WO2021015332A1 - Solar power generation and control system, and method for operating solar power generation and control system - Google Patents

Abstract

Disclosed are a system including a solar power generation system and a control server, and a method for operating a solar power generation and control system for diagnosing a failure of the solar power generation system and predicting a power generation amount by using the control server. A solar power generation system of the present invention provides: a solar power generation and control system comprising one or more solar power generation units, a control server communicably connected to the one or more solar power generation units in a wired or wireless manner, and one or more manager terminals communicably connected to the control server in a wired or wireless manner; and a method for operating the solar power generation and control system, the method including a function capable of diagnosing a failure of the solar power generation and control system and predicting a solar power generation amount by using the solar power generation and control system.

Description

Solar power generation and control system, and operating method of photovoltaic power generation and control system

The present invention relates to a method of operating a photovoltaic power generation and control system, and more particularly, a system including a photovoltaic power generation system and a control server, and a fault diagnosis and generation amount prediction of a photovoltaic power generation system using the control server. It relates to a method of operating a solar power generation and control system to implement.

As environmental issues emerge around the world, technology development for alternative energy is actively progressing, and the most representative is solar power generation. This is because solar power generation is basically environmentally friendly, pollution-free energy, and does not take much effort to receive energy.

However, photovoltaic power generation is currently difficult to be utilized as a major energy supply source. Because photovoltaic power generation has a variety of changes in power production, weather, season, wind speed, sunlight, etc. due to the difference in availability of sunlight as an energy source by time. This is because it inherently has uncertainty in the amount of electricity produced due to changes in natural conditions.

Therefore, in order to analyze the characteristics of solar energy, the correlation between climate conditions, cloud conditions, and power generation performance is derived using technology such as machine learning, and after predicting the module status using a module deterioration model, solar power generation is predicted. There is a need to develop a system that can be used in the future.

The present invention solves the problems of the prior art as described above, and provides a system capable of controlling it in solar power generation, and a fault diagnosis function and a solar power generation amount prediction function of the photovoltaic power generation system using the photovoltaic power generation system. Its purpose is to provide a method of operating a solar power generation and control system including.

The present invention in order to achieve the object of the present invention as described above,

An aspect comprising one or more photovoltaic power generation units, a control server that is communicatively connected to the at least one photovoltaic power generation unit by wire or wireless, and at least one manager terminal that is communicatively connected to the control server by wire or wirelessly As a photovoltaic power generation and control system,

The solar power generation unit solar cell power generation unit including one or more solar cells; A light amount sensor capable of measuring the amount of light; And a solar cell unit including a solar cell measuring unit capable of measuring the power generation voltage and current of the solar cells in the solar cell generating unit. A storage battery storage unit including one or more individual storage batteries to store electric energy produced by power generation by the solar cell unit; And a storage battery unit including a storage battery measuring unit capable of measuring a storage battery voltage of individual storage batteries in the storage battery storage unit and an amount of current charged in the storage battery. And a communication unit including a communication means and a program capable of communicating with the control server by wire or wirelessly,

The control server may include a failure determination unit for individually diagnosing and determining a failure of each of the one or more solar power generation units to individually control the at least one solar power generation unit; A statistics creation unit for creating statistics on one or more data and values related to the one or more solar power generation units; A data collection unit storing data by referring to an external telecommunication network; And a control unit including a power generation amount learning unit capable of predicting a future solar power generation amount using a deep neural network method. A system status DB for updating and storing the status of each of the one or more solar power generation units; A system statistics DB for storing one or more statistical information generated by the statistics generating unit; An external environment DB in which the data collection unit stores data stored by referring to an external telecommunication network; And a server DB including a learning DB for storing data generated according to the operation of the power generation learning unit 214; And it provides a photovoltaic power generation and control system including the one or more photovoltaic power generation units and a server communication unit that is connected to the at least one manager terminal in a wired or wireless manner.

In the above, the solar cell measurement unit solar cell generation voltage measurement unit capable of individually measuring the generated voltage of the one or more solar cells; And a solar cell current measuring unit capable of individually measuring currents of the one or more solar cells, wherein the storage battery measuring unit includes a storage battery voltage measuring unit capable of measuring a voltage of the one or more individual storage cells; And it includes a storage battery charging current measuring unit capable of measuring the charging current of the one or more individual storage batteries.

In the above, it is preferable that the control server includes at least two computers, and a distributed processing program is installed in at least two computers included in the control server so that the control server is implemented by the distributed processing program.

It is preferable to use the distributed processing Hadoop software framework.

In the above, it is preferable that the system statistics DB includes a power generation DB, a carbon dioxide reduction total DB, a power generation cost DB, a power generation time DB, an inverter operation rate DB, and a conversion efficiency DB.

In the above, it is preferable that the external environment DB includes solar radiation DB, precipitation DB, temperature DB, total cloud amount DB, humidity DB, and fog DB.

In the above, it is preferable that the learning DB includes a formation model DB, a training set DB, a test set DB, and a blind set DB.

A method of operating the solar power generation and control system, comprising: a power generation voltage measurement step (S100) of measuring and determining the power generation voltage (V) of any one of the photovoltaic power generation units; After the step (S100), the storage battery current amount measuring step (S110) of measuring and determining the storage battery current amount (Bat_C) of the solar power generation unit to be measured in the step (S100), and after the step (S100), the step (S100) ), the light amount measurement step (S120) of measuring and determining the light amount (R) of the solar power generation unit to be measured is performed to diagnose the failure.

In the step (S100), if the power generation voltage (V) is determined to be less than the current storage battery voltage value (Bat_V) rather than 0V, a normal operation check step (S101) is performed to confirm that the solar power generation unit to be measured is in normal operation, and the step Failure diagnosis is terminated without performing (S110, S120).

