JP2019201505A - Power system monitoring system, power system monitoring method, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a power system monitoring system, a power system monitoring method, and a program capable of providing a user with information for confirming accuracy of stability of a power system evaluated by using a model by machine learning. thing. A power system monitoring system according to an embodiment includes an estimation unit, a stability evaluation unit, and a generation unit. The estimation unit estimates the second system state based on the first system state. The stability evaluation unit inputs the second system state estimated by the estimation unit and information on an accident expected in the second system state into a learned evaluation model, thereby estimating the assumed accident. The stability of the power system equipment predicted later is derived. The generation unit generates a comparison image based on a comparison between the learning data used for learning the learned model and the second system state. [Selection diagram] Figure 1

Description

  Embodiments described herein relate generally to a power system monitoring system, a power system monitoring method, and a program.

  In the stability evaluation of an electric power system, what uses a computer simulation is known. However, when using simulation, there is a problem that it takes time for complicated mathematical analysis processing. In this connection, a technique for evaluating the stability of a power system using a neural network is known. However, after evaluating stability using a neural network or the like as in the prior art, a user who feels uneasy about the result may re-evaluate using simulation. In this case, the stability evaluation as a whole does not shorten the time, and when there are many cases where it is desired to re-evaluate, it may be difficult to re-evaluate the simulation for all.

JP-A-2-162404

  Problems to be solved by the present invention include a power system monitoring system and a power system monitoring system that can provide a user with information for confirming the accuracy of the stability of the power system evaluated using a machine learning model. A method and program are provided.

  The power system monitoring system of the embodiment includes an estimation unit, a stability evaluation unit, and a generation unit. The estimating unit is a second system state that is a system state of the power system equipment at a second time after a predetermined time from the first time, based on a first system state that is a system state of the power system equipment at the first time. Is estimated. The stability evaluation unit inputs the information on the second system state estimated by the estimation unit and information on the accident assumed in the second system state into a learned evaluation model, thereby causing the assumed accident The stability of the power system equipment predicted later is derived. The generation unit generates a comparison image based on a comparison between learning data used for learning the learned model and the second system state.

The figure which shows an example of a structure of the electric power grid | system monitoring system 10 of embodiment. The figure which shows an example of the content of the assumption accident case information 13A of embodiment. The figure which shows the example of the probability distribution of the renewable energy output prediction of embodiment. The figure which shows an example of the renewable energy output prediction information 13B of embodiment. The figure which shows an example of the content of node data. The figure which shows an example of the content of branch data. The figure which shows an example of a structure of the learning process part. The figure which shows an example of several learning data Dt and several stability St. FIG. FIG. 7 is a reference diagram for explaining an example of processing by the learning unit 43 and an evaluation model. The figure which shows an example of the estimated data De and several stability Se. The figure which shows an example of the 1st comparison image 510. FIG. The figure which shows an example of the 2nd comparison image 520. FIG. The figure which shows the other example of the 2nd comparison image 520. FIG. 4 is a flowchart showing an example of a process flow in the power system monitoring system 10. The flowchart which shows an example of the flow of the process in which the production | generation part 19 produces | generates a comparison image.

  Hereinafter, a power system monitoring system, a power system monitoring method, and a program according to embodiments will be described with reference to the drawings.

  FIG. 1 is a diagram illustrating an example of a configuration of a power system monitoring system 10 according to the embodiment. The power system monitoring system 10 is a system that estimates the stability of the future system state of the power system 21.

  The power system monitoring system 10 estimates the system state in the future (for example, 30 minutes or 60 minutes after the present time) based on the current system state of the power system 21. In addition, the power system monitoring system 10 monitors, for example, the supply and demand state of the power system 21 and estimates the control content for the power system 21 that is executed when an assumed accident occurs. The power system monitoring system 10 derives the stability of the power system 21 when an assumed accident occurs based on these estimation results. Specific evaluation items for the stability of the power system 21 include, for example, overload, voltage stability, transient stability, frequency stability, and the like. When the values of these items exceed (or fall below) a predetermined management value (threshold value), the power system monitoring system 10 causes a failure or dropout of equipment included in the power system 21 and causes a power failure. Since there is a possibility of causing this, it is determined that the stability of the power system 21 is low.

