Electric power system stabilization system, setting device of electric power system stabilization system, and method

Abstract

To improve a calculation accuracy of a system stabilization system, and suppress an excess and a deficiency of a power control amount in accordance with the improvement or a control cost.SOLUTION: An electric power system stabilization system comprises: a teacher data creation part that creates teacher data by combining data related to an accident of an electric power system with meteorological data corresponding to them; a learning model creation part that creates a learning model obtained by learning the teacher data created by the teacher data creation part; a parameter switching part that switches a parameter for estimating a modulation of a residual demand after the generation of the accident on the basis of a result that the learning model to which the meteorological data corresponded to the accident is input is output when detecting the accident of the electric power system; and a control information creation part that calculates a power control amount against at least one power supply constituting the electric power system or a load by using the parameter to be switched.SELECTED DRAWING: Figure 7

Description

[Technical field]

[0001] The present invention relates to a power system stabilization system, and a setting device and method for a power system stabilization system. [Background technology]

[0002] The amount of power generated by renewable energy sources such as solar and wind power fluctuates depending on weather conditions. With the increase in the amount of renewable energy sources introduced in recent years, there is a demand for grid stabilization systems that can deal with fluctuations due to weather conditions.

[0003] Patent Documents 1 and 2, for example, are examples of calculating the amount of power control in a power system by taking into account real-time weather conditions at the time of an accident.

[0004] Patent Document 1 discloses that the system "includes a construction unit that constructs a relationship between tide fluctuation relationship information relating to tide fluctuations in a power system and tide fluctuation information indicating the level of the tide fluctuation, a prediction unit that predicts tide fluctuations of a predetermined level or more for the tide fluctuation relationship information in a prediction time period based on the tide fluctuation relationship information and the relationship between the tide fluctuation information, and a notification unit that notifies information regarding the tide fluctuation predicted by the prediction unit."

[0005] Patent Document 2 also describes a forecasting system for an area, comprising a component forecasting processor, the component forecasting processor receiving data including weather and component data for the area from a data storage device, generating a data set for the area and a set of weather variables converted into a table format to identify weather events each including a parameter along with a component location, generating a set of weather drivers each including a given set of weather variables, using the data set to configure a machine learning (ML) model for each set of weather drivers for each component in the area, iteratively identifying, for each weather event in the area, a corresponding set of weather drivers along with weather variables including maximum instantaneous wind speed and lightning current for each component, and calculating the identified configuration for the set of weather variables. "the system is configured to: generate output values ​​for the ML model corresponding to a component; update the ML model with the output values ​​and the component data for the weather event; receive real-time observation data over a time period via a communication network of impending weather (IW) events in the region; and for each time period, iteratively identify for each component the same set of corresponding weather drivers with weather variables; update the ML model corresponding to the identified component for the set of weather variables with a last performed iteration forecast output value for the component and the observation data of the IW event; and generate output values ​​for the updated ML model predicting a component status as failed or not failed for the time period." [Prior art documents] [Patent documents]

[0006] [Patent Document 1] JP 2020-31481 A [Patent Document 2] JP 2022-115061 A Summary of the Invention [Problem to be solved by the invention]

[0007] When an accident occurs in the power grid, the power grid stabilization system estimates the fluctuation in residual demand after the accident and calculates the targets and amount of power regulation. Here, residual demand P refers to the demand obtained by subtracting the power generation Pres from renewable energy (photovoltaic and wind power) from the power demand Pl of the entire power grid. Renewable energy dropout characteristics are a parameter used to estimate the fluctuation in residual demand. Photovoltaic power generation, which is an example of a renewable energy source, varies in output depending on temperature, sunlight, etc., so the amount of renewable energy dropout also varies depending on temperature, sunlight, etc. Therefore, the estimated value of the fluctuation in residual demand after an accident occurs is affected by weather conditions.

[0008] Conventionally, the set values ​​for load shedding characteristics and renewable energy shedding characteristics have been set to certain statistical values ​​calculated from actual data. However, as the amount of renewable energy introduced in the future increases, the impact of weather conditions on the estimation of load shedding characteristics, renewable energy shedding characteristics, and fluctuations in residual demand will become greater. As a result, there are concerns that the impact on excess or shortage of power control capacity and increased control costs will also become greater.

[0009] In this regard, the tidal current fluctuation monitoring device of Patent Document 1 is an invention that aims to monitor or predict sudden tidal current fluctuations.Also, the tidal current fluctuation monitoring device of Patent Document 2 aims to predict the presence or absence of failures in components, and is not intended to address the above-mentioned concerns.

[0010] Therefore, neither of them provide information on how to deal with the difficulty of collecting data before and after an accident, or the difficulty of distinguishing between load shedding and renewable energy shedding.In addition, the information output by the learning model is a stability evaluation based on tidal current fluctuations, and is not a parameter for calculating fluctuations in residual demand during an accident, such as load shedding characteristics or renewable energy shedding characteristics.