In the step (S100), if the power generation voltage (V) is not 0V but exceeds the current storage battery voltage value (Bat_V), and in the step (S110), the storage battery current amount (Bat_C) is 0A or less, among the solar power generation units to be measured A solar cell current measuring unit problem checking step (S111) is performed in which it is determined that the current measuring unit of the solar cell is broken and notified to the administrator's terminal, and the fault diagnosis is terminated without performing the step (S120).

If the power generation voltage (V) is 0V in the step (S100), the storage battery current amount (Bat_C) is more than 0A in the step (S110), and the light amount (R) in the step (S120) is more than 20W/m² Among the solar power generation units to be measured, the current measurement unit of the solar cell is determined to have failed, and the problem confirmation step (S121) of the solar cell current measurement unit informing the administrator's terminal is performed, and the diagnosis is terminated.

The power generation voltage (V) is 0V in the step (S100), the storage battery current amount (Bat_C) is more than 0A in the step (S110), and the light amount (R) in the step (S120) is less than 20W/m2 Or, the power generation voltage (V) is 0V in the step (S100), the storage battery current amount (Bat_C) is 0A or less in the step (S110), and the light amount (R) in the step (S120) is 20W/ If it exceeds ㎡, it is determined that the voltage measurement unit and current measurement unit of the solar cell among the photovoltaic power generation units to be measured have failed, and conducts a problem check step (S122) of the solar cell voltage and current measurement unit notifying the manager's terminal and diagnoses the failure. It ends.

The power generation voltage (V) is 0V in the step (S100), the storage battery current amount (Bat_C) is 0A or less in the step (S110), and the light amount (R) in the step (S120) is 20W/m² or less. Or, in the step (S100), the power generation voltage (V) is not 0V and is less than the current storage battery voltage value (Bat_V), and in the step (S110), the storage battery current amount (Bat_C) is greater than 0A, and the step (S120) If the amount of light (R) in is 20W/m2 or more, a normal operation confirmation step (S123) is performed to confirm that the solar power generation unit to be measured is in normal operation, and the fault diagnosis is terminated.

In the step (S100), the power generation voltage (V) is not 0V but less than the current storage battery voltage value (Bat_V), the storage battery current amount (Bat_C) in the step (S110) is greater than 0A, and in the step (S120) If the amount of light R is less than 20W/m 2, it is determined that the light amount sensor of the solar power generation unit to be measured has failed, and a light amount sensor problem checking step (S124) of notifying the administrator's terminal is performed, and the diagnosis of the problem is terminated.

A method of operating the solar power generation and control system, comprising: a data procurement step (S210) in which the data collection unit collects data from an external telecommunication network and updates and stores it in the external environment DB; A data mining step (S220) of producing raw model data PD by performing data mining on the updated data in the external environment DB; A model training step (S230) performed at least once or more on the raw model data PD generated by performing the step (S220); Then, after executing the step (S230) at least once, testing is performed to generate the predictive model data (CD), and in the data mining step (S220), the processed data is also a training set, a test set, and a blind The data purification step (S221) of forming a set is performed to predict the amount of solar power generation.

In the above, the model training step (S230) is a numerical selection and input step (S231), a hidden layer (HL) and a manager selects a numerical value for the incoming raw model data (PD) and inputs the selected numerical value and learning pattern pair. The output calculation step (S232) for calculating the input weighted sum and the final output of the output layer (OL), the error calculation step (S233) for calculating the error signal value, and the connection strength variation calculation to obtain the connection strength to be used in the next learning step Step S234 is included, and if the number of times the model training step S230 is performed after the step S234 is performed is less than a predetermined number of epochs or more than a pre-input maximum error rate, the step S231 is performed again.

The photovoltaic power generation and control system in the present invention effectively determines a malfunction of the photovoltaic power generation unit and predicts the amount of power generation to ensure effective operation of the photovoltaic power generation unit.

1 is a structural diagram of a solar power generation unit of the present invention.

Figure 2 is a structural diagram of the control server of the present invention.

Figure 3 is a flow chart showing a fault diagnosis sequence in the solar power generation system of the present invention.

Figure 4 is a schematic flow chart showing the solar power generation amount prediction procedure of the present invention.

5 is a structural diagram of a neural network for predicting solar power generation according to the present invention.

Figure 6 is a model training structure diagram in the solar power generation prediction sequence of the present invention.

[Explanation of code]

100: solar power generation unit. 110: solar cell unit.

111: solar cell power generation unit. 112: light quantity sensor.

113: solar cell measuring unit. 1131: Solar cell generation voltage measurement unit.

1132: solar cell current measuring unit. 120: storage battery unit.

121: storage battery storage unit. 1221: Storage battery voltage measuring unit.

1222: Storage battery charging current measurement unit. 130: Communication Department.

200: control server. 210: control unit.

211: Failure determination section. 212: Statistics preparation department.

213: Data collection unit. 214: Power generation learning department.

220: Server DB. 221: System status DB.

2211, 2212: System individual DB. 222: System status DB.

2221: Power generation DB. 2222: Total carbon dioxide reduction DB.

2223: Power generation cost DB. 2224: Power generation time DB.

2225: Inverter operation rate DB. 2226: Conversion efficiency DB.

223: External environment DB. 2231: Insolation DB.

2232: precipitation DB. 2233: Temperature DB.

2234: Total cloud volume DB. 2235: humidity DB.

2236: fog DB. 224: Learning DB.

2241: Formation model DB. 2242: Training set DB.

2243: Test set DB. 2244: blind set DB.