[Power system 21]
First, the power system 21 will be described. The power system 21 includes, for example, a plurality of generators 22 (22a, 22b,...), A plurality of renewable energy power sources 23 (23a, 23b, 23c, 23d,...), And a plurality of consumers 24 (24a,. ...) are connected. The generator 22 is, for example, a large-scale generator that performs thermal power generation, hydroelectric power generation, nuclear power generation, and the like. The renewable energy power source 23 is a power generation facility that generates power using, for example, sunlight, wind power, or the like that is renewable energy. The renewable energy power source 23 can be constructed at various scales. Individual renewable energy power sources 23 and consumers 24 may be grouped for each region, for example, for each region set 31 (31a, 31b, 31c). The pattern of this area set 31 may be set arbitrarily, and is not limited to the illustration, and can be set in various patterns.

  The power system 21 includes a plurality of measuring devices 25 (25-1, 25-2,..., 25-m) and a plurality of control terminal devices 26 as devices for cooperating with the power system monitoring system 10. (26-1, 26-2, ..., 26-m). m is an arbitrary natural number.

  The plurality of measuring devices 25 measure the system state at each location (branch or node) included in the power system 21 and provide information indicating the measured system state (hereinafter referred to as system state information) as a power system monitoring system. 10 to send. The system state includes, for example, voltage, phase, power flow, transformer load, generator output, etc. at each branch or node. The tidal current is an index indicating the magnitude of electricity flow in various places in a certain state where electricity is flowing, and is represented by the magnitude of active power, reactive power, or the like, for example.

  Based on the control command received from the power system monitoring system 10, the control terminal device 26 controls the system control device and the generator 22 (output amount, etc.), suppresses the output of the renewable energy power source 23, and the like. The system control device includes, for example, a phase advance capacitor, a shunt reactor, an SVC (Static Var Compensator), and the like.

[Power system monitoring system 10]
Next, the power system monitoring system 10 will be described. The power system monitoring system 10 includes, for example, an input unit 11, a display unit 12, a storage unit 13, a data management unit 14, a system information collection unit 15, a system state estimation unit 16, and a stability evaluation unit 17. The control content determination unit 18, the generation unit 19, and the learning processing unit 40 are provided. The power system monitoring system 10 includes one or more processors. The power system monitoring system 10 may be a single computer device or may be a computer device distributed in two or more. For example, only the input unit 11 and the display unit 12 are realized by a terminal device such as a PC, and the storage unit 13, the data management unit 14, the system information collection unit 15, the system state estimation unit 16, and the stability evaluation unit 17. And the control content determination part 18 and the production | generation part 19 may be implement | achieved by the server apparatus etc. which are connected via a network with a terminal device.

  The input unit 11 includes, for example, some or all of various keys, buttons, dial switches, a mouse, a touch panel formed integrally with the display unit 12, and the like. The input unit 11 may be a connection unit that is electrically connected to an external device. The display unit 12 is, for example, an LCD (Liquid Crystal Display) or an organic EL (Electroluminescence) display device.

  The storage unit 13 is, for example, a RAM (Random Access Memory), a ROM (Read Only Memory), a flash memory such as an SSD, an HDD, or the like. The storage unit 13 stores, for example, information such as assumed accident case information 13A, renewable energy output prediction information 13B, demand prediction information 13C, operation plan information 13D, and first system information 13E. The storage unit 13 may be an external storage device such as NAS (Network Attached Storage) accessible by the power system monitoring system 10 via a network.

  The assumed accident case information 13 </ b> A is information related to a pre-typed assumed accident assumed in the power system 21. As typical accidents classified in advance, for example, a power line failure, a bus failure, a generator failure (failure of the generator 22), a transformer failure, and the like are typical. FIG. 2 is a diagram illustrating an example of the content of the assumed accident case information 13A of the embodiment. As shown in FIG. 2, the assumed accident case information 13 </ b> A is information in which an accident target location and an accident aspect are associated with identification information (for example, an accident number) for identifying each accident, for example. . The accident target location is a location where an accident is assumed to occur in the power system 21. The accident target location is information indicating the location, and is represented by, for example, a track number, a node name, a branch name, or the like. The accident aspect is a type of accident that occurs in the power system 21. Information such as “3Φ3LG” in the figure is a code representing an accident aspect. The combination of these accident numbers, accident response points, and accident aspects is hereinafter referred to as accident types.

  The renewable energy output prediction information 13B is information on a predicted value that is predicted as an output of the renewable energy power source 23 in the future (for example, 30 minutes or 60 minutes after the present time). The renewable energy power source 23 has an uncertainty that the output fluctuates due to the influence of the weather or the like. The renewable energy output prediction information 13B is represented by, for example, a probability distribution in consideration of uncertainty.