[0011] The present invention has been made in consideration of the above-mentioned points, and aims to improve the calculation accuracy of a power grid stabilization system and to reduce the resulting excess or deficiency in the amount of power control and control costs. [Means for solving the problem]

[0012] In view of the above, the present invention provides a power system stabilization system comprising: a teacher data creation unit that creates teacher data by combining data on accidents in the power system with corresponding meteorological data; a learning model creation unit that creates a learning model that has learned the teacher data created by the teacher data creation unit; a parameter switching unit that, when an accident in the power system is detected, switches parameters for estimating fluctuations in residual demand after the accident occurs based on the results output by the learning model to which meteorological data corresponding to the accident has been input; and a control information creation unit that uses the switched parameters to calculate the amount of power control for at least one power source or load that constitutes the power system.

[0013] The present invention also provides a setting device for a power system stabilization system including a control information creation unit that uses parameters of the power system to determine in advance an amount of electrical control required to ensure stability in the event of an assumed accident in the power system, the setting device including a teacher data creation unit that combines data on the accident in the power system with corresponding meteorological data to create teacher data, a learning model creation unit that creates a learning model that has learned the teacher data created by the teacher data creation unit, and a parameter switching unit that, upon detecting an accident in the power system, switches parameters for estimating fluctuations in residual demand after the accident occurs based on a result output by the learning model to which meteorological data corresponding to the accident has been input, and the control information creation unit calculates an amount of electrical control for at least one power source or load that constitutes the power system using the switched parameters.

[0014] The present invention also provides a power system stabilization method for determining, using a computer, in advance the amount of power control required to ensure stability in the event of an assumed accident in the power system, and controlling the power system with the amount of power control in the event of an accident in the power system, the computer combining data on the accident in the power system with corresponding meteorological data to create teacher data, creating a learning model that has learned the teacher data, and, upon detecting an accident in the power system, switching parameters for estimating fluctuations in residual demand after the accident occurs based on the results output by the learning model to which meteorological data corresponding to the accident has been input, and calculating the amount of power control for at least one power source or load that constitutes the power system using the switched parameters, thereby controlling the power system.

[0015] The present invention also provides a "method of setting up a power system stabilization system, using a computer to obtain parameters of a power system for calculating the amount of power control required to ensure stability in the event of an assumed accident in the power system, characterized in that the computer combines data on the accident in the power system with corresponding meteorological data to create teacher data, creates a learning model that learns the teacher data, and, upon detecting an accident in the power system, performs a process of switching parameters for estimating fluctuations in residual demand after the accident occurs based on the results output by the learning model to which meteorological data corresponding to the accident has been input, and provides the switched parameters for calculating the amount of power control for at least one power source or load that constitutes the power system." Effect of the Invention

[0016] According to the present invention, it is possible to improve the calculation accuracy of the grid stabilization system and suppress the excess or shortage of the power supply control amount and the control cost. This is made possible by providing a method for supplementing the actual data before and after the accident, which is easily insufficient, and a method for separating the data of the load drop and the renewable energy drop, and by quickly switching the parameters for calculating the fluctuation of the residual demand at the time of the accident, such as the load and renewable energy drop characteristics. [Brief description of the drawings]

[0017] [Figure 1] FIG. 1 is a diagram showing an example of a hardware configuration of a power grid stabilization system according to an embodiment of the present invention. [Diagram 2] FIG. 1 is a diagram showing an example of a functional configuration of a power grid stabilization system according to a first embodiment of the present invention. [Diagram 3] 4 is a flowchart showing an example of processing of the entire power grid stabilization system. [Figure 4] FIG. 4 is a diagram showing an example of a database for storing past performance information regarding instantaneous voltage drops. [Figure 5A] FIG. 2 is a diagram showing an example of a database for storing past performance information regarding a system configuration. [Figure 5B] A diagram showing an image of the system configuration database DB4. [Figure 6] FIG. 13 is a diagram showing an example of a database for storing past performance information regarding dropouts. [Figure 7] 13 is a flowchart showing an example of processing by a teacher data creation unit. [Figure 8] FIG. 2 is a diagram showing an example of a weather record database DB1. [Figure 9] FIG. 13 shows an example of parameters for calculating the fluctuation of residual demand to be estimated and switched in the parameter switching unit. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0018] Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings. However, the present invention is not limited to the embodiment. EXAMPLES

[0019] Fig. 1 is a diagram showing an example of a hardware configuration of a power grid stabilization system according to an embodiment of the present invention. In Fig. 1, a power grid stabilization system 1 configured using a computer has a configuration in which a communication unit 11, an input means 12, a display device 13, a CPU (Central Processing Unit) 14, a memory 15, and a database DB are connected via a bus 17.