230: Server communication unit. 300: terminal.

Hereinafter, the present invention will be described in more detail with reference to the accompanying drawings. The following description is intended to aid in the practice and understanding of the present invention, but is not intended to limit the present invention. Those skilled in the art will understand that various modifications and changes can be made within the spirit of the invention as set forth in the following claims.

1 is a structural diagram of a solar power generation and control system of the present invention. Hereinafter, components of the solar power generation system of the present invention will be briefly described with reference to FIG. 1.

As shown in Figure 1, the solar power generation and control system of the present invention is connected to one or more solar power generation unit 100, the at least one photovoltaic power generation unit 100 and the wired or wireless communication is possible to this In order to be able to check the control status information of the control server 200 for controlling, and the one or more photovoltaic power generation units 100 transmitted from the control server 200, communication with the control server 200 is possible. It includes at least one manager's terminal 300 connected by wire or wirelessly.

In FIG. 1, the connection with a dotted line means that they are connected to each other through wired or wireless communication.

Prior to the description, in the present invention, the photovoltaic power generation unit 100 includes one or more, but one of the photovoltaic power generation unit 100 is included in the control server 200 as disclosed in FIG. 1 for convenience of description. It will be described with an example that is connected to enable communication. Even if two or more of the photovoltaic power generation units 100 are included, each component may be the same.

In addition, in the present invention, the terminal 300 may use any terminal capable of installing programs such as a desktop PC, a smartphone, a tablet PC, or accessing a web page.

The solar power generation unit 100 includes a solar cell unit 110, a storage battery unit 120, and a communication unit 130, as shown in FIG. 1.

The solar cell unit 110 is a part that generates electric energy by actually generating electricity from sunlight, and the solar cell generator 111 including one or more solar cells 1111 and 1112, the intensity of sunlight, That is, it includes a light amount sensor 112 capable of measuring the amount of light, and a solar cell measuring unit 113 capable of measuring the power generation voltage and current of individual solar cells belonging to the solar cell power generation unit 111. .

Preferably, the solar cell measurement unit 113 includes a solar cell generation voltage measurement unit 1131 capable of measuring the generated voltage of the individual solar cells, and a solar cell current capable of measuring the current of the individual solar cells. It includes a measurement unit 1132.

In addition, the storage battery unit 120 is for storing electric energy produced by power generation by the solar battery unit 110, and includes one or more individual storage batteries 1211 and 1212 in order to actually store electric energy. It includes a storage battery storage unit 121, and a storage battery measuring unit 122 capable of measuring the storage battery voltage and the amount of current charged in the storage battery of the individual storage batteries in the storage battery storage unit 121.

Preferably, the storage battery measurement unit 122 includes a storage battery voltage measurement unit 1221 capable of measuring the voltage of the individual storage batteries, and a storage battery charging current measurement unit 1222 capable of measuring the charging current of the individual storage batteries. Includes.

Further, the communication unit 130 includes a communication means and a program capable of communicating with the control server 200 by wire or wirelessly, and a description thereof will be omitted since the communication method may be performed using a conventional method.

In addition, the solar power generation unit 100 includes all of the components used as common components in general solar power generation and ESS (Energy storage system), such as inverters, distribution boards, and converters, in addition to the above components. Components are well known for their functions and operating methods, and thus a description thereof will be omitted.

The control server 200 is a control unit 120 capable of controlling each of the one or more solar power generation units 100, which are connected to each other through wired or wireless communication, and the at least one photovoltaic power generation unit 100 It includes a server DB 220 in which each control data is stored, and a server communication unit 230 including communication means and programs capable of communicating with the one or more solar power generation units 100 by wire or wirelessly. .

2 is a structural diagram showing a specific configuration of the control server 200. Hereinafter, specific components of the control server 200 will be described with reference to FIG. 2.

Prior to the description, as described above, the photovoltaic power generation and control system of the present invention includes one or more photovoltaic power generation systems and a manager terminal. Hereinafter, for convenience of explanation, the control server 200 includes three photovoltaic devices. It will be described as an example that the power generation system (100a ~ 100c) is connected.

In addition, the control server 200 of the present invention includes one or more computers including one or more computing devices and one or more storage devices in order to actually implement the components of the control server 200 to be described below. In this case, it is preferable to use a server computer, but other types of computers such as a general desktop PC or tablet may also be used.

And in the implementation of the control server 200 of the present invention, the control server 200 may be implemented using one single computer, but for more economical implementation and smooth data processing, two or more computers may be used. It is preferable to implement the control server 200 of the present invention by organizing.

The control server 200 is a control unit 210 for individually controlling one or more solar power generation units 100a to 100c included in the present invention, and a control unit for controlling the one or more solar power generation units 100a to 100c. Server DB 220 in which various data and values are stored, and a server communication unit for communicating by wired or wireless communication with the one or more photovoltaic power generation units 100a to 100c and the administrator's terminal 300 Includes 230.

The above components 210, 220, 230 include one or more computing devices and storage devices for implementing the above functions, and one or more built-in or installed programs for operation.

When the above components 210, 220, 230 are described in more detail, the control unit 210 individually diagnoses and determines a failure of each of the one or more solar power generation units 100a to 100c. The determination unit 211, a statistics creation unit 212 for updating and storing in the server DB 220 by creating statistics on various data and values related to the one or more solar power generation units 100a to 100c, the server A data collection unit 213 for storing data and values transmitted from an external telecommunication network that may be introduced from the communication unit 230 and stored in the server DB 220 for updating and storing in the server DB 220 It includes a power generation learning unit 214 for predicting future solar power generation by using a deep neural network among deep learning algorithms based on various data and values.