  FIG. 3 is a diagram illustrating an example of the probability distribution of the renewable energy output prediction according to the embodiment. In the graph of FIG. 3, the vertical axis represents the occurrence probability, and the horizontal axis represents the renewable energy output. As shown in FIG. 3, the renewable energy output is defined as having a certain probability distribution because the output varies. Uncertainty of the renewable energy output is expressed, for example, by defining that the output range is inside the predicted value ± σ (σ: standard deviation) (68.27%).

  This range can be arbitrarily determined. For example, the range of the renewable energy output may be defined as ± 2σ or ± 3σ of the predicted value, or an index other than the standard deviation, for example, The capacity ratio of the renewable energy power source 23 may be used to define the inside of ± 10% of the predicted value. Renewable energy output prediction information 13 </ b> B including such uncertainty is stored in the storage unit 13 for each renewable energy power source 23 or region set 31.

  FIG. 4 is a diagram illustrating an example of the renewable energy output prediction information 13B according to the embodiment. As shown in FIG. 4, in this renewable energy output prediction information 13B, a predicted value of renewable energy output and a value corresponding to ± σ in the probability distribution are given for each area (region set 31). For example, the predicted value of the output related to the renewable energy power supply 23 includes a value that is larger than the predicted value by a predetermined amount (for example, predicted value + σ) and a value that is smaller than the predicted value by a predetermined amount (for example, predicted value−σ). By using such renewable energy output prediction information 13B, the power system monitoring system 10 can provide a future power system 21 for every combination pattern of system states including a renewable energy output that varies due to uncertainty. It is possible to evaluate whether or not the reliability can be maintained.

  The demand prediction information 13C is information in which the magnitude of demand in each time zone (for example, every 30 minutes) of the day is predicted, and is predicted with high accuracy from a system operator's rule of thumb. The operation plan information 13D is information on an operation plan of the power system 21 determined in advance, such as turning on / off the phase adjusting equipment, starting / stopping the generator 22, and increasing / decreasing the output of the generator 22.

  Each function unit of the data management unit 14, the system information collection unit 15, the system state estimation unit 16, the stability evaluation unit 17, the control content determination unit 18, and the generation unit 19 is, for example, a processor such as a CPU (Central Processing Unit) Is realized by executing a program (software). In addition, some or all of these functional units include hardware such as LSI (Large Scale Integration), ASIC (Application Specific Integrated Circuit), FPGA (Field-Programmable Gate Array), and GPU (Graphics Processing Unit). It may be realized by a circuit unit (including circuit), or may be realized by cooperation of software and hardware. The program may be stored in advance in a storage device such as an HDD (Hard Disk Drive) or a flash memory, or is stored in a removable storage medium such as a DVD or CD-ROM, and the storage medium is stored in the drive device. It may be installed by being attached.

  The data management unit 14 stores, for example, information input using the input unit 11 by the operator of the power system 21 in the storage unit 13. The data management unit 14 may store information received from an external device via the network in the storage unit 13.

  The system information collection unit 15 collects various types of information related to the current system state of the power system 21 (hereinafter referred to as a first system state) based on the system information received from the measurement device 25 installed in the power system 21. . The system information collection unit 15 may output the collected information to the system state estimation unit 16 or store it in the storage unit 13 as part of the first system information 13E. The first system state is a state of power supplied by the power system 21 at the present time, and includes, for example, at least one of a voltage, a phase, a load, a generator output, and a power flow at a plurality of locations in the power system 21.

  The first system information 13E includes, for example, node data and branch data. FIG. 5 is a diagram illustrating an example of the contents of node data. FIG. 6 is a diagram illustrating an example of the content of branch data. The node data shown in FIG. 5 includes the node name, voltage, phase, generator output (active power output, reactive power output), load (active power load, reactive power load), and phase adjustment equipment. Information associated with information. The branch data shown in FIG. 6 is information in which active power flow, reactive power flow, active power loss, and reactive power loss information are associated with the branch name.