[0020] The communication unit 11 includes a circuit and a communication protocol for connecting to a communication network. The communication network acquires weather information from a weather information providing system, acquires information such as the system configuration and current status from measurement sensors installed in the power system, and transmits control information to the slave stations.

[0021] The input means 12 includes devices such as a keyboard, a mouse, and a touch panel for allowing a user to input information.

[0022] The output means 13 includes a device for outputting information, such as a display device or a printer device.

[0023] The CPU 14 performs processes related to system stabilization, such as searching for data stored in the database DB or memory 15, executing computer programs, and storing the results of calculations in the memory 15.

[0024] The memory 15 is, for example, a RAM (Random Access Memory) and includes information storage means for storing computer programs, calculation results, image data, and the like.

[0025] The database DB stores various data related to system stabilization, and includes databases DB1 to DB6, which will be described later.

[0026] Fig. 2 is a diagram showing an example of a functional configuration of a power grid stabilization system according to a first embodiment of the present invention. In Fig. 2, the power grid stabilization system 1 includes a weather record database DB1 as a database (hereinafter sometimes abbreviated as DB) for storing past weather information. The power grid stabilization system 1 also includes a drop record database DB2, a voltage sag record database DB3, a power grid configuration database DB4, a power grid measurement database DB5, and an accident database DB6 as databases for storing record data related to past power grid accidents.

[0027] The power system stabilization system 1 also includes the functions of a learning model 57, a teacher data creation unit 61, a learning model creation unit 62, a parameter switching unit 63, and a control information creation unit 64. Among these, the learning model 57, the teacher data creation unit 61, and the learning model creation unit 62 are used for offline processing in the offline computation system 1a, and the parameter switching unit 63 and the control information creation unit 64 are used for online processing in the online computation system 1b.

[0028] The CPU 14 realizes these functions by executing computer programs called into the RAM.

[0029] The control information creation unit 64 calculates the fluctuation of the residual demand after an accident (also called a failure, but unified as an accident here) occurs in the power system based on the load drop characteristics, the renewable energy drop characteristics, the system measurement data, the accident data, etc., and calculates the amount of power control. Specifically, it assumes an accident in the power system, and estimates and calculates the stability of the power system when this accident actually occurs for each assumed accident. Then, it determines the number of generators to be stopped or the output suppression amount (power control amount) of the power system required to stabilize it when it becomes unstable, and the power control target equipment (power control equipment) to be stopped or output suppressed. Then, when the assumed accident actually occurs, it executes power system control using the power control amount and the power control target equipment that are preset for the assumed accident. As a result, the supply and demand balance after the accident is immediately restored, and system stability is maintained.

[0030] In addition, since the information sent from parent station 1 (power grid stabilization system) to child station 3 may not be just the "power control amount" to the generator, the "power control amount" should generally be called "control information."

[0031] In the present invention, parameters such as load drop characteristics, renewable energy drop characteristics, system measurement data, and accident data are obtained in advance in the offline calculation system 1a.

[0032] The device of the present invention functions as a power grid stabilization system as a whole, but if the control information creation unit 64 is regarded as an essential function of the power grid stabilization system, the other functions of the learning model 57, the teacher data creation unit 61, the learning model creation unit 62, and the parameter switching unit 63 can be regarded as a parameter setting device for the power grid stabilization system. This means that the effect of the present invention can be obtained by adding a parameter setting device to an existing power grid stabilization system mainly composed of the control information creation unit 64.

[0033] 2 is a system that pre-calculates parameters using past performance information, and includes a teacher data creation unit 61, a learning model creation unit 62, and a learning model 57. The past performance information is information on the power system when an accident occurred in the past, and is stored in a weather performance database DB1, a power outage performance database DB2, a voltage sag performance database DB3, and a system configuration database DB4.

[0034] In FIG. 2, the online calculation system 1b is a system that performs calculations in real time at short intervals using real-time data such as a system measurement database DB5 and real-time weather data in addition to the learning model 57 created by the offline calculation system 1a and past performance information accumulated in the system configuration database DB4 and the accident database DB6, and is provided with a parameter creation unit 63 and a control information creation unit 64.

[0035] 2, the weather information providing system 2 is a system that provides weather information such as lightning strikes, wind and rain, sunshine, and temperature, and is connected to the power grid stabilization system 1 via the communication unit 11 in Fig. 1, and appropriately supplies weather information to the power grid stabilization system 1. Of the weather information provided, past weather information D1 is accumulated in a weather record database DB1 as exemplified in Fig. 8, and real-time weather information is input to a parameter switching unit 63.

[0036] The weather record database DB1 stores information (weather record data) showing the weather conditions at the location on the map of the power system to be managed. Figure 8 shows an image of the weather record database DB1, in which the geographical area to be managed is organized into a mesh matrix such as A1-A5, B1-B5, C1-C5, E1-E5, F1-F5, and the weather conditions at each time in the area within each mesh are organized by weather type and time as the weather record database DB1.