And the DB 220 includes a system status DB (221). The system status DB 221 corresponds to each of the one or more photovoltaic power generation systems 100a to 100c, so that the status of the connected one or more photovoltaic power generation units 100a to 100c is separately updated and stored. It includes one or more individual DBs (2211, 2212....) that are storage spaces.

In addition, the DB 220 includes a system statistics DB (222). The system statistics DB 222 is for storing various statistical information and data generated by the statistics generating unit 212 according to the operation of the one or more photovoltaic power generation systems 100a to 100c. Power generation DB (2221) that separates and stores the power generation amount of one or more photovoltaic power generation systems (100a to 100c), carbon dioxide reduction total amount DB (2222) that updates and stores the total amount of carbon dioxide reduction, and power generation that separates and stores power generation costs Cost DB (2223), power generation time DB (2224) that separates and stores power generation time, and inverter operation rate DB (2225) that updates and stores operating rates of inverters included in the one or more photovoltaic power generation systems (100a to 100c) And a conversion efficiency DB 2226 for updating and storing the average value of each power conversion efficiency.

To describe in more detail each of the DBs 2221 to 2226 included in the system statistics DB 222, the power generation DB 2221 is each of the one or more photovoltaic power generation systems 100a to 100c. The generation amount, this month's generation amount, cumulative generation amount, the previous day's generation amount, the previous month's generation amount, and the previous year's generation amount are updated and saved.

In addition, the total amount of carbon dioxide reduction DB 2222 updates and stores the numerical value obtained according to the following equation (1).

Explaining Equation 1 above, the total amount of carbon dioxide reduction should be converted to the sum of the current generation amount to the hourly power consumption. Since the annual generation amount of 1 kW is 1.314 MWh, the sum of the current generation amount is multiplied by 1.314 to obtain the power consumption. Is multiplied by the power emission factor of 0.424. (Power emission factor 1kwh=0.424kgCO₂)

In addition, the generation cost DB 2223 is updated and stored, including the current generation amount, the current month generation amount, the current year generation amount, and the cumulative generation amount.

In addition, the generation time DB 2224 is updated and stored, including the current generation time, the current month generation time, the current year generation time, and the cumulative generation time.

In addition, the inverter operation rate DB 2235 is updated and stored in the form of a percentage (%) by multiplying a value obtained by dividing the number of power generation units in which the inverter is normally operated among the connected power generation systems by the total number of connected power generation units by 100.

In addition, the conversion efficiency DB 2226 stores a value obtained by dividing the sum of the final power conversion efficiency of the connected photovoltaic power generation system measured today by the total number of connected power generation systems.

And the server DB 220 includes an external environment DB (223). The external environment DB 223 is a DB for storing data stored by the data collection unit 213 by referring to an external telecommunication network.

Preferably, the external environment DB (223) is an insolation DB (2231) that separates and stores the amount of insolation provided from the outside by time interval, a precipitation DB (2232) that separates and stores precipitation by time interval, and the temperature is time interval. Temperature DB (2233) that separates and stores the amount of clouds by time interval, total cloud volume DB (2234) that separates and stores the amount of clouds by time interval, humidity DB (2235) that separates and stores humidity by time interval, and It includes a fog DB 2236 that separates and stores the degree of density by time interval.

Here, each of the weather information DBs 2231 to 2236 included in the external environment DB 223 is stored not only in the present information, but also in the past information according to set time intervals.

In addition, in addition to each of the weather information DBs 2231 to 2236 included in the external environment DB 223, additional DBs and information necessary for predicting power generation may be further included. This may fluctuate depending on which element the manager selects. Hereinafter, for convenience of explanation, each element stored in the weather information DB 2231-2236 by the manager, that is, insolation (S) and precipitation (R). , Temperature (T), total cloud volume (C), humidity (Wt), and fog (F) information will be selected and used as an example.

In addition, the server DB 220 includes a learning DB (224). The learning DB 224 is a space in which data generated according to the operation of the power generation learning unit 214 can be updated and stored, and training and testing for the formed model DB 2241 and the formed model are stored. It includes DB (2242 ~ 2244) that implements. A detailed operation and configuration of the learning DB 224 will be described later.

3 is a flowchart illustrating a procedure for diagnosing a failure of the components of the solar power generation unit 100 by the control server 200 in the solar power generation and control system of the present invention. Hereinafter, a method of diagnosing a failure of each solar power generation unit 100 by the control server 200 will be described with reference to FIG. 3.

Prior to the description, the failure diagnosis procedure of the photovoltaic power generation unit 100 disclosed in FIG. 3 is performed by the failure determination unit 211, and the administrator requests the implementation of the failure diagnosis procedure through the terminal 300. It may be performed at the time, or may be performed periodically according to a predetermined time interval.

As shown in FIG. 3, first, the control server 200 measures the power generation voltage V of the solar cell and performs a power generation voltage measurement step S100 of determining accordingly.

In the step (S100), the generated voltage (V) of the solar cell is measured and determined accordingly. First, it is determined whether the generated voltage (V) is 0V, and if the generated voltage (V) is not 0V, the storage battery is stored. It is determined whether the generated voltage V is greater than the battery voltage Bat_V by comparing the current storage battery voltage Bat_V of the individual storage batteries 1211 and 1212 in the unit 121.

At this time, the storage battery voltage Bat_V may be easily measured by the storage battery voltage measurement unit 1221.

In the step (S100), if the power generation voltage (V) is not 0V, but less than the current storage battery voltage (Bat_V) value, the solar power generation unit 100 is in normal operation and therefore a normal operation confirmation step ( S101) is executed to complete the fault diagnosis.