  The system state estimation unit 16 estimates a future system state of the power system 21 (hereinafter referred to as a second system state) based on the current system state (first system state) of the power system 21. The system state estimation unit 16 may acquire information indicating the first system state from the system information collection unit 15 or read from the storage unit 13. When the system state estimation unit 16 estimates the second system state, the system state estimation unit 16 further predicts the output of the renewable energy power source 23 (renewable energy output prediction information 13B), demand prediction related to power demand (demand prediction information 13C), An operation plan (operation plan information 13D) related to power supply in the power system 21 is also referred to. Further, the system state estimation unit 16 may estimate a plurality of second system states by changing the predicted value according to the predicted value width predicted as the output of the renewable energy power source 23.

  The stability evaluation unit 17 is assumed by inputting the second system state estimated by the system state estimation unit 16 and the accident assumed in the second system state (assumed accident case information 13A) to the evaluation model. The stability in the power system 21 after the accident is derived. When deriving the stability, the stability evaluation unit 17 may further derive the stability by inputting the renewable energy output prediction information 13B into the evaluation model. The evaluation model is a learned model generated by the learning processing unit 40, for example. The evaluation model is information generated by an external learning processing device, and may be stored in the storage unit 13. The stability is an index indicating the stability of power supply in the power system 21 and includes, for example, at least one of an overload margin, a transient stability margin, a voltage stability margin, and a frequency stability margin. .

  The control content determination unit 18 determines whether or not the stability derived by the stability evaluation unit 17 is an unstable level. For example, the control content determination unit 18 determines that the level is unstable when the stability is equal to or less than a threshold value (for example, 1.05). When it is determined that the stability is at an unstable level, the control content determination unit 18 selects a device that can control an input variable having a correlation higher than a predetermined value among the devices in the power system 21 as a control target device. Determine as. The control content determination unit 18 determines control content such that the stability is not at an unstable level for the determined control target device, and transmits a control command based on the determined control content to the control terminal device 26. The control terminal device 26 controls the system control device and the generator 22 (output amount, etc.) and suppresses the output of the renewable energy power source 23 based on the received control command.

  The generation unit 19 generates output information to be displayed on the display unit 12 and outputs the output information to the display unit 12. For example, based on the stability derived by the stability evaluation unit 17, the generation unit 19 generates output information for displaying, for example, a stability screen that displays the stability for each second system state or assumed accident type. . The generation unit 19 generates a comparison image based on the comparison between the learning data used for learning the evaluation model and the second system state, and generates output information for displaying the comparison image. The process in which the generation unit 19 generates the comparison image will be described later.

[Learning processing unit 40]
The learning processing unit 40 generates an evaluation model. FIG. 7 is a diagram illustrating an example of the configuration of the learning processing unit 40. As illustrated in FIG. 7, the learning processing unit 40 includes, for example, a data acquisition unit 41, a storage unit 42, a learning unit 43, and a derivation unit 44. The learning processing unit 40 may be realized by the power system monitoring system 10 or may be realized by an external device. Some or all of the functional units of the data acquisition unit 41, the learning unit 43, and the derivation unit 44 are realized by a processor such as a CPU executing a program (software). Some or all of these functional units may be realized by hardware such as LSI, ASIC, FPGA, GPU, or may be realized by cooperation of software and hardware. The storage unit 42 is realized by, for example, a flash memory such as a RAM, a ROM, and an SSD, an HDD, and the like. The storage unit 42 stores, for example, learning data 42A, stability 42B, evaluation model 42C, correlation coefficient information 42D, and the like.

  The data acquisition unit 41 acquires the learning data Dt and the stability St, and stores them in the storage unit 42 as part of the learning data 42A and part of the stability 42B, respectively. FIG. 8 is a diagram illustrating an example of a plurality of learning data Dt and a plurality of stability Sts. The plurality of learning data Dt includes, for example, learning data Dt (1), Dt (2)... Dt (n). n is a natural number. Each learning data Dt includes information indicating an assumed system state, information indicating an assumed accident type, and the like. For each learning data Dt, a plurality of stability Sts are derived for each assumed accident type by simulation. For example, a plurality of stability St (1) -1, St (1) -2,..., St (1) -k are k assumed accident types assumed in the system state indicated by the learning data Dt (1). Each is the stability derived by simulation.

  The learning unit 43 generates an evaluation model that is a learned model by learning the relationship between the learning data Dt acquired by the data acquisition unit 41 and the stability St. For example, the learning unit 43 optimizes the discriminator so that a result reaching a simulation result can be obtained when each piece of system information included in each learning data Dt is input as a vector element to a discriminator such as a neural network. By obtaining parameters, an evaluation model is generated and stored in the storage unit 42 as an evaluation model 42C.