[0037] In this embodiment, in accordance with legend D1a indicating the thundercloud conditions, the thundercloud conditions of each region at 10:10 and 10:20 on August 2, 2002 are stored as D1b and D1c, in accordance with legend D1d indicating the wind speed conditions, the wind speed conditions of each region at 10:10 and 10:20 on August 2, 2002 are stored as D1e and D1f, and in accordance with legend D1g indicating the sunshine conditions, the sunshine conditions of each region at 10:10 and 10:20 on August 2, 2002 are stored as D1h and D1i.

[0038] 2, a slave station 3 issues a control command to a power-controlled device 4a (such as a generator) based on control information output from a master station. In this embodiment, the master station refers to the power grid stabilization system 1.

[0039] 2, a power system 4 includes facilities such as power plants, transmission lines, distribution lines, substations, etc. Various pieces of information are provided from measurement sensors installed in the power system 4 to a power failure record database DB2, a voltage sag record database DB3, a system record database DB4, a system measurement database DB5, and an accident database DB6.

[0040] In Fig. 2, the drop record database DB2 stores past record information (drop record data D2) on drop acquired from a measurement sensor installed in a power system as shown in Fig. 6. According to the drop record database DB2 in Fig. 6, the drop record data D2 is saved as data on the date and time when the drop occurred D2a, the measurement point D2b, the active power drop rate ΔP (D2c), and the reactive power drop rate ΔQ (D2d). For example, the example in the first line records that a drop occurred at measurement point 1 on August 2, 2002 at 10:10 with an active power drop rate ΔP of 2% and a reactive power drop rate ΔQ (D2d) of 1%.

[0041] The voltage sag record database DB3 stores past record information (voltage sag record data D3) on instantaneous voltage dips (hereinafter also referred to as voltage sags) acquired from measurement sensors installed in a power grid as shown in Fig. 4. According to the voltage sag record database DB3 in Fig. 4, the voltage sag record data D3 is saved as data on the date and time D3a when a disconnection occurred, a measurement point D3b, and a maximum voltage dip rate D3c. For example, the example in the first row records that a maximum voltage dip rate of 20% occurred at measurement point 1 at 10:12 on August 2, 2002.

[0042] The system configuration database DB4 stores information (system configuration data D4) on the configuration of system equipment installed in the power system as exemplified in Fig. 5A. Specifically, according to the system configuration database DB4 in Fig. 5A, the system configuration data D4 stores data on the load and power source power flow ratio D4b, the connection relationship D4c with other power system equipment, the presence or absence of a measurement sensor D4d, and the short circuit capacity D4e for the system equipment D4a, together with the equipment installation position information D4f in the power system. When a measurement sensor is provided, information on the measurement point where the measurement sensor is installed may also be stored together with the presence or absence D4d of the measurement sensor.

[0043] For example, in the example in the first row, it is recorded that the transmission line 1 is connected to the bus bar 1, there is no measurement sensor, the short circuit capacity is 500 MVA, and the equipment is installed at the equipment installation position B1. The system configuration database DB4 including the equipment installation position information D4f can be said to show the layout of the power system to be managed on a map.

[0044] FIG. 5B shows an image of the system configuration database DB4, in which the area 54 of the power system 4 to be managed is organized into a mesh matrix such as A1-A5, B1-B5, C1-C5, E1-E5, and F1-F5, and the connection relationships of the power system configuration equipment present within each mesh are shown.

[0045] In the areas A1, A2, B1, and B2 shown enlarged on the left side of Fig. 5B, each power system component equipment is placed at each position as shown, and the maximum voltage drop is measured at measurement point 1 and measurement point 2, the load drop of load 1 is measured at measurement point 1, and the renewable energy drop of renewable energy 1 is measured at measurement point 2. Similarly, in the areas E4, E5, F4, and F5 shown enlarged on the right side of Fig. 5B, each power system component equipment is placed at each position as shown, and the maximum voltage drop is measured at measurement point 3, and the dropout of load 2 and renewable energy 2 is measured at measurement point 3.

[0046] The system measurement database DB5 stores information (system measurement data D5) related to real-time or past system currents (generator output, active power passing through substations, etc.) acquired from measurement sensors installed in the power system. The accident database DB6 stores information (accident data D6) related to past or expected accidents in the power system, including a triple-circuit short circuit accident on a transmission line and a single-circuit ground fault accident. Since the method of storing these data is arbitrary, explanations using illustrations are omitted.

[0047] Fig. 3 is a flowchart showing an example of processing of the entire power grid stabilization system. The offline calculation system 1a performs processing steps S61 and S62, and the online calculation system 1b performs processing steps S63 and S64. Hereinafter, the processing content at each processing step will be described according to the flow of Fig. 3.