In this case, the normal operation confirmation step (S101) may further include notifying the terminal 300 of the administrator that the solar power generation unit 100 is operating normally.

And in the step (S100), if the generated voltage (V) is 0V, or even if the generated voltage (V) is not 0V exceeds the current storage battery voltage (Bat_V) value, individual storage batteries in the storage battery storage unit 121 ( A storage battery current measurement step (S110) is performed to measure the amount of current (Bat_C) charged in 1211 and 1212 and determine accordingly.

At this time, the storage battery current amount Bat_C may be easily measured by the storage battery charging current measuring unit 1222.

In the step S110, the current amount of the storage battery (Bat_C) is measured, and it is determined by checking whether the current amount of the storage battery (Bat_C) exceeds 0A. The determination at this time is determined in consideration of the power generation voltage V in the previous step (S100) and the current amount of the storage battery (Bat_C) in this step (S110).

When it is determined that the power generation voltage (V) in the previous step (S100) is not 0V but is higher than the storage battery voltage (Bat_V), the battery current amount (Bat_C) is measured in the step (S110) and does not exceed 0A. If the state, that is, less than 0A, the solar cell current measuring unit problem checking step of determining that a problem has occurred in the current measuring unit 1132 of the solar cell measuring unit 113 and confirming the problem and notifying the terminal 300 of the administrator (S111 ) To diagnose the fault.

And in the state that the generation voltage (V) in the previous step (S100) is not 0V but is determined to be higher than the storage battery voltage (Bat_V), when the battery current amount (Bat_C) is measured in the step (S110), it exceeds 0A. Or when the generated voltage (V) in the previous step (S100) is measured as 0V, the storage battery current amount (Bat_C) is measured in the step (S110) and then proceeds to the next step regardless of the result.

And after the step (S110), the light amount measuring step (S120) of measuring the current amount of light (R) using the light amount sensor 112 is performed.

In the step S120, the amount of light R is measured, and it is determined whether the measured light amount R value exceeds 20W/m2.

After the step (S120), using the measured amount of light (R), and the generated voltage (V) and storage battery current amount (Bat_C) values determined in the previous steps (S100, S110), respectively, to determine whether there is a failure and to determine the location of the failure. .

If the power generation voltage (V) value measured in the power generation voltage measurement step (S100) was 0V, and the storage battery current amount (Bat_C) measured in the storage battery current amount measurement step (S110) exceeds 0A, the light amount measurement step If the light quantity (R) value measured in (S120) exceeds 20W/m², it is determined that the solar cell current measuring unit 1132 of the solar cell measuring unit 113 has failed and notified to the administrator's terminal 300 The solar cell current measuring unit problem check step (S121) is carried out to end.

In the state in which the generated voltage (V) value measured in the generation voltage measurement step (S100) was 0V, and the storage battery current amount (Bat_C) measured in the storage battery current amount measurement step (S110) exceeds 0A, the light quantity measurement step If the light intensity (R) value measured in (S120) is less than 20W/m2, it is determined that the voltage measurement unit 1131 and the current measurement unit 1132 of the solar cell measurement unit 113 have failed, and this 300) is notified to the solar cell voltage and current measuring unit problem confirmation step (S122) is carried out to end.

And in the state that the generated voltage (V) value measured in the generation voltage measurement step (S100) is 0V, and the storage battery current amount (Bat_C) measured in the storage battery current amount measurement step (S110) is 0A or less, the light quantity measurement step (S120) If the light quantity (R) value measured in) exceeds 20W/m2, the solar cell voltage and current measuring unit problem checking step (S122) is performed to terminate.

In addition, in a state in which the generated voltage (V) value measured in the generation voltage measurement step (S100) was 0V, and the storage battery current amount (Bat_C) measured in the storage battery current amount measurement step (S110) is 0A or less, the light quantity measurement step (S120) If the light amount (R) value measured in) is less than 20W/m2, the solar power generation unit 100 to be measured performs a normal operation check step (S123), which determines that it is in normal operation, and ends.

In this case, the step of confirming the normal operation (S123) may further include notifying the terminal 300 of the administrator that the photovoltaic power generation unit 100 is operating normally.

And although the generated voltage (V) value measured in the generation voltage measurement step (S100) was not 0V, the storage battery voltage (Bat_V) value was exceeded, and the storage battery current amount (Bat_C) measured in the storage battery current amount measurement step (S110) was In the state exceeding 0A, if the light amount R measured in the light amount measurement step S120 is less than or equal to 20 W/m 2, the normal operation check step S123 is performed to end.

In addition, the generation voltage (V) value measured in the generation voltage measurement step (S100) was not 0V, but the storage battery voltage (Bat_V) value was exceeded, and the storage battery current amount (Bat_C) measured in the storage battery current amount measurement step (S110) is In the state exceeding 0A, if the light amount (R) value measured in the light amount measurement step (S120) exceeds 20W/m², it is determined that there is a problem with the light amount sensor 112 and the administrator's terminal 300 It is terminated by performing the step (S124) to check the problem of the light amount sensor to be notified.

4 is a structural diagram of the operation sequence of the data collection unit 213 and the power generation learning unit 214. Hereinafter, the operation procedure of the power generation learning unit 214 will be described with reference to FIG. 4.

As described above, the data collection unit 213 updates the external environment DB 223 by collecting data from an external telecommunication network according to a predetermined time interval or in real time, and the external environment DB 223 Until the update and storage is the data procurement step (S210).