  Here, with reference to FIG. 9, an example of processing by the learning unit 43 and an evaluation model will be described. FIG. 9 is a reference diagram for explaining an example of processing and an evaluation model by the learning unit 43. The learning unit 43 generates an evaluation model Em by executing machine learning based on the plurality of learning data Dt and the plurality of stability St. The evaluation model Em derives stability by inputting input data. For example, when the input data is the estimated data De, the evaluation model Em derives the stability Se. When the input data is the learning data Dt, the evaluation model Em derives the stability Stm.

  The estimation data De includes information indicating the second system state estimated by the system state estimation unit 16, information indicating an assumed accident type used when estimating the second system state, and the like. FIG. 10 is a diagram illustrating an example of the estimation data De and a plurality of stability levels Se. The plurality of estimation data De includes, for example, estimation data De (1), De (2)... De (j). j is a natural number. Each estimation data De includes information indicating the second system state, information indicating an assumed accident type assumed in the second system state, and the like. For each estimated data De, a plurality of stability Se is derived for each assumed accident type using an evaluation model. For example, a plurality of stability levels Se (1) -1, Se (1) -2,..., Se (1) -k are k assumed accident types assumed in the system state indicated by the estimated data De (1). Each is the stability derived using the evaluation model.

  The deriving unit 44 derives a correlation coefficient ρ indicating a correlation between a part of the learning data Dt and the stability thereof for every element included in the learning data Dt for all the learning data Dt. For example, the deriving unit 44 derives the correlation coefficient ρ for each dimension included in the learning data Dt such as the node voltage and the active power of the branch. The deriving unit 44 may derive the correlation coefficient ρ for each assumed accident type. The deriving unit 44 stores the derived correlation coefficient ρ in the storage unit 42 as a part of the correlation coefficient information 42D.

[Generator 19 and Comparative Image]
The generation unit 19 generates at least one of the first comparison image and the second comparison image based on the estimation data De. The generation unit 19 may generate a first comparison image and a second comparison image for each assumed accident type, and display the generated plurality of comparison images side by side in one screen.

  First, the first comparison image will be described. FIG. 11 is a diagram illustrating an example of the first comparison image 510. For example, the first comparison image 510 is an image including a scatter diagram in which the horizontal axis indicates the node voltage and the vertical axis indicates the voltage stability margin. The node voltage is a voltage value at a certain node and is a part of the system state. The voltage stability margin is one of “load margins” included in the stability.

  The first comparison image 510 is an image representing the “relation between the system state and the stability in the simulation” and the “relation between the system state and the stability in the evaluation model” in the same scatter diagram. “Relationship between system state and stability in simulation” is a relationship between part of the system state included in the learning data Dt and stability St derived by simulation. The “relation between the system state and the stability in the evaluation model” is a relationship between a corresponding part of the estimated data De (second system state) and the stability Se derived by the stability evaluation unit 17.

  The generation unit 19 plots the coordinates indicating the “relation between the system state and the stability in the simulation”, for example, in a scatter diagram with a first graphic (diamond in the figure). In addition, the generation unit 19 plots the coordinates indicating “the relationship between the system state and the stability in the evaluation model” on the scatter diagram with the second graphic (cross in the figure). The coordinate value indicated by the first graphic is a value indicating the stability St derived by simulation based on a certain node voltage included in the learning data Dt. The coordinate value indicated by the second graphic is the stability Se derived by inputting a certain node voltage included in the estimated data De into the evaluation model Em. In addition, the accident type assumed when deriving both stability St and Se is the same accident type.

  Next, the second comparison image will be described. FIG. 12 is a diagram illustrating an example of the second comparison image 520. The second comparison image 520 is an image including a graph with the horizontal axis representing the case ID and the vertical axis representing the difference in voltage stability margin. The case ID is identification information assigned for each combination of the system state and the assumed accident type. The difference in voltage stability margin is a difference in stability St derived by inputting learning data Dt into the simulation with respect to stability Se derived by inputting estimated data De to evaluation model Em.

  The second comparison image 520 is derived, for example, by inputting learning data Dt having high similarity to the estimated data De to the stability Se derived by inputting the estimated data De to the evaluation model Em. It is the image which displayed the difference of the stable St for every assumption accident classification. Note that the second comparison image 520 may be an image in which the difference in the stability St with respect to the stability Se is displayed for each case (system state × assumed accident type).