[0048] First, in processing step S61, the teacher data creation unit 61 supplements any deficiencies in actual data (such as momentary sag actual data and tripping actual data) before and after the accident based on the classification of the tripping actual data stored in the tripping actual data database DB2 and the system configuration information in the system configuration database DB4.

[0049] FIG. 7 is a flow chart showing a specific example of the process in the process step S61 in FIG. 3, for the dropout record data D2 as an example.

[0050] In the first processing step 6101 in processing step S61 of Figure 3, the presence or absence D4d of a measurement sensor and the measurement point D4f in the system configuration database DB4 are referenced to obtain the installation locations of the measurement sensors, and the drop record data for each measurement point D2b stored in the drop record database DB2 of Figure 6 is obtained. Then, by referring to the system configuration database DB4 of Figure 5, the drop record data is classified into three types of groups: a group of load drop data, a group of renewable energy drop data, and a group of data containing a mixture of load drop and renewable energy drop.

[0051] Here, the dropout record database DB2 stores the active power dropout rate and reactive power dropout rate of each measurement point as shown in Figure 6, and the system configuration database DB4 stores the system connection relationships of loads and renewable energy and the installation locations of measurement sensors (positions of each measurement point) as shown in Figure 5A, so these can be linked to the data of measurement point D2b and processed in an appropriate association manner.

[0052] According to this, for example, in Figure 5B, the measurement sensor installed at measurement point 1 acquires the power flow state of only load 1, so it can be seen that the active power drop rate and reactive power drop rate acquired at measurement point 1 in Figure 6 are load drop data.

[0053] On the other hand, since the measurement sensor installed at measurement point 2 in Figure 5B acquires the power flow state of only renewable energy 1, it can be seen that the active power dropout rate and reactive power dropout rate acquired at measurement point 2 in Figure 6 are renewable energy dropout data.

[0054] Furthermore, in Figure 6, the measurement sensor installed at measurement point 3 acquires a power flow state in which load 2 and renewable energy 2 are mixed, so it can be seen that the active power drop rate and reactive power drop rate acquired at measurement point 2 in Figure 6 are data in which load drop and renewable energy drop are mixed (hereinafter also referred to as mixed data).

[0055] In processing step S6102, it is determined whether the actual data on load drop-out and renewable energy drop-out are included in the drop-out record data acquired in processing step S6101. If data on load drop-out and renewable energy drop-out are included, the process proceeds to processing step S6103, and if not, the process proceeds to processing step S6106.

[0056] In processing step S6103, for data in which load dropout and renewable energy dropout are mixed, it is determined whether the proportion of load dropout and renewable energy dropout in the dropout data can be estimated from the configuration of loads and power sources in the power system. For example, as shown in Fig. 5A, when the power flow ratio of the load and power source for each time and weather at measurement point 3 is known by referring to system configuration database DB4, the active power dropout rate (or reactive power dropout rate) of the load and the active power dropout rate (or reactive power dropout rate) of the renewable energy can be calculated by dividing the active power dropout rate (or reactive power dropout rate) by the power flow ratio corresponding to the time and weather at which the load dropout or renewable energy dropout occurred.

[0057] Even if the power flow ratio is unknown, if it is assumed that the load flow is much greater than the renewable energy flow because it is nighttime, the weather is cloudy, the power source capacity is very small, etc., data containing a mixture of load dropout and renewable energy dropout can be regarded as load dropout data. Alternatively, in the opposite case, that is, if it is assumed that the renewable energy flow is much greater than the load, data containing a mixture of load dropout and renewable energy dropout can be regarded as renewable energy dropout data.

[0058] In the above cases, it is determined that the load and renewable energy ratios in the missing data can be estimated from the load and power source configuration, and the process proceeds to processing step S6104. If not, the process proceeds to processing step S6105.

[0059] In process step S6104, based on the result of the determination in process step S6103, the data including both load drop and renewable energy drop is separated into load drop data and renewable energy drop data using the load and power source configuration (such as the load and power source power flow ratio).

[0060] In processing step S6105, data in which load drop and renewable energy drop are mixed and which could not be separated into load drop data and renewable energy drop data based on the determination result in processing step S6103 is discarded.

[0061] In processing step S6106, it is determined whether the current number of load drop data and renewable energy drop data (or voltage sag record data) is sufficient as teacher data. If it is determined that it is sufficient, the processing of the teacher data creation unit is terminated, and if it is determined that it is not sufficient, the processing proceeds to processing step S6106. For example, a method may be used in which if the number of load drop data and renewable energy drop data (or voltage sag record data) exceeds a preset threshold, it is determined that it is sufficient, and if not, it is determined that it is not sufficient.

