaller than the peak power so far, . That is, the first to fourth arithmetic units 223a-1, 223a-2, 223a-3, and 223a-4 subtract the current consumed power from the output power according to the real-time scheduling of the energy storage device, The output of the energy storage device can be determined by subtracting the result of the integration from the output power according to the real time scheduling of the energy storage device.

This can prevent the energy storage device from discharging more power than the resulting power peaks and maintain the battery's charge state. However, the result should not be smaller than 0 so that the charge / discharge results do not change.

28 is a schematic block diagram of a real-time peak suppression module.

Next, in contrast to the description of FIG. 27, there is a case where the sum of the output of the energy storage device and the measured load power and photovoltaic power in real time is larger than the peak measured to date. In this case, the output of the energy storage device is determined according to the remaining battery charge amount as shown in FIG.

If the battery charging amount is sufficient through the fifth calculating unit 223b-1, the higher value is selected from the two methods. If the real time battery charging state is less than the next battery charging state, the sixth and seventh The average of the two results is used through the calculation units 223b-2 and 223b-3.

3 is a schematic block diagram of a charge / discharge control module of an energy storage device according to another embodiment of the present invention.

Referring to FIG. 3, the charge / discharge control module of the energy storage device according to another embodiment of the present invention having the real-time control module includes a first control module 221, a second control module 222, 223).

22, the first control module 221 may include a next day PV prediction module 221a, a next day load prediction module 221b, and a first optimization module 221c,

23, the second control module 222 may include a real-time PV prediction module 222a, a real-time load prediction module 222b, and a second optimization module 222c.

The real time control module 223 determines the energy level of the second control module 222 based on the next day result of the first optimization module 221c of the first control module 221 and the real time optimal result of the second optimization module 222c of the second control module 222 The final output of the storage device can be controlled.

The real-time control module 223 may include the real-time SOC module 223a shown in FIG. 27 and the real-time peak suppression module 223b shown in FIG.

A method of operating the charge / discharge control module of the energy storage device according to another embodiment of the present invention will be described as method c.

Method c is able to cope more flexibly when the peak power is changed due to the deviation of prediction through the real-time control module.

In order to find out the result, the ESS output was obtained based on the measurement data from Nov. 01 to 2013. 12. 31, 2012 as in the optimization module, and the electric bill was calculated for one year as shown in FIG. Compared with methods a and b, annual gain increased by more than 1 million won. The real-time module shows that the optimization module results are improved. In the appendix, the daily peak results show that method c is superior to methods a and b.

29 and 30 are graphs showing power curves.

FIG. 29 shows the peak decrease amount by the method on Dec. 11, 2012, and FIG. 30 shows the power on the same day peak time. It can be seen that method c makes the peaks more flat than the other methods and it shows that the peaks are reduced more than the other methods. In this example, the peaks are reduced a lot in the order of methods c, b, a, and the method c is most similar to the optimal result.

As described above, according to the present invention, business models that can be obtained when an energy storage device is connected to a factory are classified into peak power saving, electricity price difference transaction, emergency power supply, demand response, and reactive power compensation.

Since the current electricity rate system for the factory is a TOU rate system, it is possible to trade electricity tariffs because different electricity rates are applied over time. Also, because peak electricity is reflected in the electricity bill, economic benefits can be obtained if the peak power is reduced.

Demand responses can also benefit from contracts with KEPCO. However, in the case of emergency power supply, the economics evaluation model is still insufficient, and in the case of reactive power compensation, there is not established a system for guaranteeing profit, which is not suitable to be used as a current business model. Therefore, we selected the function of the EMS linked to the factory ESS by the profit generation model using the peak power reduction and electricity price difference transaction suitable for the TOU plan.

 The target grid to which the energy storage device is installed is a factory complex, which includes PV, load, and energy storage devices. ESS output should be determined from PV and load power since the electric charge is settled by measuring power coming into the factory. Since the energy storage device must determine the output within a predetermined battery charge level, it is important to determine the output of the energy storage device in accordance with the prediction of the solar power generation power and the load power so as to maintain the battery charge amount.

As a result, there are four PV modules, a load prediction module, an optimization module, and a real-time control module, and the output of the energy storage device is determined in the optimization module and the real-time control module based on the results of the prediction module.

  Solar power generation forecast firstly obtains the solar radiation of sunny day according to the date and time according to the installed hardness of the solar power generator, the latitude and the installation angle of the solar panel. Next, we divided into 6 stages according to cloudiness and precipitation, and the correlation coefficient between solar radiation amount and power generation amount corresponding to each step was obtained by time zone. Based on the correlation coefficient thus obtained, we predicted the amount of electricity generated by the cloud and the time. In real - time forecasting, correlation is applied by estimating the real - time power generation based on continuous variable instead of 6 steps. As a result of prediction of solar power generation, the rRMSE value is 34.92%, which is similar to the general solar power generation rRMSE value (36%).

  We used the pattern method to predict the load using past load pattern data. Patterns were classified according to day, season, and temperature. Day of week was excluded because it had no characteristics. Because the pattern characteristics for the season and temperature are similar to the 30-day weekday load data just before the forecast date, the load pattern was formed using the 30-day weekday load data just before the forecast date. When estimating the load in real time, the correction was made using the ratio in the immediately preceding time. As a result of the load prediction, the error rate was predicted within 10%.