  The second comparison image 520 is an image that displays the similarity between the learning data Dt and the estimated data De in association with the difference between the stability Se and the stability St. For example, in the graph of the second comparison image 520, the difference between the stability St and the stability Se is expressed by coordinate values (d11, etc.). Further, in the graph of the second comparison image 520, the similarity between the learning data Dt and the estimated data De is expressed by the intensity of the icon color indicated by the coordinate value. For example, the darker the color, the more similar the learning data Dt and the estimated data De. The generation unit 19 calculates the similarity with all the learning data Dt with respect to the estimated data De. For example, the generation unit 19 calculates Euclidean distances from all the learning data Dt with respect to the estimation data De. Next, the generation unit 19 selects the top five from the shortest distance when the calculated distances are arranged in the shortest order. And the production | generation part 19 corresponds to any of the similarities 1-5 classified in five steps based on the result of having compared the value of each of the selected top five distances with the predetermined reference value. By determining whether to do so, the similarity is calculated in five stages.

  When the difference in stability is small in the second comparison image 520 (the variation between coordinate values associated with the same case ID is small), the operator who viewed the second comparison image 520 is determined to be unstable. It can be seen that the accuracy of the obtained stability Se is high. On the other hand, when the difference in stability is large in the second comparison image 520 (the variation between coordinate values associated with the same case ID is large), the operator who has seen the second comparison image 520 is unstable. It can be seen that the accuracy of the stability Se determined to be low is low.

  In case ID (F27PV1334), five coordinate values d11 to d15 are plotted. The coordinate values d11 to d15 are, for example, differences between the stability St (1) to St (5) with respect to the stability Se (1). For example, the coordinate value d11 is a difference of the stability St (1) with respect to the stability Se (1).

  Further, the generation unit 19 may determine the arrangement order of the horizontal axis (case ID) in the graph shown in FIG. 12 according to the difference in stability. In the arrangement order shown in FIG. 12, the smaller the average value of the stability difference is, the more the left side of the horizontal axis is, and the larger the average value of the stability difference is, the more the right side of the horizontal axis is. . The average value of the stability is, for example, an average value of the difference in stability between the coordinate values d11 to d15.

  Note that the comparative image 522 shown in FIG. 13 is another example of the second comparative image. The generation unit 19 may generate a comparison image 522 that displays coordinate values corresponding to all case IDs as illustrated.

  FIG. 14 is a flowchart illustrating an example of the overall processing flow in the power system monitoring system 10. In the following description, the process for generating the comparison screen will be mainly described, and the description of the process for determining the control content will be omitted.

  First, the system information collection unit 15 collects system information indicating the first system state that is the system state of the current power system 21 (step S101). The system state estimation part 16 estimates the 2nd system state which is the future system state of the electric power system 21 based on the system information which shows this 1st system state (step S102). Assuming that the stability evaluation unit 17 inputs the second system state estimated by the system state estimation unit 16 and the accident assumed in the second system state (assumed accident case information 13A) to the evaluation model Em. The stability Se in the power system 21 after the accident is performed is derived (step S103).

  Next, the production | generation part 19 produces | generates the output information for displaying the stability screen, a comparison screen, etc. on the display part 12 (step S104). Then, the generation unit 19 outputs output information for displaying the stability screen to the display unit 12. Thereby, the display part 12 displays a stability screen (step S105). Next, the generation unit 19 determines whether an instruction to display a comparison screen has been received using the input unit 11 (step S106). The generation unit 19 displays the designated comparison image on the display unit 12 according to the instruction received using the input unit 11 (step S107).

  FIG. 15 is a flowchart illustrating an example of a process flow in which the generation unit 19 generates a comparison image. The generation unit 19 determines whether or not the target range of the comparison image has been designated by the operator using the input unit 11 (step S201). The target range includes, for example, a range in which the stability is equal to or higher than a threshold value (or is equal to or lower than the threshold value), a designated assumed accident type, a system state, or a regional set. When the target range is specified, the generation unit 19 extracts the estimation data De of the specified target range (step S202). If no target range is specified in step S201, the generation unit 19 sets all the estimated data De as targets of the comparison image.