[0062] In process step S6113, it is determined whether there is a similar system. Here, a similar system is a system that has similar measurement sensor installation positions, load / power supply configuration, short circuit capacity, etc. stored in the system configuration DB. If there is a similar system, proceed to process step S6107, and if not, proceed to process step S6109.

[0063] In processing step S6107, the shortage of teacher data is made up by substituting the actual dropout data (or actual voltage sag data) of another similar system as the actual dropout data (or actual voltage sag data) of the own system. Alternatively, the shortage of teacher data may be made up by substituting the actual dropout data (or actual voltage sag data) of another similar system, which is averaged for each time or weather condition, as the actual dropout data of the own system.

[0064] In processing step S6108, it is determined whether the current number of pieces of load drop data and renewable energy drop data is sufficient as training data. If it is sufficient, the processing of the training data creation unit is terminated, and if it is not sufficient, the process proceeds to processing step S6112.

[0065] In processing step S6109, it is determined whether a sufficient amount of data can be secured as training data by using data containing a mixture of load drop and renewable energy drop. If it is determined that the amount can be secured, the process proceeds to processing step S6110, and if not, the process proceeds to processing step S6112. For example, a method may be used in which if data containing a mixture of load drop and renewable energy drop is used, it is determined that the amount of data can be secured if it exceeds a pre-set amount of data required as training data, and it is determined that the amount cannot be secured if not.

[0066] In processing step S6110, a part of the data of a mixture of load dropout and renewable energy dropout rejected in processing step S6105 is applied as training data to make up for the lack of training data. For example, a method may be used in which the data of the smallest power source capacity among the mixed dropout data is used as load dropout data, or conversely, the data of the largest power source capacity is used as renewable energy dropout data.

[0067] Alternatively, a method may be adopted in which the load drop data and the renewable energy drop data are not separated for only a specific system, and the mixed data is used as it is as training data. In this case, since the number of parameters for calculating the fluctuation of the residual demand output by the parameter switching unit may change, an alarm may be issued to notify the user of this.

[0068] In processing step S6111, it is determined whether the current number of pieces of load drop data and renewable energy drop data is sufficient as training data. If it is sufficient, the processing of the training data creation unit is terminated, and if it is not sufficient, the process proceeds to processing step S6112.

[0069] In processing step S6112, provisional performance data is created by simulation. For example, when lightning damage is a cause of momentary voltage drops and load drop / renewable energy drop, provisional performance data may be created by calculating the maximum voltage drop rate and drop rate for each area as follows:

[0070] As a specific example, let us consider a case where lightning strikes point A in area A1 shown on the left side of Figure 5B, causing busbar 1 and busbar 2 to trip. At load busbar L1, the maximum voltage sag rate is VL1A, the active power drop rate is PL1A, and the reactive power drop rate is QL1A, while at load busbar L2, the maximum voltage sag rate is VL2A, the active power drop rate is PL2A, and the reactive power drop rate is QL1A. At renewable energy-connected busbar R1, the maximum voltage sag rate is VR1A, the active power drop rate is PR1A, and the reactive power drop rate is QR1A, while at renewable energy-connected busbar R2, the maximum voltage sag rate is VR2A, the active power drop rate is PR2A, and the reactive power drop rate is QR2A.

[0071] At this time, the maximum voltage drop rate of area A1 is the average value, and the dropout rate of area A1 is the total value. At this time, the following relationships (1), (2), and (3) hold. [Number 1] Maximum voltage drop rate in area A1 = (VL1A + VL2A) / 2 (1) [Number 2] Active power dropout rate of area A1 = PR1A + PR2A (2) [Number 3] Reactive power dropout rate in area A1 = QR1A + QR2A (3) In this way, training data is created by supplementing actual data before and after accidents, which tend to be lacking, and by separating data on load dropout and renewable energy dropout.

[0072] 7 described above, the processing function of the teacher data creation unit 61 has been described in detail. In this function, data on power system accidents is combined with corresponding meteorological data to create teacher data, and at this time, actual data on past power system accidents is combined with corresponding meteorological data to create teacher data.

[0073] Also, it is preferable to classify the actual data into a plurality of groups, and combine each classified actual data or data obtained by processing each classified actual data with corresponding meteorological data to create teacher data, and the groups to be classified at this time preferably include at least two of a group of data on load drop, a group of data on renewable energy drop, or a group of data containing a mixture of load drop and renewable energy drop. Also, it is preferable to classify the actual data stored in the database into a plurality of groups based on the installation position of the sensor that measured the drop actual data, and it is preferable to store drop actual data, which is the actual results of load drop and renewable energy drop, and voltage sag actual data as actual data related to past accidents in the power system, and combine the drop actual data with the corresponding voltage sag actual data and meteorological data to create teacher data.