The source of data procurement in the step S210 may be designated by an administrator, or data may be extracted and updated by an automated search program such as a web crower. However, it is preferable to use an Apache Hadoop (hereinafter'Hadoop') software framework that can collect, store, and process large amounts of data in order to smoothly proceed with the steps below the step S210. Hadoop is a freeware Java software framework that supports distributed application programs capable of processing a large amount of data, and is used as a de facto standard among platforms for processing and analyzing big data.

The Hadoop framework as described above is a distributed programming system that uses a plurality of computers as if it were one, and when implementing the control server 200 by organizing two or more computers as described above, the Hadoop framework is used. Can be used.

In addition, if the Hadoop framework is used, all of the Hadoop distributed file system (HDFS), MapReduce, Hive, Spark, etc. that can be used as components of the Hadoop framework can be used. The description of the Hadoop framework and the components of the Hadoop framework, etc., has been disclosed in the prior art, and a further description will be omitted.

The data collection source in the step (S210) may be any place accessible to the public among external telecommunication networks, for example, it may refer to annual or monthly weather data of the area where the system of the present invention is installed. There will be, and it will be possible to refer to a database that stores data collected around the world such as PVOutput, NSRDB, Meso West, and other countries' data from other solar power systems.

With respect to the data collected through the step (S210), the power generation learning unit 240 performs a data mining step (S220) of performing data mining.

The target in the step (S220) is to process the raw data stored in the external environment DB 223 so that it can be used in the next step, and based on this, raw model data, which is primitive and uncertain predictive model data for predicting power generation ( PD) is to be created. The original model data PD may be updated and stored in the formed model DB 2241.

In order to achieve the data mining technique for achieving the above step (S220) and the model training (S230) and model testing (S24) steps that will be further described below, two or more hidden layers among the artificial neural network (ANN) structures are used. The branch uses a deep neural network to achieve the purpose of the power generation learning unit 214.

In addition, the deep neural network operates using a BP algorithm (Back Propagation), so it is preferable to implement each of the steps (S230, S240) below the step (S220) using the deep neural network using the BP algorithm. .

In the following, as shown in Figure 5, including each step (S220, S230, S240) of the present invention, the operation of the power generation learning unit 214 is the deep neural network and the BP algorithm having two depths (Depth) It will be described using as an example. Since the BP algorithm applied to the deep neural network below uses the most general and formal method, those skilled in the art will be able to easily understand the description of the deep neural network and the BP algorithm to be described below.

In the step (S220), a training set (D1), a test set (D2), and a blind set (D3) to be used in the power generation prediction model are formed through an additional data purification step (S221). The training set (D1), the test set (D2), and the blind set (D3) are respectively in the training set DB 2242, the test set DB 2243, and the blind set DB 2244 in the learning DB 224. It is stored individually, and the description of the set data D1 to D3 will be described later.

The primitive model data PD formed through the step S220 is stored in the external environment DB 223 so as to correspond to the input layer portion of a general forward deep neural network. , Temperature (T), total cloud volume (C), humidity (Wt), fog (F) information is transformed into input data (X1 to X6) normalized to a value between 0.0 and 1.0, and the formation model DB Save the update to (2241).

A model training step (S230) is performed on the raw model data (PD) formed through the step (S220).

In the step S230, as shown in FIG. 5, a structure including two depths, that is, a hidden layer HL and an output layer OL, is made to perform prediction. The step S230 is performed by selecting an epoch (the number of repetitive learning), a maximum error rate, and a learning rate, where the epoch value, the maximum error rate, and the learning rate are values arbitrarily selected by the administrator to perform the step 230. For example, 1000 epochs, maximum error rates of 0.0001, 0.001, and 0.01, and learning rates of 0.0001, 0.001, 0.01, and 0.1 can be selected to proceed.

As a result of proceeding with the maximum error rate and the learning rate arbitrarily determined as described above, a root mean square error (RMSE) is calculated, and the smallest value can be derived and used.

The training set D1 is used for learning in step S230. The training set (D1) is formed in the data purification step (S221), and in the step (S221), data according to a predetermined standard, that is, the amount of insolation (S) and precipitation (R) stored in the external environment DB (223) ), temperature (T), total cloud volume (C), humidity (Wt), and fog (F) information are classified, analyzed, and set according to pre-input criteria.

A specific procedure for performing the step S230 is shown in FIG. 6. Explaining this, a numerical selection and input step (S231) in which an administrator selects a numerical value for the raw model data PD imported in the step (S230) and inputs the selected numerical value and learning pattern pair (S231), a hidden layer (HL) And the output calculation step (S232) of calculating the input weighted sum and final output of the output layer (OL), the error calculation step (S233) of calculating the error signal value, and the amount of change in connection strength to obtain the connection strength to be used in the next learning step. A calculation step (S234) is performed.

In the numerical selection and input step (S231), first, the power generation learning unit 214 includes a connection strength (V _k ) between the primitive model data (PD) and the hidden layer (HL) of the neural network, and the hidden layer (HL). The connection strength (W _k ) between the output layers (OL) is initialized to a small arbitrary value, the normalized input data (X1 to X6) included in the original model data (PD) are loaded, and a target value (d) is selected. .

At this time, if the step (S231) performs learning from the previous, and performs the step (S234) of calculating the amount of change in the connection strength to be described later, the neural network connection strengths V _k+1 and W _k+1 to be used in the next round are received. If the numerical selection and input step (S231) is recursively performed, the values of the neural network connection strengths V _k+1 and W _k+1 determined in the previous round of the connection strength change calculation step (S234) are used.

Further, in the step S231, the administrator determines and inputs an appropriate learning rate (η) and a maximum error rate (E _max ), and the power generation learning unit 214 sequentially inputs a pair of learning patterns to change the connection strength.