  Next, the generation unit 19 refers to, for example, the storage unit 42 of the learning processing unit 40 and, for each dimension included in the learning data Dt, learning data (hereinafter referred to as learning data Dt ′) whose correlation coefficient ρ is equal to or greater than a threshold value. It is determined whether or not (step S203). When there is learning data Dt ′, the generation unit 19 determines whether display of the second comparison image is selected by the operator using the input unit 11 (step S204). When the operator has not selected the display of the second comparison image, the generation unit 19 extracts the learning data Dt ′ from the storage unit 42 of the learning processing unit 40 (Step S205), and based on the extracted learning data Dt ′. The first comparison image is generated (step S206). By doing so, it is possible to compare the learning data Dt ′ and the estimated data De that have a high correlation between the system state and the stability at the dimension level included in the system state. In FIG. 11, both the “relation between the system state and the stability in the simulation” and the “relation between the system state and the stability in the evaluation model” show the tendency indicated by the straight line L1. Thus, when the learning data Dt ′ and the estimated data De are similar in the first comparison image, it can be seen that the accuracy of the stability determined to be the unstable level is high.

  Next, the generation unit 19 determines whether or not the operator has instructed display of the second comparison image using the input unit 11 (step S207). When the operator does not instruct the display of the second comparison image, the generation unit 19 ends the process.

  When there is no learning data Dt ′ in step S203, when display of the second comparison image is selected by the driver in step S204 or step S207, the generation unit 19 refers to the storage unit 42 of the learning processing unit 40, and Learning data similar to the estimated data De (hereinafter referred to as learning data Dt ″) is extracted from all the learning data Dt (step S208). For example, the generation unit 19 derives the similarity with all the learning data Dt for each estimation data De, and extracts the learning data Dt from the highest similarity to the top five as the similar learning data Dt ″. To do. For example, the generation unit 19 derives the similarity using, for example, a calculation method such as Euclidean distance or cosine similarity. For example, the generation unit 19 may derive a similarity by converting a part of learning data of the same dimension into a vector or a matrix and normalizing the converted value.

  The generation unit 19 reads the stability corresponding to the extracted learning data Dt ″ (hereinafter referred to as the stability St ″) from the storage unit 42 of the learning processing unit 40 (step S209), and the read stability St Based on ″, a second comparison image is generated (step S210). By doing so, in terms of the difference between the stability St ″ and the stability Se and the similarity between the learning data Dt ″ and the estimated data De for each case (system state × assumed accident type), Similar learning data Dt ″ and estimated data De can be compared.

  According to at least one embodiment described above, based on the first system state that is the system state of the power system facility at the first time point, the power system facility at the second time point after a predetermined time from the first time point. An estimation unit for estimating a second system state that is a system state, the second system state estimated by the estimation unit, and information on an accident assumed in the second system state are input to a learned evaluation model A stability evaluation unit for deriving the stability of the power system equipment predicted after the assumed accident, learning data used for learning the learned model, and the second system state Providing users with information to confirm the accuracy of power system stability evaluated using a machine learning model by having a generator that generates comparison images based on comparisons Rukoto can.

  Thereby, the operator of the electric power system 21 can confirm the accuracy of the stability before re-evaluation by simulation even if the stability is an unstable level. When the operator determines that re-evaluation by simulation is unnecessary, the time required for the entire stability evaluation is shortened.

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

  Further, the generation unit 19 generates the first comparison image or the second comparison image designated by the operator using the input unit 11 without performing the processing corresponding to step S203 in the flowchart of FIG. You may make it display on. In this case, the generation unit 19 may generate only the designated comparison image, create both the first comparison image and the second comparison image, and cause the display unit 12 to display only the designated comparison image. You may do it.

  Further, the generation unit 19 determines, for example, whether or not the stability derived by the stability evaluation unit 17 is an unstable level, and based on the estimated data De determined that the stability is an unstable level. A comparison image may be generated. For example, the generation unit 19 determines that the level is unstable when the stability is equal to or less than a threshold value (for example, 1.05). In addition, when there are a plurality of stability levels determined as unstable levels in different assumed accident types, the generation unit 19 generates a first comparison image and a second comparison image for each assumed accident type, and generates the plurality of generated comparison images. May be displayed side by side in one screen.

  The generation unit 19 may determine, for each assumed accident type, whether or not the learning data Dt ′ whose correlation coefficient ρ is greater than or equal to the threshold can be extracted with reference to the storage unit 42 of the learning processing unit 40. When it is determined that the learning data Dt ′ cannot be extracted, the generation unit 19 may generate a second comparison image. Thus, even if it is difficult to determine whether or not the accuracy of the stability determined to be at the unstable level is high in the first comparison image, similar learning data Dt as in the second comparison image. By comparing ″ and the estimated data De, the operator of the power system 21 can determine the accuracy of the stability determined to be the unstable level.