[0074] Furthermore, the processing function of the teacher data creation unit 61 is preferably provided with a function for creating teacher data by integrating and statistically processing performance data in systems having similar information regarding the locations of measurement sensors, load / power source configuration, and transmission line short-circuit capacity when actual performance data is insufficient as teacher data, and also with a function for creating provisional performance data by simulation when actual performance data is insufficient as teacher data and there is no similar system.

[0075] Now, return to processing step S62 in Fig. 3. In processing step S62, the learning model creation unit 62 performs machine learning on the relationship between weather conditions and parameters (such as dropout characteristics) for calculating the fluctuation in residual demand, and creates a learning model 57. At this time, the actual data for calculating the fluctuation in residual demand (actual data on load dropout / renewable energy dropout and actual data on momentary voltage drop, etc.) created in processing step S61 and actual weather data before and after the occurrence of load dropout / renewable energy dropout and momentary voltage drop are used as teacher data.

[0076] In processing step S63, the parameter switching unit 63 applies real-time weather data to the learning model 57 created in processing step S62 at regular intervals (e.g., every 30 seconds). The learning model 57 outputs parameters (parameters for load drop characteristics and renewable energy drop characteristics) for calculating fluctuations in residual demand linked to the weather data at regular intervals. The parameter switching unit 63 updates the output values ​​of the learning model as the latest parameters. At this time, the real-time weather data applied to the learning model may be data summarizing the weather conditions of each area by level, for example, as shown in FIG. 8.

[0077] Furthermore, the parameters for calculating the fluctuation in residual demand that are updated by the parameter switching unit 63 may be parameters (ΔV0, ΔVm, ΔPm, ΔQm, ΔVres0, ΔVresm, ΔPresm, ΔQresm) of a broken line of dropout characteristics as shown in FIG.

[0078] Here, the horizontal axis of the shedding characteristics in Figure 9 is the maximum voltage drop rate, and the vertical axis is the shedding rate, and there are two types of shedding characteristics: load shedding characteristics (left side of the figure) and renewable energy shedding characteristics (right side of the figure), and the parameters for each are determined. Note that in the legend, ΔP indicates the active power shedding rate, and ΔQ indicates the reactive power shedding rate.

[0079] Here, ΔPm is the maximum value of the load's active power drop rate, ΔQm is the maximum value of the load's reactive power drop rate, ΔPresm is the maximum value of the renewable energy's active power drop rate, and ΔQm is the maximum value of the renewable energy's reactive power drop rate. ΔV0 is the maximum voltage drop rate when the load drop rate starts to rise steadily, ΔVm is the maximum voltage drop rate when the load drop rate saturates, ΔVres0 is the maximum voltage drop rate when the renewable energy drop rate starts to rise steadily, and ΔVm is the maximum voltage drop rate when the renewable energy drop rate saturates. Specific values ​​of these parameters are summarized and exemplified at the top of Figure 9.

[0080] In processing step S63, the processing function of parameter switching unit 63 is executed. In short, when an accident in the power system is detected based on information from the accident database DB6, this parameter switching unit 63 switches parameters for estimating the fluctuation in residual demand after the occurrence of the accident based on the results output by the learning model to which the weather data corresponding to the accident has been input. Specifically, it is preferable to switch at least one of the parameters for load drop or renewable energy drop to the value output by the learning model as a parameter for estimating the fluctuation in residual demand after the occurrence of an accident.

[0081] In processing step S64, the control information creation unit 64 inputs the parameters for calculating the fluctuation of the residual demand determined in processing step S63, the system measurement data, the system configuration data, and the accident data to calculate the equipment to be controlled and the amount of power control. Control information including the calculation results is transmitted to the child station. There are many methods for this calculation, such as those disclosed in JP 2018-182844 A. These methods may be used for the calculation.

[0082] In this way, we provide a method for supplementing actual data before and after an accident, which tends to be insufficient, and a method for separating data on load dropout and renewable energy dropout. In addition, by quickly switching parameters for calculating fluctuations in residual demand during an accident, such as load and renewable energy dropout characteristics, we can improve the calculation accuracy of the grid stabilization system and reduce the resulting excess or deficiency in power control and control costs.

[0083] According to the present invention, it is possible to improve the calculation accuracy of the grid stabilization system and suppress the excess or shortage of the power supply control amount and the control cost. This is made possible by providing a method for supplementing the actual data before and after the accident, which is easily insufficient, and a method for separating the data of the load drop and the renewable energy drop, and by quickly switching the parameters for calculating the fluctuation of the residual demand at the time of the accident, such as the load and renewable energy drop characteristics. [Explanation of symbols]

[0084] 1: Power grid stabilization system 11: Communications Department 12: Input method 13:Display device 14: CPU 15: Memory DB: Database 17: Bus DB1: Weather record database DB2: Dropout performance database DB3: Voltage drop performance database DB4: System configuration database DB5: System measurement database DB6: Accident Database 57: Learning model 61: Teacher data creation department 62: Learning model creation unit 63: Parameter switching section 64: Control information creation unit 1a: Offline computing system 1b: Online computing system

Claims

1. 1. A power system stabilization system comprising: a teacher data creation unit that creates teacher data by combining data on accidents in a power system with corresponding meteorological data; a learning model creation unit that creates a learning model that has learned the teacher data created by the teacher data creation unit; a parameter switching unit that, when an accident in the power system is detected, switches parameters for estimating fluctuations in residual demand after the accident occurs based on a result output by the learning model to which meteorological data corresponding to the accident has been input; and a control information creation unit that calculates a power control amount for at least one power source or load that constitutes the power system using the switched parameters.