After the step (S231), an output calculation step (S232) is performed. In the output calculation step, first, the input weighted sum (NET _z ) of the hidden layer HL is obtained through Equation 2 below.

In Equation 2, X _n is included in the raw model data PD, insolation (S), precipitation (R), temperature (T), total cloud volume (C), humidity (Wt), and fog (F ) It is any one of experimental data (X1 to X6) normalized to a value between 0.0 to 1.0, modified with respect to the information, and T means that it has been transposed.

After obtaining the input weighted sum NET _z of the hidden layer HL through Equation 2, the output Z is obtained through Equation 3 below, expressed in the form of a sigmoid function.

And the input weighted sum (NET _y ) of the output layer OL is obtained through Equation 4 below.

And the final output (y) is obtained through Equation 5 below.

The output calculation step (S232) is terminated by obtaining the final output (y) as described above.

After the end of the step (S232), an error calculation step (S233) is performed. In the error calculation step S233, the squared error E is first updated and stored through Equation 6 below. The initial squared error E is zero.

And the error signal δ _y of the output layer OL is obtained through Equation 7 below.

In addition, the error signal δ _z of the hidden layer HL is obtained through Equation 8 below.

The error calculation step 233 is terminated by obtaining the hidden layer error signal δ _z through the above procedure.

After the end of the step (S233), the connection strength change calculation step (S234) is performed. In the step (S234), the connection strength (W _k+1 ) to be used in the numerical selection and input step (S231) of the next round of learning is calculated by obtaining the change in the connection strength (ΔW) between the hidden layer and the output layer (HL, OL). Find it through 9.

In addition, in the step (S234), the change in connection strength (ΔV) between the raw model data PD serving as an input layer and the hidden layer HL is calculated to be used in the numerical selection and input step (S231) of the next round of learning. The connection strength (V _k+1 ) is obtained through Equation 10 below.

The connection strength variation calculation step 234 is terminated by obtaining the next learning connection strength (V _k+1 , W _k+1 ) through the above procedure.

The first learning is performed through the above steps (S231 to S234), and the optimum value should be found through repetitive learning through repeated execution of the steps (S231 to S234). ) If less then checks whether more than a pre-the epoch, the end compared to the calculated back to the step (S231), and if the If pre exceeds added epoch value that case, pre-maximum error rate (E _max) if the pre- If it is more than the input maximum error rate (E _max ), it returns to the step (S231) again, and if not, it goes to the next step.

The amount of learning, the number of hidden layers, and the maximum error rate do not have clear rules in selecting, and an incorrect initial value shows an overfitting problem, so the administrator is an appropriate value in selecting the amount of learning, the number of hidden layers, and the maximum error rate. By selecting, the optimal initial value can be found through repeated learning experiments.

The model testing step (S240) is performed on the data that has passed the step (S230).

In the model testing step (S240), a test set (D2) and a blind set (D3) generated according to the criteria set in the data purification step (S221) are used.

The test set (D2) is data used in a general deep learning test step and used for verification after completion of learning, and the blind set (D3) is a validation set that is additionally classified and used, and the learning rate required for learning and This data is used to reduce overfitting.

The data passing through the step S240 become predictive model data (CD) showing a high hit rate, so that the future generation amount can be effectively predicted.

Claims (16)

1. An aspect comprising one or more photovoltaic power generation units, a control server that is communicatively connected to the at least one photovoltaic power generation unit by wire or wireless, and at least one manager terminal that is communicatively connected to the control server by wire or wirelessly As a photovoltaic power generation and control system,

   The solar power generation unit solar cell power generation unit including one or more solar cells; A light amount sensor capable of measuring the amount of light; And a solar cell unit including a solar cell measuring unit capable of measuring the power generation voltage and current of the solar cells in the solar cell generating unit. A storage battery storage unit including one or more individual storage batteries to store electric energy produced by power generation by the solar cell unit; And a storage battery unit including a storage battery measuring unit capable of measuring a storage battery voltage of individual storage batteries in the storage battery storage unit and an amount of current charged in the storage battery. And a communication unit including a communication means and a program capable of communicating with the control server by wire or wirelessly,

   The control server may include a failure determination unit for individually diagnosing and determining a failure of each of the one or more solar power generation units to individually control the at least one solar power generation unit; A statistics creation unit for creating statistics on one or more data and values related to the one or more solar power generation units; A data collection unit storing data by referring to an external telecommunication network; And a control unit including a power generation amount learning unit capable of predicting a future solar power generation amount using a deep neural network method. A system status DB for updating and storing the status of each of the one or more solar power generation units; A system statistics DB for storing one or more statistical information generated by the statistics generating unit; An external environment DB in which the data collection unit stores data stored by referring to an external telecommunication network; And a server DB including a learning DB for storing data generated according to the operation of the power generation learning unit 214; And a server communication unit connected to the at least one photovoltaic power generation unit and the at least one manager terminal through wired or wireless communication.
2. The method of claim 1, wherein the solar cell measurement unit comprises: a solar cell generation voltage measurement unit capable of individually measuring the generated voltage of the one or more solar cells; And a solar cell current measuring unit capable of individually measuring the current of the one or more solar cells,