  Further, the generation unit 19 may display an operation button for receiving an instruction for executing control by the control content determination unit 18 in the comparison screen or the stability screen. When the operation button is clicked, the generation unit 19 instructs the control content determination unit 18 to control the target range of the original data (estimated data) for generating the comparison image on which the operation button was displayed.

DESCRIPTION OF SYMBOLS 10 ... Electric power system monitoring system, 11 ... Input part, 12 ... Display part, 13 ... Memory | storage part, 14 ... Data management part, 15 ... System information collection part, 16 ... System state estimation part, 17 ... Stability evaluation part, 18 ... control content determination unit, 19 ... generation unit, 21 ... power system, 22 ... generator, 23 ... renewable energy power source, 24 ... consumer, 25 ... measuring device, 26 ... control terminal device, 31 ... regional set, 40 ... learning processing unit, 41 ... data acquisition unit, 42 ... storage unit, 43 ... learning unit, 44 ... derivation unit

Claims (10)

Estimation that estimates a second system state that is a system state of the power system facility at a second time point after a predetermined time from the first time point, based on a first system state that is a system state of the power system facility at a first time point And
The second system state estimated by the estimation unit and the information on the accident assumed in the second system state are input to the learned evaluation model, so that the prediction is made after the assumed accident. A stability evaluation unit for deriving the stability of power system facilities;
A generation unit that generates a comparison image based on a comparison between learning data used for learning the learned model and the second system state;
A power system monitoring system comprising: The generation unit includes a relationship between a part of the system state included in the learning data and a stability derived by simulation based on the learning data, and a part corresponding to the part of the second system state. And an image representing the relationship between the stability derived by the stability evaluation unit and the same scatter diagram as the comparative image,
The power system monitoring system according to claim 1. The generator is
With reference to a correlation coefficient indicating a correlation between a part of the system state included in the learning data and the stability derived by the simulation based on the learning data,
The comparison image is generated based on a part of the system state having a high correlation coefficient among the system states included in the learning data.
The power system monitoring system according to claim 2. The generation unit extracts a system state similar to the second system state from the system state included in the learning data, the stability derived by simulation based on the learning data of the extracted system state, and the stability An image displaying the difference between the stability derived by the degree evaluation unit for each accident type is generated as the comparison image,
The power system monitoring system according to any one of claims 1 to 3. The generation unit further generates, as the comparison image, an image in which the similarity between the system state included in the learning data and the second system state is associated with the difference between the stability levels.
The power system monitoring system according to claim 4. The generator is
With reference to a correlation coefficient indicating a correlation between a part of the system state included in the learning data and the stability derived by the simulation based on the learning data,
In a system state included in the learning data, when a part of the system state having a high correlation coefficient does not exist, the comparison image is generated.
The power system monitoring system according to claim 5. The generator is
Relationship between a part of the system state included in the learning data and the stability derived by simulation based on the learning data, a part corresponding to the part of the second system state, and the stability evaluation An image representing the relationship with the stability derived by the unit in the same graph;
A system state similar to the second system state is extracted from the system state included in the learning data, the stability derived by simulation based on the learning data of the extracted system state, and derived by the stability evaluation unit The image of the type selected by the user is generated as the comparative image among the images displaying the difference between the stability and the accident type for each accident type.
The power system monitoring system according to any one of claims 1 to 6. A learning unit that generates the evaluation model by learning a relationship between the learning data and stability in a system state included in the learning data;
The power system monitoring system according to any one of claims 1 to 7. Computer
Based on the first system state that is the system state of the power system equipment at the first time point, the second system state that is the system state of the power system equipment at the second time point after a predetermined time from the first time point is estimated,
By inputting the estimated second system state and information on the assumed accident in the second system state into a learned evaluation model, the power system facility predicted after the assumed accident is input. Deriving stability,
Generating a comparison image based on a comparison between learning data used for learning the learned model and the second system state;
Power system monitoring method. On the computer,
Based on the first system state that is the system state of the power system facility at the first time point, the second system state that is the system state of the power system facility at the second time point after a predetermined time from the first time point is estimated,
By inputting the estimated second system state and information on the assumed accident in the second system state into a learned evaluation model, the power system facility predicted after the assumed accident is input. Derived stability,
Generating a comparison image based on a comparison between learning data used for learning the learned model and the second system state;
program.

Images