2. 2. The power system stabilization system according to claim 1, A power system stabilization system comprising a database for storing historical data on past power system accidents, wherein the teacher data creation unit creates teacher data by combining the historical data on past power system accidents stored in the database with corresponding meteorological data.

3. 3. The power system stabilization system according to claim 2, the teacher data creation unit classifies the performance data stored in the database into a plurality of groups, and creates teacher data by combining each of the classified performance data or data obtained by processing each of the classified performance data with corresponding meteorological data.

4. 4. The power system stabilization system according to claim 3, The database stores actual data on past power system accidents, which are load drop-out and renewable energy drop-out, and the groups classified by the teacher data creation unit include at least two of a group of data on load drop-out, a group of data on renewable energy drop-out, or a group of data containing a mixture of load drop-out and renewable energy drop-out.

5. 5. A power system stabilization system according to claim 4, The power system stabilization system, wherein the teacher data creation unit classifies the historical data stored in the database into a plurality of groups based on the installation positions of the sensors that measured the dropout historical data.

6. 4. The power system stabilization system according to claim 3, The database stores actual load drop data and actual voltage sag data as actual data relating to past power system accidents, and the teacher data creation unit creates teacher data by combining the actual load drop data with the corresponding actual voltage sag data and meteorological data.

7. 7. A power system stabilization system according to claim 6, The power system stabilization system is characterized in that the parameter switching unit switches at least one of parameters of load drop or renewable energy drop as a parameter for estimating a fluctuation in residual demand after an accident occurs to a value output by the learning model.

8. 2. The power system stabilization system according to claim 1, The power system stabilization system, wherein the meteorological data is data on lightning, wind and rain, sunlight, and temperature obtained from a meteorological information providing system.

9. 3. The power system stabilization system according to claim 2, The power system stabilization system is characterized in that the teacher data creation unit has a function of creating teacher data by integrating and statistically processing the past data in systems having similar information regarding the installation locations of measurement sensors, load / power source configurations, and transmission line short-circuit capacities when the past data is insufficient as teacher data.

10. 10. The power system stabilization system according to claim 9, The power system stabilization system, wherein the teacher data creation unit has a function of creating provisional past performance data by simulation when the past performance data is insufficient as teacher data and there is no similar system.

11. A setting device for a power system stabilization system including a control information creation unit that determines in advance an amount of power control required to ensure stability in the event of an assumed fault in the power system using parameters of the power system, The setting device includes a teacher data creation unit that creates teacher data by combining data related to accidents in the power system with corresponding meteorological data, a learning model creation unit that creates a learning model that learns the teacher data created by the teacher data creation unit, and a parameter switching unit that, when an accident in the power system is detected, switches parameters for estimating a fluctuation in residual demand after the accident occurs based on a result output by the learning model to which meteorological data corresponding to the accident is input; The setting device for a power system stabilization system, wherein the control information creation unit calculates a power control amount for at least one power source or load that constitutes the power system using the switched parameters.

12. A power system stabilization method for determining, in advance, a power control amount required to ensure stability in an assumed power system accident using a computer, and controlling the power system by the power control amount in the event of the power system accident, comprising: The computer combines data on accidents in the power system with corresponding meteorological data to create teacher data, creates a learning model that learns the teacher data, and, upon detecting an accident in the power system, switches parameters for estimating fluctuations in residual demand after the accident occurs based on the results output by the learning model to which meteorological data corresponding to the accident has been input, and uses the switched parameters to calculate the amount of power control for at least one power source or load that constitutes the power system, thereby controlling the power system.

13. A method for setting a power system stabilization system, which uses a computer to obtain parameters of a power system for calculating a power control amount necessary to ensure stability in an assumed accident of the power system, comprising: The computer creates teacher data by combining data on power system accidents with corresponding meteorological data, creates a learning model that learns the teacher data, and when it detects an accident in the power system, performs a process of switching parameters for estimating fluctuations in residual demand after the accident occurs based on a result output by the learning model to which meteorological data corresponding to the accident has been input; A method for setting a power system stabilization system, comprising the steps of: providing the switched parameters for calculating an amount of electrical control for at least one power source or load that constitutes the power system.

Images