   The storage battery measurement unit includes a storage battery voltage measurement unit capable of measuring the voltage of the one or more individual storage batteries; And a storage battery charging current measuring unit capable of measuring the charging current of the one or more individual storage batteries.
3. The method of claim 1, wherein the control server comprises two or more computers, and a distributed processing program is installed in at least two computers included in the control server, and the control server is implemented by the distributed processing program. , Solar power generation and control system.
4. The solar power generation and control system according to claim 3, characterized in that it is the distributed processing Hadoop software framework.
5. The photovoltaic power generation and control system according to claim 1, wherein the system statistics DB includes a power generation DB, a carbon dioxide reduction total DB, a power generation cost DB, a power generation time DB, an inverter operation rate DB, and a conversion efficiency DB.
6. The solar power generation and control system according to claim 1, wherein the external environment DB includes a solar radiation DB, a precipitation DB, a temperature DB, a total cloud DB, a humidity DB, and a fog DB.
7. The photovoltaic power generation and control system of claim 1, wherein the learning DB includes a formation model DB, a training set DB, a test set DB, and a blind set DB.
8. As a method of operating the solar power generation and control system of claim 1,

   A power generation voltage measurement step (S100) of measuring and determining the power generation voltage (V) of any one of the solar power generation units;

   After the step (S100), a storage battery current amount measuring step (S110) of measuring and determining the amount of battery current (Bat_C) of the solar power generation unit to be measured in the step (S100)

   And after the step (S100), characterized in that the fault is diagnosed by performing the light amount measurement step (S120) of measuring and determining the amount of light (R) of the solar power generation unit to be measured in the step (S100). How to operate the photovoltaic and control system.
9. The method of claim 8,

   In the step (S100), if the power generation voltage (V) is determined to be less than the current storage battery voltage value (Bat_V) rather than 0V, a normal operation check step (S101) is performed to confirm that the solar power generation unit to be measured is in normal operation, and the step (S110, S120), characterized in that the failure diagnosis is terminated without performing, solar power generation and control system operating method.
10. The measurement according to claim 8, wherein in the step (S100), the power generation voltage (V) is not 0V but exceeds the current storage battery voltage value (Bat_V), and the storage battery current amount (Bat_C) is 0A or less in the step (S110). Among the target photovoltaic power generation units, the solar cell current measurement unit, which determines that the current measurement unit of the solar cell has failed, and notifies the administrator's terminal, performs a problem checking step (S111) of the solar cell current measurement unit and diagnoses the failure without performing the step (S120). The method of operating a solar power generation and control system, characterized in that the termination.
11. The method of claim 8, wherein the power generation voltage (V) is 0V in the step (S100), the storage battery current amount (Bat_C) is greater than 0A in the step (S110), and the light amount (R) in the step (S120) If it exceeds 20W/㎡, it is determined that the current measuring unit of the solar cell among the solar power generation units to be measured has failed, and the problem confirmation step (S121) of the solar cell current measurement unit informing the administrator's terminal is performed, and the failure diagnosis is terminated. Characterized in, solar power generation and control system operating method.
12. The method of claim 8, wherein the power generation voltage (V) is 0V in the step (S100), the storage battery current amount (Bat_C) is greater than 0A in the step (S110), and the light amount (R) in the step (S120) ) Is less than 20W/m2, or the power generation voltage (V) is 0V in the step (S100), the storage battery current amount (Bat_C) is 0A or less in the step (S110), and the light amount in the step (S120) If (R) is more than 20W/m², the voltage and current measurement unit of the solar cell among the solar power generation units to be measured is determined to have failed, and the solar cell voltage and current measurement unit problem checking step (S122) of notifying the administrator's terminal. A method of operating a solar power generation and control system, characterized in that to perform and terminate the fault diagnosis.
13. The method of claim 8, wherein the power generation voltage (V) is 0V in the step (S100), the storage battery current amount (Bat_C) is 0A or less in the step (S110), and the light amount (R) in the step (S120). ) Is 20W/m² or less, or the power generation voltage (V) is not 0V in the step (S100) and is less than the current storage battery voltage value (Bat_V), and in the step (S110), the storage battery current amount (Bat_C) is greater than 0A. , If the amount of light (R) in the step (S120) is 20W/m² or more, performing a normal operation confirmation step (S123) for confirming that the solar power generation unit to be measured is in normal operation, and terminating the failure diagnosis. , Solar power generation and control system operation method.
14. The method of claim 8, wherein the power generation voltage (V) is not 0V in the step (S100) but is less than the current storage battery voltage value (Bat_V), and in the step (S110), the storage battery current amount (Bat_C) is greater than 0A, and the If the amount of light (R) in step (S120) is less than 20W/m², it is determined that the light amount sensor of the solar power generation unit to be measured has failed, and the light intensity sensor problem checking step (S124) of notifying the administrator's terminal is performed and the malfunction The method of operating a solar power generation and control system, characterized in that the diagnosis is terminated.
15. A method of operating a photovoltaic power generation and control system of claim 1, comprising: a data procurement step (S210) in which the data collection unit collects data from an external telecommunication network and stores it in the external environment DB; A data mining step (S220) of producing raw model data PD by performing data mining on the updated data in the external environment DB; A model training step (S230) performed at least once or more on the raw model data PD generated by performing the step (S220); Then, after executing the step (S230) at least once, testing is performed to generate predictive model data (CD),

    In the data mining step (S220), a data refining step (S221) of forming a training set, a test set, and a blind set, which are processed data, is performed.
16. The method of claim 15, wherein the model training step (S230) comprises a numerical selection and input step (S231) in which an administrator selects a numerical value for the incoming raw model data PD and inputs the selected numerical value and learning pattern pair (S231), a hidden layer. The output calculation step (S232) of calculating the input weighted sum and final output of the (HL) and the output layer (OL), the error calculation step (S233) of calculating the error signal value, and the connection strength to be used in the next learning step Including the calculation step (S234) of the amount of change in connection strength, and if the number of times the model training step (S230) is performed after the step (S234) is less than the number of epochs or more than the maximum error rate previously input, the step (S231) is repeated. A method of operating a solar power generation and control system, characterized in that to carry out.

Images