  The optimization module is a module that determines the output of the energy storage device that minimizes the electricity charge on the basis of the predicted load data and solar power generation data. The optimization problem is solved with dynamic programming and the optimal value can always be guaranteed since it is the way to find the answer by searching all possible paths. Also, in the process of obtaining the optimal solution, the charging state of the battery and the restriction condition of the battery output are added so that the resultant value can be obtained in a region where the battery can be physically operated. By using the optimization module, the output of the energy storage device was determined by two methods, the next day and the real time.

  The real-time control module is a module that determines the final output of the energy storage device based on the next-day, real-time optimization result obtained above. In many cases, the real-time optimization results guarantee higher profits, but in a few cases, the next-day optimization results may be more profitable if the prediction pattern deviates. In order to compensate for this, when the power suddenly increases and becomes higher than the peak, a higher value is selected among the optimization results.

However, if the battery charge is insufficient to maintain the charge of the battery, the average of the two results is used. In addition, when the output value of the energy storage device is reduced more than the peak measured to date, the charge amount of the battery is maintained by reducing the discharge amount of the energy storage device. As a result, more than 1 million won a year was generated more than when using the optimization module alone.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory and are not intended to limit the invention to the particular forms disclosed. It will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.

100: Energy storage device
110: full power supply
120, 220: charge / discharge control module
121: PV prediction module
121a: solar radiation prediction module
121b: cloud prediction module
121c: Power generation determination module
122: load prediction module
123: Optimization module
221: first control module
221a: next day PV prediction module
221b: Next-day load prediction module
221c: first optimization module
222: second control module
222a: Real-time PC prediction module
222b: Real-time load example Cmr module
222c: second optimization module
223: Real-time control module
223a: Real-time SOC maintenance module
223a-1, 223a-2, 223a-3, 223a-4:
223b: real-time peak suppression module
223b-1, 223b-2, 223b-3: fifth to seventh operation units

Claims (11)

A power charging / discharging unit for charging / discharging power between the battery and the power line; And
And a charge / discharge control module for predicting a necessary amount of power of the next day according to the photovoltaic installation information, weather information and load information, predicting the optimum generation amount of the next day and the optimum generation amount of the next day to control the electric charge / and,
The charge / discharge control module
A first control module for predicting the optimal generation amount of the next day and providing a charge /
A second control module for predicting the optimum generation amount of the day and providing a charge / discharge control result; And
And a real-time control module that performs charge / discharge control based on each control result of the first control module and the second control module and a real-time power situation.
The method according to claim 1,
The charge / discharge control module
A PV prediction module based on inputted photovoltaic generator installation information and predicting the amount of photovoltaic generation on the next day by stepping the degree of cloudiness among the inputted meteorological information;
A load prediction module for predicting an amount of load power of the next day based on inputted load information; And
An optimization module for controlling power charging / discharging of the power charging unit according to the prediction information of the PV prediction module and the load prediction module,
≪ / RTI >
3. The method of claim 2,
Wherein the PV prediction module predicts a solar power generation amount according to a value obtained by multiplying a set solar radiation amount by a weighted value according to a stepwise light amount.
The method of claim 3,
The PV prediction module
A solar radiation amount prediction module for estimating the solar radiation amount according to the installed solar generator installation information and the date and time;
A light quantity prediction module for predicting the light quantity according to the input weather information and the predicted solar radiation amount from the solar radiation amount prediction module; And
A power generation amount determination module for predicting a solar power generation amount in accordance with a predicted solar radiation amount from the solar radiation amount prediction module and a predicted cloud amount from the lightness prediction module,
≪ / RTI >
3. The method of claim 2,
Wherein the optimization module determines a battery output amount according to dynamic programming of prediction information of the PV prediction module and the load prediction module.
delete The method according to claim 1,
The first control module
A next day PV prediction module based on the input photovoltaic generator installation information, for predicting the amount of solar power generation on the next day by stepping the degree of cloudiness among the input weather information of the next day;
A next load prediction module for predicting the load of the next day based on past load data; And
A first optimization module for determining an amount of battery output of the next day according to prediction information of the next day PV prediction module and the next day load prediction module,
≪ / RTI >
The method according to claim 1,
The second control module
A real-time PV prediction module based on inputted real-time photovoltaic power generation data, for predicting the amount of solar power generation on a given time basis by stepping the degree of cloudiness among input weather information;
A real time load prediction module for predicting the load on the day based on the real time load data and the past load data; And
A second optimization module for determining the amount of battery power of the day according to the prediction information of the real-time PV prediction module and the real-
≪ / RTI >
The method according to claim 1,
The real-
A real time SOC maintenance module for providing a charge / discharge control signal based on a measured power peak, current power consumption, and real time charge / discharge scheduling output; And
Real-time peak suppression module that provides a charge / discharge control signal based on the next-day charge / discharge scheduling output and the real-time charge /
≪ / RTI >
10. The method of claim 9,
The real-
Controlling the output of the power charging unit according to the charging / discharging control of the real-time SOC maintaining module when the measured power peak is greater than the difference between the current consumed power and the real-time charge / discharge scheduling output,
Wherein the control unit controls the output of the power charging unit according to charge / discharge control of the real time peak suppression module when the measured power peak is smaller than the difference between the current consumption power and the real time charge / discharge scheduling output.
11. The method of claim 10,
When the measured power peak is smaller than the difference between the current consumed power and the real-time charge / discharge scheduling output
When the real-time battery charge state is equal to or greater than the predicted next-day battery charge state, the output of the power charge discharge unit is controlled based on the maximum output of the next day charge / discharge scheduling output and the real-
And controls the output of the power charging unit based on a half of a sum of the next day charge / discharge scheduling output and the real time charge / discharge scheduling output when the real time battery charge state is less than the predicted next day battery charge state.

Images